Skip to content

Instantly share code, notes, and snippets.

@ginward

ginward/ckpt_xiu_monthly.ipynb

Last active March 17, 2020 19:58
Show Gist options
  • Save ginward/50994388dfd5571dc557b97cc3f99bdf to your computer and use it in GitHub Desktop.
Save ginward/50994388dfd5571dc557b97cc3f99bdf to your computer and use it in GitHub Desktop.
Download ZIP
ckpt_xiu_monthly.ipynb
Display the source blob
Display the rendered blob
Raw
ckpt_xiu_monthly.ipynb
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.1"
},
"colab": {
"name": "ckpt_xiu_monthly.ipynb",
"provenance": [],
"collapsed_sections": [],
"toc_visible": true,
"machine_shape": "hm",
"include_colab_link": true
},
"accelerator": "GPU"
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"<a href=\"https://colab.research.google.com/gist/ginward/50994388dfd5571dc557b97cc3f99bdf/ckpt_xiu_monthly.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "YX8cgPsuPUd6",
"colab_type": "text"
},
"source": [
"# 1. Introduction"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "QCPSfZqhPUd9",
"colab_type": "text"
},
"source": [
"This notebook is to replicate the main result of the paper: \n",
"Gu, Shihao, Bryan T. Kelly, and Dacheng Xiu. \"Autoencoder asset pricing models.\" Available at SSRN (2019).\n",
"\n",
"Please refer to the requirement.txt for environment configuration.\n",
"\n",
"This version also implements checkpointing."
]
},
{
"cell_type": "code",
"metadata": {
"id": "2ia_JKmlPUd-",
"colab_type": "code",
"colab": {}
},
"source": [
"import numpy as np\n",
"import pandas as pd\n",
"import h5py\n",
"from matplotlib import pyplot as plt\n",
"from scipy import stats\n",
"import torch as t\n",
"from torch import nn\n",
"import os\n",
"from torch import optim\n",
"import torch.utils.data as Data\n",
"import warnings\n",
"warnings.filterwarnings(\"ignore\")\n",
"t.manual_seed(1)\n",
"t.cuda.manual_seed(1)"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "7xd1MpSTPUeB",
"colab_type": "code",
"outputId": "aad6d94d-8b23-453a-a479-d2de56b05f0d",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"source": [
"print(t.__version__)"
],
"execution_count": 0,
"outputs": [
{
"output_type": "stream",
"text": [
"1.4.0\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "6urp-yfyPUeF",
"colab_type": "code",
"outputId": "be912f96-9f7d-425e-9883-6819b09edb25",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"source": [
"print(t.version.cuda)"
],
"execution_count": 0,
"outputs": [
{
"output_type": "stream",
"text": [
"10.1\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "6z1BCbEcMG1w",
"colab_type": "code",
"outputId": "15f939ad-8b2e-475d-bf2a-aa37ca899bb2",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 306
}
},
"source": [
"gpu_info = !nvidia-smi\n",
"gpu_info = '\\n'.join(gpu_info)\n",
"if gpu_info.find('failed') >= 0:\n",
" print('Select the Runtime → \"Change runtime type\" menu to enable a GPU accelerator, ')\n",
" print('and then re-execute this cell.')\n",
"else:\n",
" print(gpu_info)"
],
"execution_count": 0,
"outputs": [
{
"output_type": "stream",
"text": [
"Tue Mar 17 17:11:15 2020 \n",
"+-----------------------------------------------------------------------------+\n",
"| NVIDIA-SMI 440.59 Driver Version: 418.67 CUDA Version: 10.1 |\n",
"|-------------------------------+----------------------+----------------------+\n",
"| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |\n",
"| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |\n",
"|===============================+======================+======================|\n",
"| 0 Tesla P100-PCIE... Off | 00000000:00:04.0 Off | 0 |\n",
"| N/A 52C P0 28W / 250W | 0MiB / 16280MiB | 0% Default |\n",
"+-------------------------------+----------------------+----------------------+\n",
" \n",
"+-----------------------------------------------------------------------------+\n",
"| Processes: GPU Memory |\n",
"| GPU PID Type Process name Usage |\n",
"|=============================================================================|\n",
"| No running processes found |\n",
"+-----------------------------------------------------------------------------+\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "psJfSgLcPUeI",
"colab_type": "text"
},
"source": [
"If you should run this code locally, you should make sure `t.cuda.is_available() == True`, otherwise you should install the right pytorch matches your cuda version.\n",
"If you want to run the code only on CPU, then please remove every `.cuda()` in the code."
]
},
{
"cell_type": "code",
"metadata": {
"id": "RZbBFDRxPUeI",
"colab_type": "code",
"outputId": "c4c87845-a926-4dfd-9b8f-802ba76500d7",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"source": [
"t.cuda.is_available()"
],
"execution_count": 0,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"True"
]
},
"metadata": {
"tags": []
},
"execution_count": 5
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "xma_fGFIPUeL",
"colab_type": "text"
},
"source": [
"# 2. Data "
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "RZ6JEeHlPUeM",
"colab_type": "text"
},
"source": [
"## 2.1 Overview"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "MrCl5DjBPUeN",
"colab_type": "text"
},
"source": [
"We used the `datashare.csv` updated by the author of this paper"
]
},
{
"cell_type": "code",
"metadata": {
"id": "mPkbG5XaUf-h",
"colab_type": "code",
"colab": {}
},
"source": [
"# Connect with Google Drive\n",
"from google.colab import drive\n",
"drive.mount('/content/drive', force_remount = True)"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "yOBynFNbWKb9",
"colab_type": "code",
"outputId": "10f30eb8-75bc-478b-d43c-776037fe60d6",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"source": [
"cd drive/My Drive/Colab Notebooks"
],
"execution_count": 0,
"outputs": [
{
"output_type": "stream",
"text": [
"/content/drive/My Drive/Colab Notebooks\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "tzpgOXZuPUeO",
"colab_type": "code",
"colab": {}
},
"source": [
"data = pd.read_csv('./xiu_month_rf_hpr.csv')"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "FlMw8B41DpNN",
"colab_type": "code",
"colab": {}
},
"source": [
"#we don't want return to contain nan\n",
"data=data[data['return'].isnull()==False]"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "cT-3qO_MJtB0",
"colab_type": "code",
"colab": {}
},
"source": [
"#RF is in percentages\n",
"data['RF'] = data['RF']/100"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "vnrHUF45JSiG",
"colab_type": "code",
"colab": {}
},
"source": [
"data['return'] = data['return'] - data['RF']"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "PHeNtc0EJdzl",
"colab_type": "code",
"colab": {}
},
"source": [
"#drop the unnecessary columns\n",
"data = data.drop(columns = [\"MONTH\", \"RF\"])"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "CkBau5sjPUeR",
"colab_type": "code",
"colab": {}
},
"source": [
"#shift the return to future return\n",
"data['return']=data['return'].shift(periods=1)"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "V5C8-W_SMzN5",
"colab_type": "code",
"colab": {}
},
"source": [
"#we don't want return to contain nan\n",
"data=data[data['return'].isnull()==False]"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "XdzhwhSRQ9Z5",
"colab_type": "code",
"outputId": "510b5a42-5929-4d62-eb3a-c7350da05bbb",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 289
}
},
"source": [
"data.columns"
],
"execution_count": 0,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"Index(['permno', 'DATE', 'mvel1', 'beta', 'betasq', 'chmom', 'dolvol',\n",
" 'idiovol', 'indmom', 'mom1m', 'mom6m', 'mom12m', 'mom36m', 'pricedelay',\n",
" 'turn', 'absacc', 'acc', 'age', 'agr', 'bm', 'bm_ia', 'cashdebt',\n",
" 'cashpr', 'cfp', 'cfp_ia', 'chatoia', 'chcsho', 'chempia', 'chinv',\n",
" 'chpmia', 'convind', 'currat', 'depr', 'divi', 'divo', 'dy', 'egr',\n",
" 'ep', 'gma', 'grcapx', 'grltnoa', 'herf', 'hire', 'invest', 'lev',\n",
" 'lgr', 'mve_ia', 'operprof', 'orgcap', 'pchcapx_ia', 'pchcurrat',\n",
" 'pchdepr', 'pchgm_pchsale', 'pchquick', 'pchsale_pchinvt',\n",
" 'pchsale_pchrect', 'pchsale_pchxsga', 'pchsaleinv', 'pctacc', 'ps',\n",
" 'quick', 'rd', 'rd_mve', 'rd_sale', 'realestate', 'roic', 'salecash',\n",
" 'saleinv', 'salerec', 'secured', 'securedind', 'sgr', 'sin', 'sp',\n",
" 'tang', 'tb', 'aeavol', 'cash', 'chtx', 'cinvest', 'ear', 'nincr',\n",
" 'roaq', 'roavol', 'roeq', 'rsup', 'stdacc', 'stdcf', 'ms', 'baspread',\n",
" 'ill', 'maxret', 'retvol', 'std_dolvol', 'std_turn', 'zerotrade',\n",
" 'sic2', 'PERMNO', 'date', 'return'],\n",
" dtype='object')"
]
},
"metadata": {
"tags": []
},
"execution_count": 15
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "ZFfYUw7jPUeT",
"colab_type": "code",
"outputId": "4b2c9394-7fc7-4705-a7c5-604855e8f89f",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"source": [
"data.shape"
],
"execution_count": 0,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"(3739448, 100)"
]
},
"metadata": {
"tags": []
},
"execution_count": 16
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "0KhP6wl_y5Pu",
"colab_type": "code",
"colab": {}
},
"source": [
""
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "5VLqDjQqPUeZ",
"colab_type": "code",
"colab": {}
},
"source": [
"summary = data.describe()"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "w1hc8hFaPUeb",
"colab_type": "code",
"outputId": "0b16b58c-88f6-4c81-c198-eb5a92b85ea5",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 397
}
},
"source": [
"summary"
],
"execution_count": 0,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>permno</th>\n",
" <th>DATE</th>\n",
" <th>mvel1</th>\n",
" <th>beta</th>\n",
" <th>betasq</th>\n",
" <th>chmom</th>\n",
" <th>dolvol</th>\n",
" <th>idiovol</th>\n",
" <th>indmom</th>\n",
" <th>mom1m</th>\n",
" <th>mom6m</th>\n",
" <th>mom12m</th>\n",
" <th>mom36m</th>\n",
" <th>pricedelay</th>\n",
" <th>turn</th>\n",
" <th>absacc</th>\n",
" <th>acc</th>\n",
" <th>age</th>\n",
" <th>agr</th>\n",
" <th>bm</th>\n",
" <th>bm_ia</th>\n",
" <th>cashdebt</th>\n",
" <th>cashpr</th>\n",
" <th>cfp</th>\n",
" <th>cfp_ia</th>\n",
" <th>chatoia</th>\n",
" <th>chcsho</th>\n",
" <th>chempia</th>\n",
" <th>chinv</th>\n",
" <th>chpmia</th>\n",
" <th>convind</th>\n",
" <th>currat</th>\n",
" <th>depr</th>\n",
" <th>divi</th>\n",
" <th>divo</th>\n",
" <th>dy</th>\n",
" <th>egr</th>\n",
" <th>ep</th>\n",
" <th>gma</th>\n",
" <th>grcapx</th>\n",
" <th>...</th>\n",
" <th>ps</th>\n",
" <th>quick</th>\n",
" <th>rd</th>\n",
" <th>rd_mve</th>\n",
" <th>rd_sale</th>\n",
" <th>realestate</th>\n",
" <th>roic</th>\n",
" <th>salecash</th>\n",
" <th>saleinv</th>\n",
" <th>salerec</th>\n",
" <th>secured</th>\n",
" <th>securedind</th>\n",
" <th>sgr</th>\n",
" <th>sin</th>\n",
" <th>sp</th>\n",
" <th>tang</th>\n",
" <th>tb</th>\n",
" <th>aeavol</th>\n",
" <th>cash</th>\n",
" <th>chtx</th>\n",
" <th>cinvest</th>\n",
" <th>ear</th>\n",
" <th>nincr</th>\n",
" <th>roaq</th>\n",
" <th>roavol</th>\n",
" <th>roeq</th>\n",
" <th>rsup</th>\n",
" <th>stdacc</th>\n",
" <th>stdcf</th>\n",
" <th>ms</th>\n",
" <th>baspread</th>\n",
" <th>ill</th>\n",
" <th>maxret</th>\n",
" <th>retvol</th>\n",
" <th>std_dolvol</th>\n",
" <th>std_turn</th>\n",
" <th>zerotrade</th>\n",
" <th>sic2</th>\n",
" <th>PERMNO</th>\n",
" <th>return</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>count</th>\n",
" <td>3.739448e+06</td>\n",
" <td>3.739448e+06</td>\n",
" <td>3.736495e+06</td>\n",
" <td>3.371274e+06</td>\n",
" <td>3.371274e+06</td>\n",
" <td>3.428516e+06</td>\n",
" <td>3.393674e+06</td>\n",
" <td>3.371274e+06</td>\n",
" <td>3.739335e+06</td>\n",
" <td>3.710613e+06</td>\n",
" <td>3.596218e+06</td>\n",
" <td>3.428516e+06</td>\n",
" <td>2.836227e+06</td>\n",
" <td>3.371202e+06</td>\n",
" <td>3.395522e+06</td>\n",
" <td>2.315500e+06</td>\n",
" <td>2.315500e+06</td>\n",
" <td>2.819051e+06</td>\n",
" <td>2.584590e+06</td>\n",
" <td>2.770186e+06</td>\n",
" <td>2.770186e+06</td>\n",
" <td>2.675175e+06</td>\n",
" <td>2.773472e+06</td>\n",
" <td>2.498440e+06</td>\n",
" <td>2.498440e+06</td>\n",
" <td>2.344492e+06</td>\n",
" <td>2.580992e+06</td>\n",
" <td>2.551149e+06</td>\n",
" <td>2.501191e+06</td>\n",
" <td>2.544197e+06</td>\n",
" <td>2.819051e+06</td>\n",
" <td>2.697751e+06</td>\n",
" <td>2.623932e+06</td>\n",
" <td>2.584633e+06</td>\n",
" <td>2.584633e+06</td>\n",
" <td>2.805990e+06</td>\n",
" <td>2.557353e+06</td>\n",
" <td>2.816012e+06</td>\n",
" <td>2.578034e+06</td>\n",
" <td>2.223305e+06</td>\n",
" <td>...</td>\n",
" <td>2.584633e+06</td>\n",
" <td>2.670802e+06</td>\n",
" <td>2.584633e+06</td>\n",
" <td>1.322812e+06</td>\n",
" <td>1.305316e+06</td>\n",
" <td>945405.000000</td>\n",
" <td>2.675117e+06</td>\n",
" <td>2.781557e+06</td>\n",
" <td>2.185371e+06</td>\n",
" <td>2.694111e+06</td>\n",
" <td>1.428519e+06</td>\n",
" <td>2.819051e+06</td>\n",
" <td>2.549270e+06</td>\n",
" <td>2.819051e+06</td>\n",
" <td>2.808559e+06</td>\n",
" <td>2.636985e+06</td>\n",
" <td>2.485428e+06</td>\n",
" <td>2.213614e+06</td>\n",
" <td>2.138232e+06</td>\n",
" <td>2.050588e+06</td>\n",
" <td>2.028068e+06</td>\n",
" <td>2.241263e+06</td>\n",
" <td>2.243545e+06</td>\n",
" <td>2.159743e+06</td>\n",
" <td>1.842151e+06</td>\n",
" <td>2.220430e+06</td>\n",
" <td>2.180953e+06</td>\n",
" <td>1.427430e+06</td>\n",
" <td>1.427430e+06</td>\n",
" <td>2.180630e+06</td>\n",
" <td>3.738812e+06</td>\n",
" <td>3.432781e+06</td>\n",
" <td>3.738908e+06</td>\n",
" <td>3.736341e+06</td>\n",
" <td>3.425130e+06</td>\n",
" <td>3.435372e+06</td>\n",
" <td>3.431336e+06</td>\n",
" <td>3.485191e+06</td>\n",
" <td>3.739448e+06</td>\n",
" <td>3.739448e+06</td>\n",
" </tr>\n",
" <tr>\n",
" <th>mean</th>\n",
" <td>5.640640e+04</td>\n",
" <td>1.992516e+07</td>\n",
" <td>1.079295e+06</td>\n",
" <td>1.009825e+00</td>\n",
" <td>1.443811e+00</td>\n",
" <td>1.390837e-03</td>\n",
" <td>1.079274e+01</td>\n",
" <td>6.043689e-02</td>\n",
" <td>1.245873e-01</td>\n",
" <td>9.164804e-03</td>\n",
" <td>4.885550e-02</td>\n",
" <td>1.176537e-01</td>\n",
" <td>3.139641e-01</td>\n",
" <td>1.566428e-01</td>\n",
" <td>1.004903e+00</td>\n",
" <td>9.028293e-02</td>\n",
" <td>-2.009567e-02</td>\n",
" <td>1.126533e+01</td>\n",
" <td>-1.570820e-01</td>\n",
" <td>2.492919e+00</td>\n",
" <td>-5.026388e-01</td>\n",
" <td>2.821344e-02</td>\n",
" <td>-5.203027e-01</td>\n",
" <td>7.714399e-02</td>\n",
" <td>-1.520239e-01</td>\n",
" <td>-4.727783e-04</td>\n",
" <td>1.137399e-01</td>\n",
" <td>-9.211914e-02</td>\n",
" <td>1.274625e-02</td>\n",
" <td>1.734495e-01</td>\n",
" <td>1.403582e-01</td>\n",
" <td>3.738560e+00</td>\n",
" <td>2.546599e-01</td>\n",
" <td>3.110654e-02</td>\n",
" <td>3.034048e-02</td>\n",
" <td>2.094585e-02</td>\n",
" <td>1.408368e-01</td>\n",
" <td>-1.709670e-02</td>\n",
" <td>3.574233e-01</td>\n",
" <td>9.058863e-01</td>\n",
" <td>...</td>\n",
" <td>4.167191e+00</td>\n",
" <td>3.047020e+00</td>\n",
" <td>1.267689e-01</td>\n",
" <td>5.891470e-02</td>\n",
" <td>4.931038e-01</td>\n",
" <td>0.261367</td>\n",
" <td>-6.433974e-02</td>\n",
" <td>4.996550e+01</td>\n",
" <td>2.695078e+01</td>\n",
" <td>1.158898e+01</td>\n",
" <td>5.621686e-01</td>\n",
" <td>4.031154e-01</td>\n",
" <td>1.914831e-01</td>\n",
" <td>8.879584e-03</td>\n",
" <td>2.215269e+00</td>\n",
" <td>5.380046e-01</td>\n",
" <td>-1.630156e-01</td>\n",
" <td>8.094319e-01</td>\n",
" <td>1.569021e-01</td>\n",
" <td>1.008276e-03</td>\n",
" <td>2.018008e-01</td>\n",
" <td>3.227083e-03</td>\n",
" <td>1.015102e+00</td>\n",
" <td>3.200990e-04</td>\n",
" <td>2.630615e-02</td>\n",
" <td>7.392625e-03</td>\n",
" <td>1.921570e-02</td>\n",
" <td>6.254564e+00</td>\n",
" <td>1.277835e+01</td>\n",
" <td>3.625508e+00</td>\n",
" <td>5.400234e-02</td>\n",
" <td>5.082129e-06</td>\n",
" <td>6.982922e-02</td>\n",
" <td>3.062196e-02</td>\n",
" <td>8.671846e-01</td>\n",
" <td>4.074175e+00</td>\n",
" <td>1.509971e+00</td>\n",
" <td>4.741589e+01</td>\n",
" <td>5.640640e+04</td>\n",
" <td>7.267635e-03</td>\n",
" </tr>\n",
" <tr>\n",
" <th>std</th>\n",
" <td>2.732097e+04</td>\n",
" <td>1.420988e+05</td>\n",
" <td>4.825887e+06</td>\n",
" <td>6.480901e-01</td>\n",
" <td>1.723233e+00</td>\n",
" <td>5.388213e-01</td>\n",
" <td>2.922874e+00</td>\n",
" <td>3.660089e-02</td>\n",
" <td>2.861385e-01</td>\n",
" <td>1.473740e-01</td>\n",
" <td>3.501918e-01</td>\n",
" <td>5.601429e-01</td>\n",
" <td>9.156545e-01</td>\n",
" <td>1.353978e+00</td>\n",
" <td>1.047787e+01</td>\n",
" <td>9.846507e-02</td>\n",
" <td>1.272371e-01</td>\n",
" <td>1.009857e+01</td>\n",
" <td>4.204604e-01</td>\n",
" <td>2.676182e+01</td>\n",
" <td>2.446237e+01</td>\n",
" <td>1.543410e+01</td>\n",
" <td>8.348633e+01</td>\n",
" <td>1.332607e+00</td>\n",
" <td>6.283583e+00</td>\n",
" <td>2.192646e-01</td>\n",
" <td>3.242773e-01</td>\n",
" <td>9.348005e-01</td>\n",
" <td>5.621459e-02</td>\n",
" <td>5.779889e+00</td>\n",
" <td>3.473584e-01</td>\n",
" <td>1.230473e+01</td>\n",
" <td>3.934959e-01</td>\n",
" <td>1.736057e-01</td>\n",
" <td>1.715224e-01</td>\n",
" <td>3.718645e-02</td>\n",
" <td>6.511498e-01</td>\n",
" <td>3.566570e-01</td>\n",
" <td>3.356517e-01</td>\n",
" <td>4.528311e+00</td>\n",
" <td>...</td>\n",
" <td>1.715702e+00</td>\n",
" <td>1.230901e+01</td>\n",
" <td>3.327139e-01</td>\n",
" <td>1.040411e-01</td>\n",
" <td>3.791595e+00</td>\n",
" <td>0.281477</td>\n",
" <td>8.379559e-01</td>\n",
" <td>1.559034e+02</td>\n",
" <td>7.878391e+01</td>\n",
" <td>5.241357e+01</td>\n",
" <td>5.224417e-01</td>\n",
" <td>4.905237e-01</td>\n",
" <td>5.803090e-01</td>\n",
" <td>9.381226e-02</td>\n",
" <td>3.611889e+00</td>\n",
" <td>1.533588e-01</td>\n",
" <td>1.809665e+00</td>\n",
" <td>2.065480e+00</td>\n",
" <td>2.028530e-01</td>\n",
" <td>1.150926e-02</td>\n",
" <td>2.492995e+01</td>\n",
" <td>7.901209e-02</td>\n",
" <td>1.370279e+00</td>\n",
" <td>5.178176e-02</td>\n",
" <td>4.758519e-02</td>\n",
" <td>2.096277e-01</td>\n",
" <td>2.142203e-01</td>\n",
" <td>7.498682e+01</td>\n",
" <td>1.355331e+02</td>\n",
" <td>1.674303e+00</td>\n",
" <td>7.253443e-02</td>\n",
" <td>2.663448e-05</td>\n",
" <td>7.038525e-02</td>\n",
" <td>2.528540e-02</td>\n",
" <td>4.007274e-01</td>\n",
" <td>9.178549e+00</td>\n",
" <td>3.536545e+00</td>\n",
" <td>1.970392e+01</td>\n",
" <td>2.732097e+04</td>\n",
" <td>1.728344e-01</td>\n",
" </tr>\n",
" <tr>\n",
" <th>min</th>\n",
" <td>1.000000e+04</td>\n",
" <td>1.957033e+07</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-1.933279e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-9.062534e+00</td>\n",
" <td>-3.060271e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-7.794856e-01</td>\n",
" <td>-6.981132e-01</td>\n",
" <td>-1.000000e+00</td>\n",
" <td>-1.000000e+00</td>\n",
" <td>-1.000000e+00</td>\n",
" <td>-8.229368e+02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-1.257607e+00</td>\n",
" <td>1.000000e+00</td>\n",
" <td>-6.033304e+00</td>\n",
" <td>-2.439120e+01</td>\n",
" <td>-7.759533e+02</td>\n",
" <td>-1.024800e+04</td>\n",
" <td>-8.364906e+03</td>\n",
" <td>-2.057420e+02</td>\n",
" <td>-1.912332e+02</td>\n",
" <td>-1.314998e+00</td>\n",
" <td>-8.998305e-01</td>\n",
" <td>-5.622415e+01</td>\n",
" <td>-3.003145e-01</td>\n",
" <td>-1.574528e+02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>1.418063e-03</td>\n",
" <td>-9.838288e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-3.313476e+00</td>\n",
" <td>-6.199143e+00</td>\n",
" <td>-1.821396e+01</td>\n",
" <td>-9.219229e-01</td>\n",
" <td>-4.047032e+02</td>\n",
" <td>...</td>\n",
" <td>0.000000e+00</td>\n",
" <td>1.418063e-03</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-3.411239e-02</td>\n",
" <td>-9.038462e+01</td>\n",
" <td>0.000000</td>\n",
" <td>-1.717197e+01</td>\n",
" <td>-1.591636e+03</td>\n",
" <td>-1.066224e+02</td>\n",
" <td>-2.179600e+04</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-1.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-3.594196e+01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-4.219112e+01</td>\n",
" <td>-1.000000e+00</td>\n",
" <td>-1.431838e-01</td>\n",
" <td>-1.607357e-01</td>\n",
" <td>-2.833333e+03</td>\n",
" <td>-4.754902e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-5.334442e-01</td>\n",
" <td>4.202608e-06</td>\n",
" <td>-1.720769e+02</td>\n",
" <td>-3.343933e+01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>5.354650e-06</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-1.979275e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-7.335907e-02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>1.954854e-12</td>\n",
" <td>1.000000e+00</td>\n",
" <td>1.000000e+04</td>\n",
" <td>-9.912950e-01</td>\n",
" </tr>\n",
" <tr>\n",
" <th>25%</th>\n",
" <td>3.052500e+04</td>\n",
" <td>1.982093e+07</td>\n",
" <td>2.032800e+04</td>\n",
" <td>5.400761e-01</td>\n",
" <td>2.975018e-01</td>\n",
" <td>-2.380378e-01</td>\n",
" <td>8.760727e+00</td>\n",
" <td>3.428336e-02</td>\n",
" <td>-4.939807e-02</td>\n",
" <td>-6.144062e-02</td>\n",
" <td>-1.355932e-01</td>\n",
" <td>-1.919505e-01</td>\n",
" <td>-2.126540e-01</td>\n",
" <td>-5.269740e-02</td>\n",
" <td>1.900418e-01</td>\n",
" <td>2.969798e-02</td>\n",
" <td>-7.354222e-02</td>\n",
" <td>4.000000e+00</td>\n",
" <td>-2.030585e-01</td>\n",
" <td>3.376279e-01</td>\n",
" <td>-4.110030e-01</td>\n",
" <td>1.410281e-02</td>\n",
" <td>-8.480907e+00</td>\n",
" <td>-4.018772e-02</td>\n",
" <td>-1.441469e-01</td>\n",
" <td>-7.099654e-02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-1.804556e-01</td>\n",
" <td>-1.534082e-03</td>\n",
" <td>-1.304695e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>1.228087e+00</td>\n",
" <td>9.547563e-02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-2.729350e-02</td>\n",
" <td>-2.553201e-03</td>\n",
" <td>1.143995e-01</td>\n",
" <td>-3.721840e-01</td>\n",
" <td>...</td>\n",
" <td>3.000000e+00</td>\n",
" <td>8.793050e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>3.133341e-03</td>\n",
" <td>2.423276e-03</td>\n",
" <td>0.107750</td>\n",
" <td>6.581906e-03</td>\n",
" <td>2.635117e+00</td>\n",
" <td>4.606254e+00</td>\n",
" <td>3.735176e+00</td>\n",
" <td>1.497460e-02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>-1.065269e-02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>4.256670e-01</td>\n",
" <td>4.697502e-01</td>\n",
" <td>-7.153101e-01</td>\n",
" <td>-2.646674e-01</td>\n",
" <td>2.256222e-02</td>\n",
" <td>-1.279341e-03</td>\n",
" <td>-2.838249e-02</td>\n",
" <td>-3.225806e-02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>1.322650e-05</td>\n",
" <td>5.141335e-03</td>\n",
" <td>8.120633e-04</td>\n",
" <td>-4.400480e-03</td>\n",
" <td>7.941321e-02</td>\n",
" <td>8.458948e-02</td>\n",
" <td>2.000000e+00</td>\n",
" <td>1.931990e-02</td>\n",
" <td>1.070679e-08</td>\n",
" <td>2.777778e-02</td>\n",
" <td>1.444932e-02</td>\n",
" <td>5.540967e-01</td>\n",
" <td>7.449102e-01</td>\n",
" <td>2.009551e-08</td>\n",
" <td>3.400000e+01</td>\n",
" <td>3.052500e+04</td>\n",
" <td>-6.532400e-02</td>\n",
" </tr>\n",
" <tr>\n",
" <th>50%</th>\n",
" <td>6.177800e+04</td>\n",
" <td>1.994063e+07</td>\n",
" <td>8.572500e+04</td>\n",
" <td>9.384522e-01</td>\n",
" <td>8.842564e-01</td>\n",
" <td>-4.718135e-03</td>\n",
" <td>1.066154e+01</td>\n",
" <td>5.120070e-02</td>\n",
" <td>1.042282e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>2.208386e-02</td>\n",
" <td>5.331414e-02</td>\n",
" <td>1.520912e-01</td>\n",
" <td>6.873652e-02</td>\n",
" <td>4.587454e-01</td>\n",
" <td>6.230405e-02</td>\n",
" <td>-1.916127e-02</td>\n",
" <td>8.000000e+00</td>\n",
" <td>-7.571500e-02</td>\n",
" <td>6.494110e-01</td>\n",
" <td>-1.083135e-01</td>\n",
" <td>1.287089e-01</td>\n",
" <td>-6.661708e-01</td>\n",
" <td>4.781421e-02</td>\n",
" <td>-6.330727e-03</td>\n",
" <td>1.589537e-03</td>\n",
" <td>6.847698e-03</td>\n",
" <td>-5.982460e-02</td>\n",
" <td>2.894090e-04</td>\n",
" <td>-3.168992e-03</td>\n",
" <td>0.000000e+00</td>\n",
" <td>1.978973e+00</td>\n",
" <td>1.490439e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>4.161961e-03</td>\n",
" <td>7.802550e-02</td>\n",
" <td>5.058908e-02</td>\n",
" <td>3.006515e-01</td>\n",
" <td>1.338296e-01</td>\n",
" <td>...</td>\n",
" <td>4.000000e+00</td>\n",
" <td>1.296124e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>2.496483e-02</td>\n",
" <td>2.425175e-02</td>\n",
" <td>0.227216</td>\n",
" <td>6.736905e-02</td>\n",
" <td>1.005086e+01</td>\n",
" <td>7.609959e+00</td>\n",
" <td>5.895310e+00</td>\n",
" <td>5.376000e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>9.561810e-02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>1.033256e+00</td>\n",
" <td>5.477468e-01</td>\n",
" <td>-1.265953e-01</td>\n",
" <td>2.550494e-01</td>\n",
" <td>6.937483e-02</td>\n",
" <td>6.297230e-05</td>\n",
" <td>-1.184269e-03</td>\n",
" <td>5.297698e-04</td>\n",
" <td>1.000000e+00</td>\n",
" <td>7.819408e-03</td>\n",
" <td>1.134994e-02</td>\n",
" <td>2.398843e-02</td>\n",
" <td>1.446924e-02</td>\n",
" <td>1.317511e-01</td>\n",
" <td>1.436099e-01</td>\n",
" <td>4.000000e+00</td>\n",
" <td>3.276041e-02</td>\n",
" <td>1.291728e-07</td>\n",
" <td>4.878049e-02</td>\n",
" <td>2.349544e-02</td>\n",
" <td>8.029089e-01</td>\n",
" <td>1.695987e+00</td>\n",
" <td>5.636672e-08</td>\n",
" <td>4.800000e+01</td>\n",
" <td>6.177800e+04</td>\n",
" <td>-3.337000e-03</td>\n",
" </tr>\n",
" <tr>\n",
" <th>75%</th>\n",
" <td>8.068000e+04</td>\n",
" <td>2.004023e+07</td>\n",
" <td>4.118188e+05</td>\n",
" <td>1.388240e+00</td>\n",
" <td>1.930812e+00</td>\n",
" <td>2.335132e-01</td>\n",
" <td>1.278733e+01</td>\n",
" <td>7.686123e-02</td>\n",
" <td>2.611684e-01</td>\n",
" <td>6.666667e-02</td>\n",
" <td>1.813771e-01</td>\n",
" <td>3.071035e-01</td>\n",
" <td>5.797258e-01</td>\n",
" <td>3.222820e-01</td>\n",
" <td>1.087716e+00</td>\n",
" <td>1.141448e-01</td>\n",
" <td>4.696611e-02</td>\n",
" <td>1.600000e+01</td>\n",
" <td>1.586623e-02</td>\n",
" <td>1.124785e+00</td>\n",
" <td>2.505711e-01</td>\n",
" <td>2.820772e-01</td>\n",
" <td>4.930041e+00</td>\n",
" <td>1.235594e-01</td>\n",
" <td>8.755702e-02</td>\n",
" <td>7.816332e-02</td>\n",
" <td>6.497040e-02</td>\n",
" <td>2.976349e-02</td>\n",
" <td>2.251888e-02</td>\n",
" <td>5.664668e-02</td>\n",
" <td>0.000000e+00</td>\n",
" <td>3.209117e+00</td>\n",
" <td>2.630835e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>3.052534e-02</td>\n",
" <td>1.954101e-01</td>\n",
" <td>8.943522e-02</td>\n",
" <td>5.261533e-01</td>\n",
" <td>9.173273e-01</td>\n",
" <td>...</td>\n",
" <td>5.000000e+00</td>\n",
" <td>2.282214e+00</td>\n",
" <td>0.000000e+00</td>\n",
" <td>6.950250e-02</td>\n",
" <td>9.843776e-02</td>\n",
" <td>0.375574</td>\n",
" <td>1.337349e-01</td>\n",
" <td>3.565057e+01</td>\n",
" <td>1.717171e+01</td>\n",
" <td>8.917961e+00</td>\n",
" <td>1.000000e+00</td>\n",
" <td>1.000000e+00</td>\n",
" <td>2.371986e-01</td>\n",
" <td>0.000000e+00</td>\n",
" <td>2.435432e+00</td>\n",
" <td>6.116974e-01</td>\n",
" <td>3.853083e-01</td>\n",
" <td>1.105991e+00</td>\n",
" <td>2.069173e-01</td>\n",
" <td>3.174626e-03</td>\n",
" <td>2.111255e-02</td>\n",
" <td>3.675214e-02</td>\n",
" <td>2.000000e+00</td>\n",
" <td>1.978208e-02</td>\n",
" <td>2.711450e-02</td>\n",
" <td>4.376447e-02</td>\n",
" <td>5.448232e-02</td>\n",
" <td>2.518106e-01</td>\n",
" <td>2.943518e-01</td>\n",
" <td>5.000000e+00</td>\n",
" <td>5.894172e-02</td>\n",
" <td>1.264674e-06</td>\n",
" <td>8.571429e-02</td>\n",
" <td>3.830652e-02</td>\n",
" <td>1.118145e+00</td>\n",
" <td>3.927887e+00</td>\n",
" <td>9.545456e-01</td>\n",
" <td>6.300000e+01</td>\n",
" <td>8.068000e+04</td>\n",
" <td>6.276325e-02</td>\n",
" </tr>\n",
" <tr>\n",
" <th>max</th>\n",
" <td>9.343600e+04</td>\n",
" <td>2.016123e+07</td>\n",
" <td>1.221146e+08</td>\n",
" <td>3.987207e+00</td>\n",
" <td>1.589782e+01</td>\n",
" <td>8.083012e+00</td>\n",
" <td>1.900568e+01</td>\n",
" <td>2.720346e-01</td>\n",
" <td>7.682736e+00</td>\n",
" <td>2.172414e+00</td>\n",
" <td>7.844445e+00</td>\n",
" <td>1.168096e+01</td>\n",
" <td>1.685246e+01</td>\n",
" <td>9.918375e+02</td>\n",
" <td>1.734793e+04</td>\n",
" <td>1.257607e+00</td>\n",
" <td>1.155264e+00</td>\n",
" <td>5.400000e+01</td>\n",
" <td>8.265002e-01</td>\n",
" <td>2.460433e+03</td>\n",
" <td>1.564199e+03</td>\n",
" <td>4.256410e+01</td>\n",
" <td>1.139067e+04</td>\n",
" <td>1.477503e+02</td>\n",
" <td>3.657082e+02</td>\n",
" <td>2.663142e+00</td>\n",
" <td>7.208766e+00</td>\n",
" <td>6.629468e+01</td>\n",
" <td>3.991813e-01</td>\n",
" <td>1.530145e+02</td>\n",
" <td>1.000000e+00</td>\n",
" <td>8.330556e+02</td>\n",
" <td>1.509737e+01</td>\n",
" <td>1.000000e+00</td>\n",
" <td>1.000000e+00</td>\n",
" <td>1.159430e+00</td>\n",
" <td>1.385515e+01</td>\n",
" <td>8.404797e-01</td>\n",
" <td>6.138484e+00</td>\n",
" <td>4.866364e+02</td>\n",
" <td>...</td>\n",
" <td>9.000000e+00</td>\n",
" <td>8.330556e+02</td>\n",
" <td>1.000000e+00</td>\n",
" <td>2.370752e+00</td>\n",
" <td>1.584300e+02</td>\n",
" <td>57.709343</td>\n",
" <td>1.047405e+01</td>\n",
" <td>5.906600e+03</td>\n",
" <td>3.024533e+03</td>\n",
" <td>6.870769e+02</td>\n",
" <td>5.000000e+00</td>\n",
" <td>1.000000e+00</td>\n",
" <td>1.791538e+01</td>\n",
" <td>1.000000e+00</td>\n",
" <td>5.493151e+01</td>\n",
" <td>1.000000e+00</td>\n",
" <td>2.924935e+01</td>\n",
" <td>5.121570e+02</td>\n",
" <td>9.786834e-01</td>\n",
" <td>1.765701e-01</td>\n",
" <td>6.463467e+03</td>\n",
" <td>5.578913e-01</td>\n",
" <td>8.000000e+00</td>\n",
" <td>1.460327e+00</td>\n",
" <td>8.201621e-01</td>\n",
" <td>7.039832e+00</td>\n",
" <td>5.670456e+00</td>\n",
" <td>7.496989e+03</td>\n",
" <td>8.751301e+03</td>\n",
" <td>8.000000e+00</td>\n",
" <td>1.078788e+00</td>\n",
" <td>2.479339e-03</td>\n",
" <td>1.701520e+00</td>\n",
" <td>4.379252e-01</td>\n",
" <td>2.860123e+00</td>\n",
" <td>6.517474e+02</td>\n",
" <td>2.008697e+01</td>\n",
" <td>9.900000e+01</td>\n",
" <td>9.343600e+04</td>\n",
" <td>2.399660e+01</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>8 rows × 99 columns</p>\n",
"</div>"
],
"text/plain": [
" permno DATE ... PERMNO return\n",
"count 3.739448e+06 3.739448e+06 ... 3.739448e+06 3.739448e+06\n",
"mean 5.640640e+04 1.992516e+07 ... 5.640640e+04 7.267635e-03\n",
"std 2.732097e+04 1.420988e+05 ... 2.732097e+04 1.728344e-01\n",
"min 1.000000e+04 1.957033e+07 ... 1.000000e+04 -9.912950e-01\n",
"25% 3.052500e+04 1.982093e+07 ... 3.052500e+04 -6.532400e-02\n",
"50% 6.177800e+04 1.994063e+07 ... 6.177800e+04 -3.337000e-03\n",
"75% 8.068000e+04 2.004023e+07 ... 8.068000e+04 6.276325e-02\n",
"max 9.343600e+04 2.016123e+07 ... 9.343600e+04 2.399660e+01\n",
"\n",
"[8 rows x 99 columns]"
]
},
"metadata": {
"tags": []
},
"execution_count": 18
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "mPHNtLRMPUed",
"colab_type": "text"
},
"source": [
"Different stocks have different record #"
]
},
{
"cell_type": "code",
"metadata": {
"id": "EJTPsQh3PUee",
"colab_type": "code",
"outputId": "0ddcc628-a8a7-488a-f837-bd8f7b2a487d",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 470
}
},
"source": [
"data.groupby('permno').count()"
],
"execution_count": 0,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>DATE</th>\n",
" <th>mvel1</th>\n",
" <th>beta</th>\n",
" <th>betasq</th>\n",
" <th>chmom</th>\n",
" <th>dolvol</th>\n",
" <th>idiovol</th>\n",
" <th>indmom</th>\n",
" <th>mom1m</th>\n",
" <th>mom6m</th>\n",
" <th>mom12m</th>\n",
" <th>mom36m</th>\n",
" <th>pricedelay</th>\n",
" <th>turn</th>\n",
" <th>absacc</th>\n",
" <th>acc</th>\n",
" <th>age</th>\n",
" <th>agr</th>\n",
" <th>bm</th>\n",
" <th>bm_ia</th>\n",
" <th>cashdebt</th>\n",
" <th>cashpr</th>\n",
" <th>cfp</th>\n",
" <th>cfp_ia</th>\n",
" <th>chatoia</th>\n",
" <th>chcsho</th>\n",
" <th>chempia</th>\n",
" <th>chinv</th>\n",
" <th>chpmia</th>\n",
" <th>convind</th>\n",
" <th>currat</th>\n",
" <th>depr</th>\n",
" <th>divi</th>\n",
" <th>divo</th>\n",
" <th>dy</th>\n",
" <th>egr</th>\n",
" <th>ep</th>\n",
" <th>gma</th>\n",
" <th>grcapx</th>\n",
" <th>grltnoa</th>\n",
" <th>...</th>\n",
" <th>quick</th>\n",
" <th>rd</th>\n",
" <th>rd_mve</th>\n",
" <th>rd_sale</th>\n",
" <th>realestate</th>\n",
" <th>roic</th>\n",
" <th>salecash</th>\n",
" <th>saleinv</th>\n",
" <th>salerec</th>\n",
" <th>secured</th>\n",
" <th>securedind</th>\n",
" <th>sgr</th>\n",
" <th>sin</th>\n",
" <th>sp</th>\n",
" <th>tang</th>\n",
" <th>tb</th>\n",
" <th>aeavol</th>\n",
" <th>cash</th>\n",
" <th>chtx</th>\n",
" <th>cinvest</th>\n",
" <th>ear</th>\n",
" <th>nincr</th>\n",
" <th>roaq</th>\n",
" <th>roavol</th>\n",
" <th>roeq</th>\n",
" <th>rsup</th>\n",
" <th>stdacc</th>\n",
" <th>stdcf</th>\n",
" <th>ms</th>\n",
" <th>baspread</th>\n",
" <th>ill</th>\n",
" <th>maxret</th>\n",
" <th>retvol</th>\n",
" <th>std_dolvol</th>\n",
" <th>std_turn</th>\n",
" <th>zerotrade</th>\n",
" <th>sic2</th>\n",
" <th>PERMNO</th>\n",
" <th>date</th>\n",
" <th>return</th>\n",
" </tr>\n",
" <tr>\n",
" <th>permno</th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" <th></th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>10000</th>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>3</td>\n",
" <td>3</td>\n",
" <td>4</td>\n",
" <td>14</td>\n",
" <td>3</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>10</td>\n",
" <td>4</td>\n",
" <td>0</td>\n",
" <td>3</td>\n",
" <td>14</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>...</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" <td>15</td>\n",
" </tr>\n",
" <tr>\n",
" <th>10001</th>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>358</td>\n",
" <td>358</td>\n",
" <td>360</td>\n",
" <td>370</td>\n",
" <td>358</td>\n",
" <td>371</td>\n",
" <td>370</td>\n",
" <td>366</td>\n",
" <td>360</td>\n",
" <td>336</td>\n",
" <td>358</td>\n",
" <td>369</td>\n",
" <td>343</td>\n",
" <td>343</td>\n",
" <td>355</td>\n",
" <td>343</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>331</td>\n",
" <td>343</td>\n",
" <td>343</td>\n",
" <td>343</td>\n",
" <td>343</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>343</td>\n",
" <td>343</td>\n",
" <td>343</td>\n",
" <td>343</td>\n",
" <td>355</td>\n",
" <td>343</td>\n",
" <td>331</td>\n",
" <td>343</td>\n",
" <td>...</td>\n",
" <td>355</td>\n",
" <td>343</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>0</td>\n",
" <td>355</td>\n",
" <td>343</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>355</td>\n",
" <td>313</td>\n",
" <td>296</td>\n",
" <td>296</td>\n",
" <td>296</td>\n",
" <td>296</td>\n",
" <td>296</td>\n",
" <td>296</td>\n",
" <td>296</td>\n",
" <td>269</td>\n",
" <td>296</td>\n",
" <td>296</td>\n",
" <td>269</td>\n",
" <td>269</td>\n",
" <td>293</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" <td>371</td>\n",
" </tr>\n",
" <tr>\n",
" <th>10002</th>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" <td>311</td>\n",
" <td>311</td>\n",
" <td>313</td>\n",
" <td>323</td>\n",
" <td>311</td>\n",
" <td>324</td>\n",
" <td>323</td>\n",
" <td>319</td>\n",
" <td>313</td>\n",
" <td>289</td>\n",
" <td>311</td>\n",
" <td>322</td>\n",
" <td>91</td>\n",
" <td>91</td>\n",
" <td>223</td>\n",
" <td>211</td>\n",
" <td>223</td>\n",
" <td>223</td>\n",
" <td>223</td>\n",
" <td>223</td>\n",
" <td>91</td>\n",
" <td>91</td>\n",
" <td>199</td>\n",
" <td>211</td>\n",
" <td>211</td>\n",
" <td>211</td>\n",
" <td>211</td>\n",
" <td>223</td>\n",
" <td>223</td>\n",
" <td>223</td>\n",
" <td>211</td>\n",
" <td>211</td>\n",
" <td>223</td>\n",
" <td>211</td>\n",
" <td>223</td>\n",
" <td>211</td>\n",
" <td>187</td>\n",
" <td>0</td>\n",
" <td>...</td>\n",
" <td>223</td>\n",
" <td>211</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>223</td>\n",
" <td>223</td>\n",
" <td>211</td>\n",
" <td>223</td>\n",
" <td>0</td>\n",
" <td>223</td>\n",
" <td>211</td>\n",
" <td>223</td>\n",
" <td>223</td>\n",
" <td>223</td>\n",
" <td>19</td>\n",
" <td>217</td>\n",
" <td>217</td>\n",
" <td>210</td>\n",
" <td>210</td>\n",
" <td>217</td>\n",
" <td>217</td>\n",
" <td>217</td>\n",
" <td>175</td>\n",
" <td>217</td>\n",
" <td>217</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>217</td>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" <td>323</td>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" <td>324</td>\n",
" </tr>\n",
" <tr>\n",
" <th>10003</th>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>105</td>\n",
" <td>105</td>\n",
" <td>107</td>\n",
" <td>116</td>\n",
" <td>105</td>\n",
" <td>118</td>\n",
" <td>117</td>\n",
" <td>113</td>\n",
" <td>107</td>\n",
" <td>83</td>\n",
" <td>105</td>\n",
" <td>116</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>...</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>5</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" <td>118</td>\n",
" </tr>\n",
" <tr>\n",
" <th>10005</th>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>52</td>\n",
" <td>52</td>\n",
" <td>54</td>\n",
" <td>56</td>\n",
" <td>52</td>\n",
" <td>65</td>\n",
" <td>64</td>\n",
" <td>60</td>\n",
" <td>54</td>\n",
" <td>30</td>\n",
" <td>52</td>\n",
" <td>63</td>\n",
" <td>36</td>\n",
" <td>36</td>\n",
" <td>48</td>\n",
" <td>36</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>24</td>\n",
" <td>36</td>\n",
" <td>36</td>\n",
" <td>36</td>\n",
" <td>36</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>36</td>\n",
" <td>36</td>\n",
" <td>48</td>\n",
" <td>36</td>\n",
" <td>48</td>\n",
" <td>36</td>\n",
" <td>24</td>\n",
" <td>36</td>\n",
" <td>...</td>\n",
" <td>48</td>\n",
" <td>36</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>0</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>36</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>48</td>\n",
" <td>3</td>\n",
" <td>3</td>\n",
" <td>3</td>\n",
" <td>3</td>\n",
" <td>3</td>\n",
" <td>3</td>\n",
" <td>3</td>\n",
" <td>0</td>\n",
" <td>3</td>\n",
" <td>3</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>3</td>\n",
" <td>65</td>\n",
" <td>58</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>46</td>\n",
" <td>65</td>\n",
" <td>58</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" </tr>\n",
" <tr>\n",
" <th>...</th>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" <td>...</td>\n",
" </tr>\n",
" <tr>\n",
" <th>93432</th>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>9</td>\n",
" <td>0</td>\n",
" <td>11</td>\n",
" <td>10</td>\n",
" <td>6</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>9</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>...</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>2</td>\n",
" <td>2</td>\n",
" <td>2</td>\n",
" <td>2</td>\n",
" <td>2</td>\n",
" <td>2</td>\n",
" <td>2</td>\n",
" <td>0</td>\n",
" <td>2</td>\n",
" <td>2</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>0</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" <td>11</td>\n",
" </tr>\n",
" <tr>\n",
" <th>93433</th>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>64</td>\n",
" <td>64</td>\n",
" <td>66</td>\n",
" <td>76</td>\n",
" <td>64</td>\n",
" <td>77</td>\n",
" <td>76</td>\n",
" <td>72</td>\n",
" <td>66</td>\n",
" <td>42</td>\n",
" <td>64</td>\n",
" <td>75</td>\n",
" <td>53</td>\n",
" <td>53</td>\n",
" <td>65</td>\n",
" <td>53</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>41</td>\n",
" <td>53</td>\n",
" <td>53</td>\n",
" <td>53</td>\n",
" <td>53</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>53</td>\n",
" <td>53</td>\n",
" <td>65</td>\n",
" <td>53</td>\n",
" <td>65</td>\n",
" <td>53</td>\n",
" <td>41</td>\n",
" <td>53</td>\n",
" <td>...</td>\n",
" <td>65</td>\n",
" <td>53</td>\n",
" <td>60</td>\n",
" <td>60</td>\n",
" <td>60</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>0</td>\n",
" <td>65</td>\n",
" <td>17</td>\n",
" <td>65</td>\n",
" <td>53</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>51</td>\n",
" <td>51</td>\n",
" <td>45</td>\n",
" <td>48</td>\n",
" <td>51</td>\n",
" <td>51</td>\n",
" <td>51</td>\n",
" <td>23</td>\n",
" <td>51</td>\n",
" <td>51</td>\n",
" <td>23</td>\n",
" <td>23</td>\n",
" <td>46</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" <td>77</td>\n",
" </tr>\n",
" <tr>\n",
" <th>93434</th>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>67</td>\n",
" <td>76</td>\n",
" <td>65</td>\n",
" <td>78</td>\n",
" <td>77</td>\n",
" <td>73</td>\n",
" <td>67</td>\n",
" <td>43</td>\n",
" <td>65</td>\n",
" <td>76</td>\n",
" <td>60</td>\n",
" <td>60</td>\n",
" <td>72</td>\n",
" <td>60</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>48</td>\n",
" <td>60</td>\n",
" <td>60</td>\n",
" <td>60</td>\n",
" <td>60</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>60</td>\n",
" <td>60</td>\n",
" <td>72</td>\n",
" <td>60</td>\n",
" <td>72</td>\n",
" <td>60</td>\n",
" <td>48</td>\n",
" <td>60</td>\n",
" <td>...</td>\n",
" <td>72</td>\n",
" <td>60</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>36</td>\n",
" <td>72</td>\n",
" <td>60</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>72</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>62</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>38</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>38</td>\n",
" <td>38</td>\n",
" <td>65</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" </tr>\n",
" <tr>\n",
" <th>93435</th>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>9</td>\n",
" <td>9</td>\n",
" <td>11</td>\n",
" <td>21</td>\n",
" <td>9</td>\n",
" <td>22</td>\n",
" <td>21</td>\n",
" <td>17</td>\n",
" <td>11</td>\n",
" <td>0</td>\n",
" <td>9</td>\n",
" <td>20</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>0</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>0</td>\n",
" <td>10</td>\n",
" <td>...</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>10</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>13</td>\n",
" <td>8</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" <td>22</td>\n",
" </tr>\n",
" <tr>\n",
" <th>93436</th>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>65</td>\n",
" <td>65</td>\n",
" <td>67</td>\n",
" <td>76</td>\n",
" <td>65</td>\n",
" <td>78</td>\n",
" <td>77</td>\n",
" <td>73</td>\n",
" <td>67</td>\n",
" <td>43</td>\n",
" <td>65</td>\n",
" <td>76</td>\n",
" <td>54</td>\n",
" <td>54</td>\n",
" <td>66</td>\n",
" <td>54</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>42</td>\n",
" <td>54</td>\n",
" <td>54</td>\n",
" <td>54</td>\n",
" <td>54</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>54</td>\n",
" <td>54</td>\n",
" <td>66</td>\n",
" <td>54</td>\n",
" <td>66</td>\n",
" <td>54</td>\n",
" <td>42</td>\n",
" <td>36</td>\n",
" <td>...</td>\n",
" <td>66</td>\n",
" <td>54</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>54</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>66</td>\n",
" <td>71</td>\n",
" <td>71</td>\n",
" <td>65</td>\n",
" <td>71</td>\n",
" <td>71</td>\n",
" <td>71</td>\n",
" <td>71</td>\n",
" <td>41</td>\n",
" <td>71</td>\n",
" <td>71</td>\n",
" <td>41</td>\n",
" <td>41</td>\n",
" <td>66</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>77</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" <td>78</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>29853 rows × 99 columns</p>\n",
"</div>"
],
"text/plain": [
" DATE mvel1 beta betasq chmom ... zerotrade sic2 PERMNO date return\n",
"permno ... \n",
"10000 15 15 3 3 4 ... 15 15 15 15 15\n",
"10001 371 371 358 358 360 ... 371 371 371 371 371\n",
"10002 324 324 311 311 313 ... 324 324 324 324 324\n",
"10003 118 118 105 105 107 ... 118 118 118 118 118\n",
"10005 65 65 52 52 54 ... 58 65 65 65 65\n",
"... ... ... ... ... ... ... ... ... ... ... ...\n",
"93432 11 11 0 0 0 ... 11 11 11 11 11\n",
"93433 77 77 64 64 66 ... 77 77 77 77 77\n",
"93434 78 78 65 65 67 ... 78 78 78 78 78\n",
"93435 22 22 9 9 11 ... 22 22 22 22 22\n",
"93436 78 78 65 65 67 ... 78 78 78 78 78\n",
"\n",
"[29853 rows x 99 columns]"
]
},
"metadata": {
"tags": []
},
"execution_count": 19
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "IEtoLXgBPUeh",
"colab_type": "text"
},
"source": [
"## 2.2 Missing Values"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "F17S9eJPPUei",
"colab_type": "text"
},
"source": [
"mean Missing % is around 30%"
]
},
{
"cell_type": "code",
"metadata": {
"id": "d8CR4o-VPUej",
"colab_type": "code",
"outputId": "fe902cbf-47c1-4b5f-9616-26ed9d8e7e9d",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 374
}
},
"source": [
"(1-summary.loc['count'].sort_values()/3760208).plot.bar()\n",
"plt.title('Missing % for each column')"
],
"execution_count": 0,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"Text(0.5, 1.0, 'Missing % for each column')"
]
},
"metadata": {
"tags": []
},
"execution_count": 20
},
{
"output_type": "display_data",
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAFUCAYAAADf+HxmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"tags": []
}
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "KHXlMBwI4RFO",
"colab_type": "code",
"colab": {}
},
"source": [
"#we do not want to rank normalise the return or PERMNO\n",
"tmp_df=data[['return', \"PERMNO\"]]\n",
"#redundant columns\n",
"train_df = data.drop(columns=['permno','date','return', 'PERMNO'])"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "GjeqD6JQHYHW",
"colab_type": "code",
"outputId": "2fbfbfc1-c4d9-48f0-e785-ec54e0c5cccf",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 338
}
},
"source": [
"train_df.head()"
],
"execution_count": 0,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>DATE</th>\n",
" <th>mvel1</th>\n",
" <th>beta</th>\n",
" <th>betasq</th>\n",
" <th>chmom</th>\n",
" <th>dolvol</th>\n",
" <th>idiovol</th>\n",
" <th>indmom</th>\n",
" <th>mom1m</th>\n",
" <th>mom6m</th>\n",
" <th>mom12m</th>\n",
" <th>mom36m</th>\n",
" <th>pricedelay</th>\n",
" <th>turn</th>\n",
" <th>absacc</th>\n",
" <th>acc</th>\n",
" <th>age</th>\n",
" <th>agr</th>\n",
" <th>bm</th>\n",
" <th>bm_ia</th>\n",
" <th>cashdebt</th>\n",
" <th>cashpr</th>\n",
" <th>cfp</th>\n",
" <th>cfp_ia</th>\n",
" <th>chatoia</th>\n",
" <th>chcsho</th>\n",
" <th>chempia</th>\n",
" <th>chinv</th>\n",
" <th>chpmia</th>\n",
" <th>convind</th>\n",
" <th>currat</th>\n",
" <th>depr</th>\n",
" <th>divi</th>\n",
" <th>divo</th>\n",
" <th>dy</th>\n",
" <th>egr</th>\n",
" <th>ep</th>\n",
" <th>gma</th>\n",
" <th>grcapx</th>\n",
" <th>grltnoa</th>\n",
" <th>...</th>\n",
" <th>pchsaleinv</th>\n",
" <th>pctacc</th>\n",
" <th>ps</th>\n",
" <th>quick</th>\n",
" <th>rd</th>\n",
" <th>rd_mve</th>\n",
" <th>rd_sale</th>\n",
" <th>realestate</th>\n",
" <th>roic</th>\n",
" <th>salecash</th>\n",
" <th>saleinv</th>\n",
" <th>salerec</th>\n",
" <th>secured</th>\n",
" <th>securedind</th>\n",
" <th>sgr</th>\n",
" <th>sin</th>\n",
" <th>sp</th>\n",
" <th>tang</th>\n",
" <th>tb</th>\n",
" <th>aeavol</th>\n",
" <th>cash</th>\n",
" <th>chtx</th>\n",
" <th>cinvest</th>\n",
" <th>ear</th>\n",
" <th>nincr</th>\n",
" <th>roaq</th>\n",
" <th>roavol</th>\n",
" <th>roeq</th>\n",
" <th>rsup</th>\n",
" <th>stdacc</th>\n",
" <th>stdcf</th>\n",
" <th>ms</th>\n",
" <th>baspread</th>\n",
" <th>ill</th>\n",
" <th>maxret</th>\n",
" <th>retvol</th>\n",
" <th>std_dolvol</th>\n",
" <th>std_turn</th>\n",
" <th>zerotrade</th>\n",
" <th>sic2</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>19860331</td>\n",
" <td>11960.000000</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.255908</td>\n",
" <td>-0.257143</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>...</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.055511</td>\n",
" <td>1.891760e-06</td>\n",
" <td>0.044776</td>\n",
" <td>0.031004</td>\n",
" <td>1.021089</td>\n",
" <td>1.079774</td>\n",
" <td>1.023392e-07</td>\n",
" <td>39.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>19860430</td>\n",
" <td>16330.000000</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>7.897668</td>\n",
" <td>NaN</td>\n",
" <td>0.368892</td>\n",
" <td>0.365385</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.251252</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>...</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.037231</td>\n",
" <td>7.315091e-07</td>\n",
" <td>0.145161</td>\n",
" <td>0.044548</td>\n",
" <td>1.033817</td>\n",
" <td>1.745333</td>\n",
" <td>7.467463e-08</td>\n",
" <td>39.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>19860530</td>\n",
" <td>15172.000000</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>8.472954</td>\n",
" <td>NaN</td>\n",
" <td>0.388370</td>\n",
" <td>-0.098592</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.251604</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>...</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.048336</td>\n",
" <td>1.215981e-06</td>\n",
" <td>0.022727</td>\n",
" <td>0.011246</td>\n",
" <td>1.184555</td>\n",
" <td>1.502285</td>\n",
" <td>7.649551e-08</td>\n",
" <td>39.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>19860630</td>\n",
" <td>11793.859375</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>8.250098</td>\n",
" <td>NaN</td>\n",
" <td>0.400748</td>\n",
" <td>-0.222656</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.273223</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>...</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.062245</td>\n",
" <td>2.744328e-06</td>\n",
" <td>0.115702</td>\n",
" <td>0.038863</td>\n",
" <td>0.959128</td>\n",
" <td>1.756198</td>\n",
" <td>7.360224e-08</td>\n",
" <td>39.0</td>\n",
" </tr>\n",
" <tr>\n",
" <th>5</th>\n",
" <td>19860731</td>\n",
" <td>11734.593750</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>8.113567</td>\n",
" <td>NaN</td>\n",
" <td>0.476698</td>\n",
" <td>-0.005025</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.272432</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>...</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.049174</td>\n",
" <td>1.270483e-06</td>\n",
" <td>0.042553</td>\n",
" <td>0.020357</td>\n",
" <td>1.044263</td>\n",
" <td>1.239524</td>\n",
" <td>2.000000e+00</td>\n",
" <td>39.0</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>5 rows × 96 columns</p>\n",
"</div>"
],
"text/plain": [
" DATE mvel1 beta ... std_turn zerotrade sic2\n",
"1 19860331 11960.000000 NaN ... 1.079774 1.023392e-07 39.0\n",
"2 19860430 16330.000000 NaN ... 1.745333 7.467463e-08 39.0\n",
"3 19860530 15172.000000 NaN ... 1.502285 7.649551e-08 39.0\n",
"4 19860630 11793.859375 NaN ... 1.756198 7.360224e-08 39.0\n",
"5 19860731 11734.593750 NaN ... 1.239524 2.000000e+00 39.0\n",
"\n",
"[5 rows x 96 columns]"
]
},
"metadata": {
"tags": []
},
"execution_count": 22
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "UYZyjDCi-Cdm",
"colab_type": "code",
"colab": {}
},
"source": [
"for col in train_df.columns:\n",
" train_df[col]=train_df.groupby('DATE')[col].apply(lambda x:x.fillna(x.median()))"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "dWFbH0G1Npmh",
"colab_type": "text"
},
"source": [
"Some columns still have missing values. It is because those metrics were not recorded in early years.We filled thos missing value with the total mean of the column"
]
},
{
"cell_type": "code",
"metadata": {
"id": "PlqZ1xulFRtI",
"colab_type": "code",
"colab": {}
},
"source": [
"#replace all missing samples with 0\n",
"train_df.fillna(value=0, inplace=True)"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "YASkyyi5PUe1",
"colab_type": "code",
"colab": {}
},
"source": [
"ranked_train_df = train_df.groupby('DATE').rank(method='average', ascending=True, pct=True)"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "evs3I9eiPUe3",
"colab_type": "code",
"colab": {}
},
"source": [
"ranked_train_df = pd.concat([ranked_train_df, train_df['DATE']], axis=1)"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "I67q7qHctWho",
"colab_type": "code",
"colab": {}
},
"source": [
"ranked_train_df = pd.concat([ranked_train_df, tmp_df], axis=1)\n",
"train_df = pd.concat([train_df, tmp_df], axis=1)"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "ObmLNipKPUe5",
"colab_type": "text"
},
"source": [
"Finally, in this paper, the author rank-normalize the each cross-sectional characteristics. So we did the same. We make sure the sum of the second axis==1"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "_OMn7ZxhPUe7",
"colab_type": "text"
},
"source": [
"# 3. Model"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "eguSKm-MPUe8",
"colab_type": "text"
},
"source": [
"We build the model based on the PyTorch v1.4. The implementation is simple, and contains two classes.\n",
"* `Perceptron`: a sequence of Neural Network\n",
"* `Antoencoder`: The main model.\n",
"\n",
"There are several modifications:\n",
"* I am a bit confused about eq(16). Because Z is rank-normalied, so I think x = r*Z is enough.\n",
"* We can flexibly change the structure of beta nueral network by changing the parameters.\n",
"* The paper did not mention the initializer, so I just used the default setting\n",
"\n",
"Both are subclasses of torch.nn.Module. Please refer to my comments for details."
]
},
{
"cell_type": "code",
"metadata": {
"id": "oYbx_zecPUe9",
"colab_type": "code",
"colab": {}
},
"source": [
"class Perceptron(nn.Module):\n",
"\n",
" def __init__(self, in_, out_, hidden_):\n",
" \"\"\"\n",
" This is to build a multi-layer network\n",
" :param in_:int, input dimensions\n",
" :param out_:int, output dimensions\n",
" :param hidden_: list or tuple, # neurons for each hidden layer(exclude the input and output)\n",
" \"\"\"\n",
" super(Perceptron, self).__init__()\n",
" self.sequential = nn.Sequential()\n",
" for i in range(len(hidden_)+1): # output layer is not included in hidden_layer, so #layers = #hidden+1\n",
" # define the dimension of each layer\n",
" if i == 0:\n",
" input_ = in_\n",
" output_ = hidden_[i]\n",
" elif i == len(hidden_):\n",
" input_ = hidden_[i-1]\n",
" output_ = out_\n",
" else:\n",
" input_ = hidden_[i-1]\n",
" output_ = hidden_[i]\n",
"\n",
" # no batchnorm or the activation for the last layer\n",
" if i == len(hidden_):\n",
" self.sequential.add_module('linear'+str(i), nn.Linear(input_, output_))\n",
" else:\n",
" #dropout layer\n",
" #self.sequential.add_module('dropout'+str(i), nn.Dropout(p=0.3))\n",
" # add the linear layer\n",
" self.sequential.add_module('linear'+str(i), nn.Linear(input_, output_))\n",
" # add the batch norm layer\n",
" self.sequential.add_module('batchnorm'+str(i), nn.BatchNorm1d(output_))\n",
" #dropout layer\n",
" #self.sequential.add_module('dropout'+str(i), nn.Dropout(p=0.1))\n",
" # add the activation layer\n",
" self.sequential.add_module('relu'+str(i), nn.ReLU())\n",
"\n",
" def forward(self, x):\n",
" return self.sequential(x)\n",
"\n",
"class Linear(nn.Module):\n",
" def __init__(self, in_, out_):\n",
" \"\"\"\n",
" This is to build a linear network\n",
" :param in_:int, input dimensions\n",
" :param out_:int, output dimensions\n",
" \"\"\"\n",
" super(Linear, self).__init__()\n",
" self.linear = nn.Linear(in_, out_)\n",
" self.dropout = nn.Dropout(p=0.3)\n",
" def forward(self, x):\n",
" x = self.linear(x)\n",
" #x = self.dropout(x)\n",
" return x\n",
"\n",
"\n",
"class Autoencoder(nn.Module):\n",
"\n",
" def __init__(self, P, K, hidden_):\n",
" \"\"\"\n",
" This is to build the autoencoder neural netowrk with a multi-layer beta network and a single layer factor network\n",
" :param P:int, # characteristics\n",
" :param K:int, # output factors\n",
" :param hidden_: list or tuple, # neurons for each hidden layer for the beta network(exclude the input and output)\n",
" :param dropout_p: the dropout rate\n",
" \"\"\"\n",
" nn.Module.__init__(self)\n",
" self.beta_net = Perceptron(P, K, hidden_) # for beta nn, input Z: N*P, output BETA: N*K\n",
" self.factor_net = Linear(P, K) # for factor nn, only one linear layer, input r: N*1, output f: K*1\n",
"\n",
" def forward(self, z, r):\n",
" \"\"\"\n",
" :param z: N*P tensor\n",
" :param r: 1*N tensor\n",
" :return: P*1 tensor\n",
" \"\"\"\n",
" beta = self.beta_net(z)\n",
"\n",
" zz = t.mm(t.t(z), z) #P*P Matrix\n",
" zz_ = t.pinverse(zz)\n",
" zr = t.transpose(t.mm(r,z),0,1) # P*1\n",
" x = t.transpose(t.mm(zz_, zr),0,1) # 1*p\n",
" #x = t.mm(r, z)\n",
" factor = self.factor_net(x)\n",
"\n",
" return t.mm(factor, beta.t())"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "cIgj6JycPUe_",
"colab_type": "text"
},
"source": [
"# 4. Experiment"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "YKM3Aig0PUfA",
"colab_type": "text"
},
"source": [
"## 4.1 Training"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "15BNNnXBPUfA",
"colab_type": "text"
},
"source": [
"We trained the CA0 - CA3 mentioned in the paper. Below are functions for the training and evaluation process.\n",
"\n",
"I choose the 0-636th month as the training set, and the 636-677th as validation set. "
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "EdfdmNV5XnOW",
"colab_type": "text"
},
"source": [
"**Define** the evaluation function. "
]
},
{
"cell_type": "code",
"metadata": {
"id": "kwA7wes7Xgkp",
"colab_type": "code",
"colab": {}
},
"source": [
"def R_square(pred, target):\n",
"\n",
" return 1 - np.sum(np.square(target-pred)) / np.sum(np.square(target))"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "mjwmFuJev9Cm",
"colab_type": "code",
"colab": {}
},
"source": [
"ca1 = Autoencoder(95, 6, [32]).cuda()\n",
"\n",
"loss_fn = nn.MSELoss()\n",
"\n",
"decay = 1e-5 # weight decay for L1 LASSO.\n",
"\n",
"optimizer = optim.Adam(ca1.parameters(), lr=1e-3, betas=(0.9,0.999), eps=1e-8)\n",
"\n",
"epoch = 0\n",
"\n",
"if os.path.exists(\"./ck.pt\"):\n",
" checkpoint = t.load(\"./ck.pt\")\n",
" ca1.load_state_dict(checkpoint['model_state_dict'])\n",
" optimizer.load_state_dict(checkpoint['optimizer_state_dict'])\n",
" epoch = checkpoint['epoch']"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "10MpeO_pHqlG",
"colab_type": "code",
"colab": {}
},
"source": [
"def train(ca1, loader, optimizer, epoch, r_square_list):\n",
" '''\n",
" Function for training\n",
" '''\n",
" ca1.train()\n",
" for step, (batch_x, batch_y) in enumerate(loader):\n",
" optimizer.zero_grad()\n",
"\n",
" # prepare the data\n",
" z = batch_x\n",
" r = batch_y[np.newaxis, ...]\n",
" # forward\n",
" r_pred = ca1(z, r)\n",
" loss = loss_fn(r_pred, r)\n",
" for param in ca1.parameters():\n",
" loss += decay * t.sum(t.abs(param.float())) # torch has no integrated L1 regulizations, so I manually wrote this part\n",
"\n",
" loss.backward()\n",
" optimizer.step()\n",
" if step % 250 == 0:\n",
" print('Train Epoch: {} [{}/{} ({:.0f}%)]\\tLoss: {:.6f}'.format(\n",
" epoch, step * len(batch_x), len(loader.dataset),\n",
" 100. * step / len(loader), loss.item()))\n",
" r_square_list.append(R_square(r_pred.detach().cpu().numpy(), r.detach().cpu().numpy()))\n",
" print('Insample R^2 for iter: %d is %f' % (epoch+1, np.mean(r_square_list)))\n",
" print(\"\\n\")"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "p5h5yMN7I8Ie",
"colab_type": "code",
"colab": {}
},
"source": [
"def test(ca1, test_loader, epoch, r_square_list):\n",
" '''\n",
" Function for Cross-validation\n",
" '''\n",
" with t.no_grad():\n",
" ca1.eval()\n",
" test_loss=0\n",
" for step, (batch_x, batch_y) in enumerate(test_loader):\n",
" z = batch_x\n",
" r = batch_y[np.newaxis, ...]\n",
" r_pred = ca1(z, r)\n",
" r_square_list.append(R_square(r_pred.detach().cpu().numpy(), r.detach().cpu().numpy()))\n",
" test_loss += loss_fn(r_pred, r).item()\n",
" if step % 250 == 0:\n",
" print('Test set: Average loss: {:f}'.format(test_loss))\n",
" print('Test-sample R^2 for iter: %d is %f' % (epoch+1, np.mean(r_square_list)))\n",
" print(\"\\n\")"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "gTkaf8M1xDVR",
"colab_type": "code",
"colab": {}
},
"source": [
"def out(ca1, test_loader, epoch, r_square_list):\n",
" '''\n",
" Function for Cross-validation\n",
" '''\n",
" with t.no_grad():\n",
" ca1.eval()\n",
" test_loss=0\n",
" for step, (batch_x, batch_y) in enumerate(test_loader):\n",
" z = batch_x\n",
" r = batch_y[np.newaxis, ...]\n",
" r_pred = ca1(z, r)\n",
" r_square_list.append(R_square(r_pred.detach().cpu().numpy(), r.detach().cpu().numpy()))\n",
" print('Out-sample R^2 for iter: %d is %f' % (epoch+1, np.mean(r_square_list)))\n",
" print(\"\\n\")"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "IRoviu_Fuyls",
"colab_type": "code",
"colab": {}
},
"source": [
"def train_expansive(ca1, optimizer, epoch, r_square_list, test_rsq_list, out_rsq_list):\n",
" '''\n",
" Increase the training sample by one year for each refit\n",
" '''\n",
" begin_yr = 1957\n",
" train_begin_yr = 1974\n",
" train_end_yr = 2004\n",
" #Trainning dataset\n",
" for year_end in range(train_begin_yr, train_end_yr+1):\n",
" print(\"Training\" + \" \" + str(year_end)+\"\\n\")\n",
" begin_date = int(str(begin_yr)+\"0000\")\n",
" end_date = int(str(year_end)+\"0000\")\n",
" \n",
" X = ranked_train_df[(ranked_train_df['DATE'] <= end_date) & (ranked_train_df['DATE'] >= begin_date)].drop(['PERMNO', 'DATE', 'return'], axis=1).values\n",
" y = train_df[(train_df['DATE']<=end_date) & (train_df['DATE']>=begin_date)]['return'].values\n",
" X = t.from_numpy(X).float().cuda()\n",
" y = t.from_numpy(y).float().cuda()\n",
" torch_dataset = Data.TensorDataset(X, y)\n",
" \n",
" test_begin_year = year_end\n",
" test_end_year = test_begin_year + 12\n",
" test_begin_date = int(str(test_begin_year)+\"0000\")\n",
" test_end_date = int(str(test_end_year)+\"0000\")\n",
" \n",
" #Testing Dataset\n",
" X_test = ranked_train_df[(ranked_train_df['DATE']>test_begin_date) & (ranked_train_df['DATE']<=test_end_date) ].drop(['PERMNO', 'DATE', 'return'], axis=1).values\n",
" y_test = train_df[(train_df['DATE']>test_begin_date) & (train_df['DATE']<=test_end_date) ]['return'].values\n",
" X_test = t.from_numpy(X_test).float().cuda()\n",
" y_test = t.from_numpy(y_test).float().cuda()\n",
" test_dataset = Data.TensorDataset(X_test, y_test)\n",
"\n",
" out_begin_date = test_end_date\n",
" out_end_year = test_end_year + 1\n",
" out_end_date = int(str(out_end_year)+\"0000\")\n",
" #out of sample dataset\n",
" X_out = ranked_train_df[(ranked_train_df['DATE']>=out_begin_date) & ((ranked_train_df['DATE']<=(out_end_date+1))) ].drop(['PERMNO', 'DATE', 'return'], axis=1).values\n",
" y_out = train_df[(train_df['DATE']>=out_begin_date) & (train_df['DATE']<=out_end_date+1)]['return'].values\n",
" X_out = t.from_numpy(X_out).float().cuda()\n",
" y_out = t.from_numpy(y_out).float().cuda()\n",
" out_dataset = Data.TensorDataset(X_out, y_out)\n",
" \n",
" loader = Data.DataLoader(\n",
" dataset=torch_dataset,\n",
" batch_size=10000,\n",
" shuffle=True,\n",
" num_workers=0)\n",
" \n",
" #Test Sample Estimation Dataset\n",
" test_loader = Data.DataLoader(\n",
" dataset=test_dataset,\n",
" batch_size=len(test_dataset),\n",
" shuffle=False,\n",
" num_workers=0)\n",
" \n",
" #Out-of-sample Estimation Dataset\n",
" out_loader = Data.DataLoader(\n",
" dataset=out_dataset,\n",
" batch_size=len(out_dataset),\n",
" shuffle=False,\n",
" num_workers=0)\n",
"\n",
" train(ca1, loader, optimizer, epoch, r_square_list)\n",
" test(ca1, test_loader , epoch, test_rsq_list)\n",
" out(ca1, out_loader , epoch, out_rsq_list)"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "7NduRBhuK136",
"colab_type": "code",
"colab": {}
},
"source": [
"def save_checkpoint(ca1, optimizer, epoch):\n",
" t.save({\n",
" 'epoch': epoch,\n",
" 'model_state_dict': ca1.state_dict(),\n",
" 'optimizer_state_dict': optimizer.state_dict()\n",
" }, \"./ck.pt\")"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"scrolled": true,
"id": "aO9LkKtkPUfj",
"colab_type": "code",
"outputId": "8319cca2-58f7-405c-e2eb-ef6f1a442af8",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 1000
}
},
"source": [
"for epoch in range(epoch, 100):\n",
" r_square_list = []\n",
" test_rsq_list=[]\n",
" out_rsq_list=[]\n",
" train_expansive(ca1, optimizer, epoch, r_square_list, test_rsq_list, out_rsq_list)\n",
" #if epoch % 2 == 0: \n",
" print(\"Checkpointing ...\")\n",
" save_checkpoint(ca1, optimizer, epoch)"
],
"execution_count": 0,
"outputs": [
{
"output_type": "stream",
"text": [
"Training 1974\n",
"\n",
"Train Epoch: 14 [0/415222 (0%)]\tLoss: 0.011569\n",
"Insample R^2 for iter: 15 is 0.617356\n",
"\n",
"\n",
"Test set: Average loss: 0.007993\n",
"Test-sample R^2 for iter: 15 is 0.688269\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.643267\n",
"\n",
"\n",
"Training 1975\n",
"\n",
"Train Epoch: 14 [0/475849 (0%)]\tLoss: 0.006288\n",
"Insample R^2 for iter: 15 is 0.652540\n",
"\n",
"\n",
"Test set: Average loss: 0.008069\n",
"Test-sample R^2 for iter: 15 is 0.691451\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.613886\n",
"\n",
"\n",
"Training 1976\n",
"\n",
"Train Epoch: 14 [0/534401 (0%)]\tLoss: 0.006819\n",
"Insample R^2 for iter: 15 is 0.661362\n",
"\n",
"\n",
"Test set: Average loss: 0.008445\n",
"Test-sample R^2 for iter: 15 is 0.688222\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.632177\n",
"\n",
"\n",
"Training 1977\n",
"\n",
"Train Epoch: 14 [0/593250 (0%)]\tLoss: 0.006109\n",
"Insample R^2 for iter: 15 is 0.668915\n",
"\n",
"\n",
"Test set: Average loss: 0.007648\n",
"Test-sample R^2 for iter: 15 is 0.695320\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.657974\n",
"\n",
"\n",
"Training 1978\n",
"\n",
"Train Epoch: 14 [0/652479 (0%)]\tLoss: 0.006131\n",
"Insample R^2 for iter: 15 is 0.676912\n",
"\n",
"\n",
"Test set: Average loss: 0.007804\n",
"Test-sample R^2 for iter: 15 is 0.699633\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.662804\n",
"\n",
"\n",
"Training 1979\n",
"\n",
"Train Epoch: 14 [0/710435 (0%)]\tLoss: 0.005751\n",
"Insample R^2 for iter: 15 is 0.683200\n",
"\n",
"\n",
"Test set: Average loss: 0.008211\n",
"Test-sample R^2 for iter: 15 is 0.701923\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.663845\n",
"\n",
"\n",
"Training 1980\n",
"\n",
"Train Epoch: 14 [0/767839 (0%)]\tLoss: 0.005472\n",
"Insample R^2 for iter: 15 is 0.688910\n",
"\n",
"\n",
"Test set: Average loss: 0.009009\n",
"Test-sample R^2 for iter: 15 is 0.703611\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.660794\n",
"\n",
"\n",
"Training 1981\n",
"\n",
"Train Epoch: 14 [0/825704 (0%)]\tLoss: 0.006428\n",
"Insample R^2 for iter: 15 is 0.694098\n",
"\n",
"\n",
"Test set: Average loss: 0.010003\n",
"Test-sample R^2 for iter: 15 is 0.702782\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.661785\n",
"\n",
"\n",
"Training 1982\n",
"\n",
"Train Epoch: 14 [0/887496 (0%)]\tLoss: 0.006266\n",
"Insample R^2 for iter: 15 is 0.698593\n",
"\n",
"\n",
"Test set: Average loss: 0.010151\n",
"Test-sample R^2 for iter: 15 is 0.702431\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.670856\n",
"\n",
"\n",
"Training 1983\n",
"\n",
"Train Epoch: 14 [0/951438 (0%)]\tLoss: 0.005644\n",
"Insample R^2 for iter: 15 is 0.702945\n",
"\n",
"\n",
"Test set: Average loss: 0.009331\n",
"Test-sample R^2 for iter: 15 is 0.703921\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.683577\n",
"\n",
"\n",
"Training 1984\n",
"\n",
"Train Epoch: 14 [0/1019788 (0%)]\tLoss: 0.005661\n",
"Insample R^2 for iter: 15 is 0.707165\n",
"\n",
"\n",
"Test set: Average loss: 0.008525\n",
"Test-sample R^2 for iter: 15 is 0.706898\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.692728\n",
"\n",
"\n",
"Training 1985\n",
"\n",
"Train Epoch: 14 [0/1094741 (0%)]\tLoss: 0.005670\n",
"Insample R^2 for iter: 15 is 0.711299\n",
"\n",
"\n",
"Test set: Average loss: 0.008047\n",
"Test-sample R^2 for iter: 15 is 0.710963\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.702559\n",
"\n",
"\n",
"Training 1986\n",
"\n",
"Train Epoch: 14 [0/1169628 (0%)]\tLoss: 0.006286\n",
"Insample R^2 for iter: 15 is 0.715561\n",
"\n",
"\n",
"Test set: Average loss: 0.008295\n",
"Test-sample R^2 for iter: 15 is 0.713812\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.702776\n",
"\n",
"\n",
"Training 1987\n",
"\n",
"Train Epoch: 14 [0/1246063 (0%)]\tLoss: 0.006881\n",
"Insample R^2 for iter: 15 is 0.719060\n",
"\n",
"\n",
"Test set: Average loss: 0.008260\n",
"Test-sample R^2 for iter: 15 is 0.716959\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.707487\n",
"\n",
"\n",
"Training 1988\n",
"\n",
"Train Epoch: 14 [0/1328981 (0%)]\tLoss: 0.006765\n",
"Insample R^2 for iter: 15 is 0.721536\n",
"\n",
"\n",
"Test set: Average loss: 0.008570\n",
"Test-sample R^2 for iter: 15 is 0.719817\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.707443\n",
"\n",
"\n",
"Training 1989\n",
"\n",
"Train Epoch: 14 [0/1413142 (0%)]\tLoss: 0.005334\n",
"Insample R^2 for iter: 15 is 0.724378\n",
"\n",
"\n",
"Test set: Average loss: 0.010005\n",
"Test-sample R^2 for iter: 15 is 0.721292\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.705859\n",
"\n",
"\n",
"Training 1990\n",
"\n",
"Train Epoch: 14 [0/1494994 (0%)]\tLoss: 0.005765\n",
"Insample R^2 for iter: 15 is 0.727219\n",
"\n",
"\n",
"Test set: Average loss: 0.011330\n",
"Test-sample R^2 for iter: 15 is 0.721883\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.710861\n",
"\n",
"\n",
"Training 1991\n",
"\n",
"Train Epoch: 14 [0/1575765 (0%)]\tLoss: 0.005522\n",
"Insample R^2 for iter: 15 is 0.729890\n",
"\n",
"\n",
"Test set: Average loss: 0.010912\n",
"Test-sample R^2 for iter: 15 is 0.723081\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.713792\n",
"\n",
"\n",
"Training 1992\n",
"\n",
"Train Epoch: 14 [0/1655543 (0%)]\tLoss: 0.005916\n",
"Insample R^2 for iter: 15 is 0.732061\n",
"\n",
"\n",
"Test set: Average loss: 0.010955\n",
"Test-sample R^2 for iter: 15 is 0.723646\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.715586\n",
"\n",
"\n",
"Training 1993\n",
"\n",
"Train Epoch: 14 [0/1737145 (0%)]\tLoss: 0.006173\n",
"Insample R^2 for iter: 15 is 0.733718\n",
"\n",
"\n",
"Test set: Average loss: 0.009929\n",
"Test-sample R^2 for iter: 15 is 0.724811\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.711324\n",
"\n",
"\n",
"Training 1994\n",
"\n",
"Train Epoch: 14 [0/1823681 (0%)]\tLoss: 0.006492\n",
"Insample R^2 for iter: 15 is 0.735336\n",
"\n",
"\n",
"Test set: Average loss: 0.009980\n",
"Test-sample R^2 for iter: 15 is 0.725385\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.706335\n",
"\n",
"\n",
"Training 1995\n",
"\n",
"Train Epoch: 14 [0/1919298 (0%)]\tLoss: 0.006853\n",
"Insample R^2 for iter: 15 is 0.736903\n",
"\n",
"\n",
"Test set: Average loss: 0.010112\n",
"Test-sample R^2 for iter: 15 is 0.725669\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.700626\n",
"\n",
"\n",
"Training 1996\n",
"\n",
"Train Epoch: 14 [0/2017668 (0%)]\tLoss: 0.005218\n",
"Insample R^2 for iter: 15 is 0.738746\n",
"\n",
"\n",
"Test set: Average loss: 0.010075\n",
"Test-sample R^2 for iter: 15 is 0.725826\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.701528\n",
"\n",
"\n",
"Training 1997\n",
"\n",
"Train Epoch: 14 [0/2121588 (0%)]\tLoss: 0.006356\n",
"Insample R^2 for iter: 15 is 0.740611\n",
"\n",
"\n",
"Test set: Average loss: 0.010434\n",
"Test-sample R^2 for iter: 15 is 0.725676\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.700731\n",
"\n",
"\n",
"Training 1998\n",
"\n",
"Train Epoch: 14 [0/2229613 (0%)]\tLoss: 0.007510\n",
"Insample R^2 for iter: 15 is 0.742498\n",
"\n",
"\n",
"Test set: Average loss: 0.011179\n",
"Test-sample R^2 for iter: 15 is 0.725499\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.702534\n",
"\n",
"\n",
"Training 1999\n",
"\n",
"Train Epoch: 14 [0/2336372 (0%)]\tLoss: 0.007796\n",
"Insample R^2 for iter: 15 is 0.744068\n",
"\n",
"\n",
"Test set: Average loss: 0.010764\n",
"Test-sample R^2 for iter: 15 is 0.725260\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.703023\n",
"\n",
"\n",
"Training 2000\n",
"\n",
"Train Epoch: 14 [0/2437321 (0%)]\tLoss: 0.005948\n",
"Insample R^2 for iter: 15 is 0.745482\n",
"\n",
"\n",
"Test set: Average loss: 0.010367\n",
"Test-sample R^2 for iter: 15 is 0.724660\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.703292\n",
"\n",
"\n",
"Training 2001\n",
"\n",
"Train Epoch: 14 [0/2536121 (0%)]\tLoss: 0.007171\n",
"Train Epoch: 14 [2500000/2536121 (98%)]\tLoss: 0.007451\n",
"Insample R^2 for iter: 15 is 0.746591\n",
"\n",
"\n",
"Test set: Average loss: 0.009239\n",
"Test-sample R^2 for iter: 15 is 0.723877\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.698854\n",
"\n",
"\n",
"Training 2002\n",
"\n",
"Train Epoch: 14 [0/2628379 (0%)]\tLoss: 0.006617\n",
"Train Epoch: 14 [2500000/2628379 (95%)]\tLoss: 0.008679\n",
"Insample R^2 for iter: 15 is 0.747305\n",
"\n",
"\n",
"Test set: Average loss: 0.008068\n",
"Test-sample R^2 for iter: 15 is 0.722980\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.696642\n",
"\n",
"\n",
"Training 2003\n",
"\n",
"Train Epoch: 14 [0/2714364 (0%)]\tLoss: 0.006816\n",
"Train Epoch: 14 [2500000/2714364 (92%)]\tLoss: 0.017163\n",
"Insample R^2 for iter: 15 is 0.748236\n",
"\n",
"\n",
"Test set: Average loss: 0.008446\n",
"Test-sample R^2 for iter: 15 is 0.720709\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.696246\n",
"\n",
"\n",
"Training 2004\n",
"\n",
"Train Epoch: 14 [0/2795471 (0%)]\tLoss: 0.007348\n",
"Train Epoch: 14 [2500000/2795471 (89%)]\tLoss: 0.007906\n",
"Insample R^2 for iter: 15 is 0.749230\n",
"\n",
"\n",
"Test set: Average loss: 0.009132\n",
"Test-sample R^2 for iter: 15 is 0.717184\n",
"\n",
"\n",
"Out-sample R^2 for iter: 15 is 0.696535\n",
"\n",
"\n",
"Checkpointing ...\n",
"Training 1974\n",
"\n",
"Train Epoch: 15 [0/415222 (0%)]\tLoss: 0.016229\n",
"Insample R^2 for iter: 16 is 0.596179\n",
"\n",
"\n",
"Test set: Average loss: 0.008213\n",
"Test-sample R^2 for iter: 16 is 0.679717\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.652917\n",
"\n",
"\n",
"Training 1975\n",
"\n",
"Train Epoch: 15 [0/475849 (0%)]\tLoss: 0.005822\n",
"Insample R^2 for iter: 16 is 0.641223\n",
"\n",
"\n",
"Test set: Average loss: 0.007369\n",
"Test-sample R^2 for iter: 16 is 0.700421\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.640637\n",
"\n",
"\n",
"Training 1976\n",
"\n",
"Train Epoch: 15 [0/534401 (0%)]\tLoss: 0.006189\n",
"Insample R^2 for iter: 16 is 0.652780\n",
"\n",
"\n",
"Test set: Average loss: 0.007274\n",
"Test-sample R^2 for iter: 16 is 0.708915\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.677407\n",
"\n",
"\n",
"Training 1977\n",
"\n",
"Train Epoch: 15 [0/593250 (0%)]\tLoss: 0.006609\n",
"Insample R^2 for iter: 16 is 0.662499\n",
"\n",
"\n",
"Test set: Average loss: 0.007296\n",
"Test-sample R^2 for iter: 16 is 0.714092\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.705850\n",
"\n",
"\n",
"Training 1978\n",
"\n",
"Train Epoch: 15 [0/652479 (0%)]\tLoss: 0.005802\n",
"Insample R^2 for iter: 16 is 0.671816\n",
"\n",
"\n",
"Test set: Average loss: 0.007312\n",
"Test-sample R^2 for iter: 16 is 0.718224\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.709849\n",
"\n",
"\n",
"Training 1979\n",
"\n",
"Train Epoch: 15 [0/710435 (0%)]\tLoss: 0.005861\n",
"Insample R^2 for iter: 16 is 0.679044\n",
"\n",
"\n",
"Test set: Average loss: 0.007560\n",
"Test-sample R^2 for iter: 16 is 0.721208\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.704147\n",
"\n",
"\n",
"Training 1980\n",
"\n",
"Train Epoch: 15 [0/767839 (0%)]\tLoss: 0.005730\n",
"Insample R^2 for iter: 16 is 0.685609\n",
"\n",
"\n",
"Test set: Average loss: 0.008584\n",
"Test-sample R^2 for iter: 16 is 0.722069\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.697339\n",
"\n",
"\n",
"Training 1981\n",
"\n",
"Train Epoch: 15 [0/825704 (0%)]\tLoss: 0.006041\n",
"Insample R^2 for iter: 16 is 0.691242\n",
"\n",
"\n",
"Test set: Average loss: 0.009352\n",
"Test-sample R^2 for iter: 16 is 0.721400\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.697469\n",
"\n",
"\n",
"Training 1982\n",
"\n",
"Train Epoch: 15 [0/887496 (0%)]\tLoss: 0.005690\n",
"Insample R^2 for iter: 16 is 0.696001\n",
"\n",
"\n",
"Test set: Average loss: 0.009655\n",
"Test-sample R^2 for iter: 16 is 0.720609\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.704799\n",
"\n",
"\n",
"Training 1983\n",
"\n",
"Train Epoch: 15 [0/951438 (0%)]\tLoss: 0.005975\n",
"Insample R^2 for iter: 16 is 0.700804\n",
"\n",
"\n",
"Test set: Average loss: 0.008636\n",
"Test-sample R^2 for iter: 16 is 0.722387\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.714922\n",
"\n",
"\n",
"Training 1984\n",
"\n",
"Train Epoch: 15 [0/1019788 (0%)]\tLoss: 0.007610\n",
"Insample R^2 for iter: 16 is 0.705358\n",
"\n",
"\n",
"Test set: Average loss: 0.008006\n",
"Test-sample R^2 for iter: 16 is 0.725143\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.721045\n",
"\n",
"\n",
"Training 1985\n",
"\n",
"Train Epoch: 15 [0/1094741 (0%)]\tLoss: 0.005905\n",
"Insample R^2 for iter: 16 is 0.709785\n",
"\n",
"\n",
"Test set: Average loss: 0.008153\n",
"Test-sample R^2 for iter: 16 is 0.727421\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.728032\n",
"\n",
"\n",
"Training 1986\n",
"\n",
"Train Epoch: 15 [0/1169628 (0%)]\tLoss: 0.005482\n",
"Insample R^2 for iter: 16 is 0.714182\n",
"\n",
"\n",
"Test set: Average loss: 0.008187\n",
"Test-sample R^2 for iter: 16 is 0.729257\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.727141\n",
"\n",
"\n",
"Training 1987\n",
"\n",
"Train Epoch: 15 [0/1246063 (0%)]\tLoss: 0.005484\n",
"Insample R^2 for iter: 16 is 0.718108\n",
"\n",
"\n",
"Test set: Average loss: 0.008477\n",
"Test-sample R^2 for iter: 16 is 0.730846\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.729114\n",
"\n",
"\n",
"Training 1988\n",
"\n",
"Train Epoch: 15 [0/1328981 (0%)]\tLoss: 0.006215\n",
"Insample R^2 for iter: 16 is 0.720674\n",
"\n",
"\n",
"Test set: Average loss: 0.008449\n",
"Test-sample R^2 for iter: 16 is 0.733003\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.728888\n",
"\n",
"\n",
"Training 1989\n",
"\n",
"Train Epoch: 15 [0/1413142 (0%)]\tLoss: 0.006362\n",
"Insample R^2 for iter: 16 is 0.723482\n",
"\n",
"\n",
"Test set: Average loss: 0.010596\n",
"Test-sample R^2 for iter: 16 is 0.732706\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.725558\n",
"\n",
"\n",
"Training 1990\n",
"\n",
"Train Epoch: 15 [0/1494994 (0%)]\tLoss: 0.005113\n",
"Insample R^2 for iter: 16 is 0.726410\n",
"\n",
"\n",
"Test set: Average loss: 0.010977\n",
"Test-sample R^2 for iter: 16 is 0.733118\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.730109\n",
"\n",
"\n",
"Training 1991\n",
"\n",
"Train Epoch: 15 [0/1575765 (0%)]\tLoss: 0.007434\n",
"Insample R^2 for iter: 16 is 0.729158\n",
"\n",
"\n",
"Test set: Average loss: 0.011257\n",
"Test-sample R^2 for iter: 16 is 0.733241\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.731873\n",
"\n",
"\n",
"Training 1992\n",
"\n",
"Train Epoch: 15 [0/1655543 (0%)]\tLoss: 0.006213\n",
"Insample R^2 for iter: 16 is 0.731485\n",
"\n",
"\n",
"Test set: Average loss: 0.010529\n",
"Test-sample R^2 for iter: 16 is 0.733815\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.732582\n",
"\n",
"\n",
"Training 1993\n",
"\n",
"Train Epoch: 15 [0/1737145 (0%)]\tLoss: 0.006189\n",
"Insample R^2 for iter: 16 is 0.733293\n",
"\n",
"\n",
"Test set: Average loss: 0.009694\n",
"Test-sample R^2 for iter: 16 is 0.734772\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.727377\n",
"\n",
"\n",
"Training 1994\n",
"\n",
"Train Epoch: 15 [0/1823681 (0%)]\tLoss: 0.006569\n",
"Insample R^2 for iter: 16 is 0.735053\n",
"\n",
"\n",
"Test set: Average loss: 0.009952\n",
"Test-sample R^2 for iter: 16 is 0.734906\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.722018\n",
"\n",
"\n",
"Training 1995\n",
"\n",
"Train Epoch: 15 [0/1919298 (0%)]\tLoss: 0.006197\n",
"Insample R^2 for iter: 16 is 0.736618\n",
"\n",
"\n",
"Test set: Average loss: 0.009815\n",
"Test-sample R^2 for iter: 16 is 0.735115\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.712213\n",
"\n",
"\n",
"Training 1996\n",
"\n",
"Train Epoch: 15 [0/2017668 (0%)]\tLoss: 0.005896\n",
"Insample R^2 for iter: 16 is 0.738455\n",
"\n",
"\n",
"Test set: Average loss: 0.009724\n",
"Test-sample R^2 for iter: 16 is 0.735271\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.711400\n",
"\n",
"\n",
"Training 1997\n",
"\n",
"Train Epoch: 15 [0/2121588 (0%)]\tLoss: 0.006231\n",
"Insample R^2 for iter: 16 is 0.740348\n",
"\n",
"\n",
"Test set: Average loss: 0.010266\n",
"Test-sample R^2 for iter: 16 is 0.734914\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.711108\n",
"\n",
"\n",
"Training 1998\n",
"\n",
"Train Epoch: 15 [0/2229613 (0%)]\tLoss: 0.005627\n",
"Insample R^2 for iter: 16 is 0.742131\n",
"\n",
"\n",
"Test set: Average loss: 0.011199\n",
"Test-sample R^2 for iter: 16 is 0.734348\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.712873\n",
"\n",
"\n",
"Training 1999\n",
"\n",
"Train Epoch: 15 [0/2336372 (0%)]\tLoss: 0.007626\n",
"Insample R^2 for iter: 16 is 0.743705\n",
"\n",
"\n",
"Test set: Average loss: 0.011136\n",
"Test-sample R^2 for iter: 16 is 0.733395\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.713967\n",
"\n",
"\n",
"Training 2000\n",
"\n",
"Train Epoch: 15 [0/2437321 (0%)]\tLoss: 0.009701\n",
"Insample R^2 for iter: 16 is 0.745249\n",
"\n",
"\n",
"Test set: Average loss: 0.010423\n",
"Test-sample R^2 for iter: 16 is 0.732435\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.713190\n",
"\n",
"\n",
"Training 2001\n",
"\n",
"Train Epoch: 15 [0/2536121 (0%)]\tLoss: 0.006253\n",
"Train Epoch: 15 [2500000/2536121 (98%)]\tLoss: 0.006635\n",
"Insample R^2 for iter: 16 is 0.746375\n",
"\n",
"\n",
"Test set: Average loss: 0.009706\n",
"Test-sample R^2 for iter: 16 is 0.730839\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.705508\n",
"\n",
"\n",
"Training 2002\n",
"\n",
"Train Epoch: 15 [0/2628379 (0%)]\tLoss: 0.009370\n",
"Train Epoch: 15 [2500000/2628379 (95%)]\tLoss: 0.008954\n",
"Insample R^2 for iter: 16 is 0.747238\n",
"\n",
"\n",
"Test set: Average loss: 0.009868\n",
"Test-sample R^2 for iter: 16 is 0.727378\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.695985\n",
"\n",
"\n",
"Training 2003\n",
"\n",
"Train Epoch: 15 [0/2714364 (0%)]\tLoss: 0.006326\n",
"Train Epoch: 15 [2500000/2714364 (92%)]\tLoss: 0.006599\n",
"Insample R^2 for iter: 16 is 0.748260\n",
"\n",
"\n",
"Test set: Average loss: 0.008335\n",
"Test-sample R^2 for iter: 16 is 0.725110\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.696579\n",
"\n",
"\n",
"Training 2004\n",
"\n",
"Train Epoch: 15 [0/2795471 (0%)]\tLoss: 0.006271\n",
"Train Epoch: 15 [2500000/2795471 (89%)]\tLoss: 0.008344\n",
"Insample R^2 for iter: 16 is 0.749210\n",
"\n",
"\n",
"Test set: Average loss: 0.008609\n",
"Test-sample R^2 for iter: 16 is 0.722161\n",
"\n",
"\n",
"Out-sample R^2 for iter: 16 is 0.697299\n",
"\n",
"\n",
"Checkpointing ...\n",
"Training 1974\n",
"\n",
"Train Epoch: 16 [0/415222 (0%)]\tLoss: 0.011671\n",
"Insample R^2 for iter: 17 is 0.617919\n",
"\n",
"\n",
"Test set: Average loss: 0.007431\n",
"Test-sample R^2 for iter: 17 is 0.710187\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.680282\n",
"\n",
"\n",
"Training 1975\n",
"\n",
"Train Epoch: 16 [0/475849 (0%)]\tLoss: 0.005747\n",
"Insample R^2 for iter: 17 is 0.652067\n",
"\n",
"\n",
"Test set: Average loss: 0.007532\n",
"Test-sample R^2 for iter: 17 is 0.712574\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.660646\n",
"\n",
"\n",
"Training 1976\n",
"\n",
"Train Epoch: 16 [0/534401 (0%)]\tLoss: 0.007221\n",
"Insample R^2 for iter: 17 is 0.660394\n",
"\n",
"\n",
"Test set: Average loss: 0.007421\n",
"Test-sample R^2 for iter: 17 is 0.715159\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.687548\n",
"\n",
"\n",
"Training 1977\n",
"\n",
"Train Epoch: 16 [0/593250 (0%)]\tLoss: 0.006688\n",
"Insample R^2 for iter: 17 is 0.668336\n",
"\n",
"\n",
"Test set: Average loss: 0.007469\n",
"Test-sample R^2 for iter: 17 is 0.717180\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.708889\n",
"\n",
"\n",
"Training 1978\n",
"\n",
"Train Epoch: 16 [0/652479 (0%)]\tLoss: 0.005815\n",
"Insample R^2 for iter: 17 is 0.675973\n",
"\n",
"\n",
"Test set: Average loss: 0.007344\n",
"Test-sample R^2 for iter: 17 is 0.720464\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.711104\n",
"\n",
"\n",
"Training 1979\n",
"\n",
"Train Epoch: 16 [0/710435 (0%)]\tLoss: 0.006054\n",
"Insample R^2 for iter: 17 is 0.682130\n",
"\n",
"\n",
"Test set: Average loss: 0.008056\n",
"Test-sample R^2 for iter: 17 is 0.720186\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.705108\n",
"\n",
"\n",
"Training 1980\n",
"\n",
"Train Epoch: 16 [0/767839 (0%)]\tLoss: 0.005431\n",
"Insample R^2 for iter: 17 is 0.688093\n",
"\n",
"\n",
"Test set: Average loss: 0.009356\n",
"Test-sample R^2 for iter: 17 is 0.717690\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.697241\n",
"\n",
"\n",
"Training 1981\n",
"\n",
"Train Epoch: 16 [0/825704 (0%)]\tLoss: 0.005897\n",
"Insample R^2 for iter: 17 is 0.693342\n",
"\n",
"\n",
"Test set: Average loss: 0.010149\n",
"Test-sample R^2 for iter: 17 is 0.714548\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.692611\n",
"\n",
"\n",
"Training 1982\n",
"\n",
"Train Epoch: 16 [0/887496 (0%)]\tLoss: 0.006201\n",
"Insample R^2 for iter: 17 is 0.697738\n",
"\n",
"\n",
"Test set: Average loss: 0.010297\n",
"Test-sample R^2 for iter: 17 is 0.712409\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.695844\n",
"\n",
"\n",
"Training 1983\n",
"\n",
"Train Epoch: 16 [0/951438 (0%)]\tLoss: 0.005856\n",
"Insample R^2 for iter: 17 is 0.702159\n",
"\n",
"\n",
"Test set: Average loss: 0.008913\n",
"Test-sample R^2 for iter: 17 is 0.714166\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.706587\n",
"\n",
"\n",
"Training 1984\n",
"\n",
"Train Epoch: 16 [0/1019788 (0%)]\tLoss: 0.005808\n",
"Insample R^2 for iter: 17 is 0.706404\n",
"\n",
"\n",
"Test set: Average loss: 0.008483\n",
"Test-sample R^2 for iter: 17 is 0.716329\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.713243\n",
"\n",
"\n",
"Training 1985\n",
"\n",
"Train Epoch: 16 [0/1094741 (0%)]\tLoss: 0.008782\n",
"Insample R^2 for iter: 17 is 0.710635\n",
"\n",
"\n",
"Test set: Average loss: 0.008236\n",
"Test-sample R^2 for iter: 17 is 0.719129\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.720551\n",
"\n",
"\n",
"Training 1986\n",
"\n",
"Train Epoch: 16 [0/1169628 (0%)]\tLoss: 0.005556\n",
"Insample R^2 for iter: 17 is 0.714954\n",
"\n",
"\n",
"Test set: Average loss: 0.007884\n",
"Test-sample R^2 for iter: 17 is 0.722310\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.719886\n",
"\n",
"\n",
"Training 1987\n",
"\n",
"Train Epoch: 16 [0/1246063 (0%)]\tLoss: 0.006290\n",
"Insample R^2 for iter: 17 is 0.718784\n",
"\n",
"\n",
"Test set: Average loss: 0.008238\n",
"Test-sample R^2 for iter: 17 is 0.724896\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.723765\n",
"\n",
"\n",
"Training 1988\n",
"\n",
"Train Epoch: 16 [0/1328981 (0%)]\tLoss: 0.006689\n",
"Insample R^2 for iter: 17 is 0.721310\n",
"\n",
"\n",
"Test set: Average loss: 0.008596\n",
"Test-sample R^2 for iter: 17 is 0.727176\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.722782\n",
"\n",
"\n",
"Training 1989\n",
"\n",
"Train Epoch: 16 [0/1413142 (0%)]\tLoss: 0.006059\n",
"Insample R^2 for iter: 17 is 0.723995\n",
"\n",
"\n",
"Test set: Average loss: 0.009316\n",
"Test-sample R^2 for iter: 17 is 0.729296\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.722111\n",
"\n",
"\n",
"Training 1990\n",
"\n",
"Train Epoch: 16 [0/1494994 (0%)]\tLoss: 0.005832\n",
"Insample R^2 for iter: 17 is 0.727031\n",
"\n",
"\n",
"Test set: Average loss: 0.011136\n",
"Test-sample R^2 for iter: 17 is 0.729687\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.726566\n",
"\n",
"\n",
"Training 1991\n",
"\n",
"Train Epoch: 16 [0/1575765 (0%)]\tLoss: 0.008340\n",
"Insample R^2 for iter: 17 is 0.729896\n",
"\n",
"\n",
"Test set: Average loss: 0.011511\n",
"Test-sample R^2 for iter: 17 is 0.729669\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.726969\n",
"\n",
"\n",
"Training 1992\n",
"\n",
"Train Epoch: 16 [0/1655543 (0%)]\tLoss: 0.010153\n",
"Insample R^2 for iter: 17 is 0.732050\n",
"\n",
"\n",
"Test set: Average loss: 0.010629\n",
"Test-sample R^2 for iter: 17 is 0.730303\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.728494\n",
"\n",
"\n",
"Training 1993\n",
"\n",
"Train Epoch: 16 [0/1737145 (0%)]\tLoss: 0.005272\n",
"Insample R^2 for iter: 17 is 0.734042\n",
"\n",
"\n",
"Test set: Average loss: 0.009946\n",
"Test-sample R^2 for iter: 17 is 0.731113\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.724435\n",
"\n",
"\n",
"Training 1994\n",
"\n",
"Train Epoch: 16 [0/1823681 (0%)]\tLoss: 0.006979\n",
"Insample R^2 for iter: 17 is 0.735739\n",
"\n",
"\n",
"Test set: Average loss: 0.009502\n",
"Test-sample R^2 for iter: 17 is 0.731986\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.719590\n",
"\n",
"\n",
"Training 1995\n",
"\n",
"Train Epoch: 16 [0/1919298 (0%)]\tLoss: 0.007171\n",
"Insample R^2 for iter: 17 is 0.737391\n",
"\n",
"\n",
"Test set: Average loss: 0.009518\n",
"Test-sample R^2 for iter: 17 is 0.732687\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.711998\n",
"\n",
"\n",
"Training 1996\n",
"\n",
"Train Epoch: 16 [0/2017668 (0%)]\tLoss: 0.005789\n",
"Insample R^2 for iter: 17 is 0.739181\n",
"\n",
"\n",
"Test set: Average loss: 0.010064\n",
"Test-sample R^2 for iter: 17 is 0.732551\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.711741\n",
"\n",
"\n",
"Training 1997\n",
"\n",
"Train Epoch: 16 [0/2121588 (0%)]\tLoss: 0.005559\n",
"Insample R^2 for iter: 17 is 0.740910\n",
"\n",
"\n",
"Test set: Average loss: 0.010159\n",
"Test-sample R^2 for iter: 17 is 0.732425\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.710680\n",
"\n",
"\n",
"Training 1998\n",
"\n",
"Train Epoch: 16 [0/2229613 (0%)]\tLoss: 0.005209\n",
"Insample R^2 for iter: 17 is 0.742749\n",
"\n",
"\n",
"Test set: Average loss: 0.011604\n",
"Test-sample R^2 for iter: 17 is 0.731555\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.712418\n",
"\n",
"\n",
"Training 1999\n",
"\n",
"Train Epoch: 16 [0/2336372 (0%)]\tLoss: 0.007230\n",
"Insample R^2 for iter: 17 is 0.744170\n",
"\n",
"\n",
"Test set: Average loss: 0.010768\n",
"Test-sample R^2 for iter: 17 is 0.731079\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.712667\n",
"\n",
"\n",
"Training 2000\n",
"\n",
"Train Epoch: 16 [0/2437321 (0%)]\tLoss: 0.005961\n",
"Insample R^2 for iter: 17 is 0.745606\n",
"\n",
"\n",
"Test set: Average loss: 0.010427\n",
"Test-sample R^2 for iter: 17 is 0.730201\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.714656\n",
"\n",
"\n",
"Training 2001\n",
"\n",
"Train Epoch: 16 [0/2536121 (0%)]\tLoss: 0.008125\n",
"Train Epoch: 16 [2500000/2536121 (98%)]\tLoss: 0.007054\n",
"Insample R^2 for iter: 17 is 0.746838\n",
"\n",
"\n",
"Test set: Average loss: 0.010008\n",
"Test-sample R^2 for iter: 17 is 0.728337\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.705928\n",
"\n",
"\n",
"Training 2002\n",
"\n",
"Train Epoch: 16 [0/2628379 (0%)]\tLoss: 0.007166\n",
"Train Epoch: 16 [2500000/2628379 (95%)]\tLoss: 0.013044\n",
"Insample R^2 for iter: 17 is 0.747739\n",
"\n",
"\n",
"Test set: Average loss: 0.008582\n",
"Test-sample R^2 for iter: 17 is 0.726621\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.701114\n",
"\n",
"\n",
"Training 2003\n",
"\n",
"Train Epoch: 16 [0/2714364 (0%)]\tLoss: 0.007160\n",
"Train Epoch: 16 [2500000/2714364 (92%)]\tLoss: 0.006645\n",
"Insample R^2 for iter: 17 is 0.748675\n",
"\n",
"\n",
"Test set: Average loss: 0.009025\n",
"Test-sample R^2 for iter: 17 is 0.723440\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.700162\n",
"\n",
"\n",
"Training 2004\n",
"\n",
"Train Epoch: 16 [0/2795471 (0%)]\tLoss: 0.006950\n",
"Train Epoch: 16 [2500000/2795471 (89%)]\tLoss: 0.007957\n",
"Insample R^2 for iter: 17 is 0.749578\n",
"\n",
"\n",
"Test set: Average loss: 0.007456\n",
"Test-sample R^2 for iter: 17 is 0.722127\n",
"\n",
"\n",
"Out-sample R^2 for iter: 17 is 0.701730\n",
"\n",
"\n",
"Checkpointing ...\n",
"Training 1974\n",
"\n",
"Train Epoch: 17 [0/415222 (0%)]\tLoss: 0.010624\n",
"Insample R^2 for iter: 18 is 0.622937\n",
"\n",
"\n",
"Test set: Average loss: 0.007826\n",
"Test-sample R^2 for iter: 18 is 0.694793\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.678417\n",
"\n",
"\n",
"Training 1975\n",
"\n",
"Train Epoch: 17 [0/475849 (0%)]\tLoss: 0.005431\n",
"Insample R^2 for iter: 18 is 0.656737\n",
"\n",
"\n",
"Test set: Average loss: 0.008508\n",
"Test-sample R^2 for iter: 18 is 0.686401\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.658765\n",
"\n",
"\n",
"Training 1976\n",
"\n",
"Train Epoch: 17 [0/534401 (0%)]\tLoss: 0.005746\n",
"Insample R^2 for iter: 18 is 0.663918\n",
"\n",
"\n",
"Test set: Average loss: 0.007322\n",
"Test-sample R^2 for iter: 18 is 0.698963\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.689965\n",
"\n",
"\n",
"Training 1977\n",
"\n",
"Train Epoch: 17 [0/593250 (0%)]\tLoss: 0.006706\n",
"Insample R^2 for iter: 18 is 0.671306\n",
"\n",
"\n",
"Test set: Average loss: 0.007363\n",
"Test-sample R^2 for iter: 18 is 0.706011\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.711039\n",
"\n",
"\n",
"Training 1978\n",
"\n",
"Train Epoch: 17 [0/652479 (0%)]\tLoss: 0.006035\n",
"Insample R^2 for iter: 18 is 0.678732\n",
"\n",
"\n",
"Test set: Average loss: 0.007670\n",
"Test-sample R^2 for iter: 18 is 0.709158\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.714086\n",
"\n",
"\n",
"Training 1979\n",
"\n",
"Train Epoch: 17 [0/710435 (0%)]\tLoss: 0.005475\n",
"Insample R^2 for iter: 18 is 0.683485\n",
"\n",
"\n",
"Test set: Average loss: 0.007715\n",
"Test-sample R^2 for iter: 18 is 0.712748\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.705872\n",
"\n",
"\n",
"Training 1980\n",
"\n",
"Train Epoch: 17 [0/767839 (0%)]\tLoss: 0.006132\n",
"Insample R^2 for iter: 18 is 0.688647\n",
"\n",
"\n",
"Test set: Average loss: 0.008607\n",
"Test-sample R^2 for iter: 18 is 0.714715\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.703662\n",
"\n",
"\n",
"Training 1981\n",
"\n",
"Train Epoch: 17 [0/825704 (0%)]\tLoss: 0.005659\n",
"Insample R^2 for iter: 18 is 0.693737\n",
"\n",
"\n",
"Test set: Average loss: 0.009928\n",
"Test-sample R^2 for iter: 18 is 0.712783\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.700581\n",
"\n",
"\n",
"Training 1982\n",
"\n",
"Train Epoch: 17 [0/887496 (0%)]\tLoss: 0.005109\n",
"Insample R^2 for iter: 18 is 0.698065\n",
"\n",
"\n",
"Test set: Average loss: 0.010102\n",
"Test-sample R^2 for iter: 18 is 0.711482\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.705944\n",
"\n",
"\n",
"Training 1983\n",
"\n",
"Train Epoch: 17 [0/951438 (0%)]\tLoss: 0.005880\n",
"Insample R^2 for iter: 18 is 0.702486\n",
"\n",
"\n",
"Test set: Average loss: 0.008890\n",
"Test-sample R^2 for iter: 18 is 0.713402\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.716236\n",
"\n",
"\n",
"Training 1984\n",
"\n",
"Train Epoch: 17 [0/1019788 (0%)]\tLoss: 0.005720\n",
"Insample R^2 for iter: 18 is 0.707032\n",
"\n",
"\n",
"Test set: Average loss: 0.008484\n",
"Test-sample R^2 for iter: 18 is 0.715630\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.721724\n",
"\n",
"\n",
"Training 1985\n",
"\n",
"Train Epoch: 17 [0/1094741 (0%)]\tLoss: 0.005641\n",
"Insample R^2 for iter: 18 is 0.711422\n",
"\n",
"\n",
"Test set: Average loss: 0.007832\n",
"Test-sample R^2 for iter: 18 is 0.719512\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.728565\n",
"\n",
"\n",
"Training 1986\n",
"\n",
"Train Epoch: 17 [0/1169628 (0%)]\tLoss: 0.005968\n",
"Insample R^2 for iter: 18 is 0.715670\n",
"\n",
"\n",
"Test set: Average loss: 0.007596\n",
"Test-sample R^2 for iter: 18 is 0.723338\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.728012\n",
"\n",
"\n",
"Training 1987\n",
"\n",
"Train Epoch: 17 [0/1246063 (0%)]\tLoss: 0.021119\n",
"Insample R^2 for iter: 18 is 0.719484\n",
"\n",
"\n",
"Test set: Average loss: 0.008902\n",
"Test-sample R^2 for iter: 18 is 0.724459\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.729408\n",
"\n",
"\n",
"Training 1988\n",
"\n",
"Train Epoch: 17 [0/1328981 (0%)]\tLoss: 0.005822\n",
"Insample R^2 for iter: 18 is 0.722084\n",
"\n",
"\n",
"Test set: Average loss: 0.008877\n",
"Test-sample R^2 for iter: 18 is 0.726244\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.728741\n",
"\n",
"\n",
"Training 1989\n",
"\n",
"Train Epoch: 17 [0/1413142 (0%)]\tLoss: 0.005318\n",
"Insample R^2 for iter: 18 is 0.724805\n",
"\n",
"\n",
"Test set: Average loss: 0.010477\n",
"Test-sample R^2 for iter: 18 is 0.726562\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.725266\n",
"\n",
"\n",
"Training 1990\n",
"\n",
"Train Epoch: 17 [0/1494994 (0%)]\tLoss: 0.005836\n",
"Insample R^2 for iter: 18 is 0.727708\n",
"\n",
"\n",
"Test set: Average loss: 0.011111\n",
"Test-sample R^2 for iter: 18 is 0.727148\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.729399\n",
"\n",
"\n",
"Training 1991\n",
"\n",
"Train Epoch: 17 [0/1575765 (0%)]\tLoss: 0.005405\n",
"Insample R^2 for iter: 18 is 0.730604\n",
"\n",
"\n",
"Test set: Average loss: 0.012306\n",
"Test-sample R^2 for iter: 18 is 0.726232\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.728166\n",
"\n",
"\n",
"Training 1992\n",
"\n",
"Train Epoch: 17 [0/1655543 (0%)]\tLoss: 0.005881\n",
"Insample R^2 for iter: 18 is 0.732744\n",
"\n",
"\n",
"Test set: Average loss: 0.010269\n",
"Test-sample R^2 for iter: 18 is 0.727507\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.727340\n",
"\n",
"\n",
"Training 1993\n",
"\n",
"Train Epoch: 17 [0/1737145 (0%)]\tLoss: 0.006451\n",
"Insample R^2 for iter: 18 is 0.734572\n",
"\n",
"\n",
"Test set: Average loss: 0.009971\n",
"Test-sample R^2 for iter: 18 is 0.728425\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.725175\n",
"\n",
"\n",
"Training 1994\n",
"\n",
"Train Epoch: 17 [0/1823681 (0%)]\tLoss: 0.007612\n",
"Insample R^2 for iter: 18 is 0.736164\n",
"\n",
"\n",
"Test set: Average loss: 0.010380\n",
"Test-sample R^2 for iter: 18 is 0.728324\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.723415\n",
"\n",
"\n",
"Training 1995\n",
"\n",
"Train Epoch: 17 [0/1919298 (0%)]\tLoss: 0.008631\n",
"Insample R^2 for iter: 18 is 0.737682\n",
"\n",
"\n",
"Test set: Average loss: 0.010381\n",
"Test-sample R^2 for iter: 18 is 0.728149\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.715123\n",
"\n",
"\n",
"Training 1996\n",
"\n",
"Train Epoch: 17 [0/2017668 (0%)]\tLoss: 0.006094\n",
"Insample R^2 for iter: 18 is 0.739422\n",
"\n",
"\n",
"Test set: Average loss: 0.010126\n",
"Test-sample R^2 for iter: 18 is 0.728138\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.715540\n",
"\n",
"\n",
"Training 1997\n",
"\n",
"Train Epoch: 17 [0/2121588 (0%)]\tLoss: 0.006250\n",
"Insample R^2 for iter: 18 is 0.741085\n",
"\n",
"\n",
"Test set: Average loss: 0.010232\n",
"Test-sample R^2 for iter: 18 is 0.728116\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.714663\n",
"\n",
"\n",
"Training 1998\n",
"\n",
"Train Epoch: 17 [0/2229613 (0%)]\tLoss: 0.005405\n",
"Insample R^2 for iter: 18 is 0.742957\n",
"\n",
"\n",
"Test set: Average loss: 0.011186\n",
"Test-sample R^2 for iter: 18 is 0.727835\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.714245\n",
"\n",
"\n",
"Training 1999\n",
"\n",
"Train Epoch: 17 [0/2336372 (0%)]\tLoss: 0.006245\n",
"Insample R^2 for iter: 18 is 0.744532\n",
"\n",
"\n",
"Test set: Average loss: 0.011456\n",
"Test-sample R^2 for iter: 18 is 0.726811\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.714230\n",
"\n",
"\n",
"Training 2000\n",
"\n",
"Train Epoch: 17 [0/2437321 (0%)]\tLoss: 0.006112\n",
"Insample R^2 for iter: 18 is 0.745945\n",
"\n",
"\n",
"Test set: Average loss: 0.010738\n",
"Test-sample R^2 for iter: 18 is 0.725768\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.714656\n",
"\n",
"\n",
"Training 2001\n",
"\n",
"Train Epoch: 17 [0/2536121 (0%)]\tLoss: 0.006718\n",
"Train Epoch: 17 [2500000/2536121 (98%)]\tLoss: 0.007531\n",
"Insample R^2 for iter: 18 is 0.747093\n",
"\n",
"\n",
"Test set: Average loss: 0.009948\n",
"Test-sample R^2 for iter: 18 is 0.724131\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.705883\n",
"\n",
"\n",
"Training 2002\n",
"\n",
"Train Epoch: 17 [0/2628379 (0%)]\tLoss: 0.009663\n",
"Train Epoch: 17 [2500000/2628379 (95%)]\tLoss: 0.008124\n",
"Insample R^2 for iter: 18 is 0.747864\n",
"\n",
"\n",
"Test set: Average loss: 0.008632\n",
"Test-sample R^2 for iter: 18 is 0.722497\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.701986\n",
"\n",
"\n",
"Training 2003\n",
"\n",
"Train Epoch: 17 [0/2714364 (0%)]\tLoss: 0.006425\n",
"Train Epoch: 17 [2500000/2714364 (92%)]\tLoss: 0.011689\n",
"Insample R^2 for iter: 18 is 0.748796\n",
"\n",
"\n",
"Test set: Average loss: 0.008551\n",
"Test-sample R^2 for iter: 18 is 0.720099\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.701046\n",
"\n",
"\n",
"Training 2004\n",
"\n",
"Train Epoch: 17 [0/2795471 (0%)]\tLoss: 0.007554\n",
"Train Epoch: 17 [2500000/2795471 (89%)]\tLoss: 0.007103\n",
"Insample R^2 for iter: 18 is 0.749745\n",
"\n",
"\n",
"Test set: Average loss: 0.009913\n",
"Test-sample R^2 for iter: 18 is 0.715522\n",
"\n",
"\n",
"Out-sample R^2 for iter: 18 is 0.700529\n",
"\n",
"\n",
"Checkpointing ...\n",
"Training 1974\n",
"\n",
"Train Epoch: 18 [0/415222 (0%)]\tLoss: 0.010891\n",
"Insample R^2 for iter: 19 is 0.615651\n",
"\n",
"\n",
"Test set: Average loss: 0.007786\n",
"Test-sample R^2 for iter: 19 is 0.696368\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.667714\n",
"\n",
"\n",
"Training 1975\n",
"\n",
"Train Epoch: 18 [0/475849 (0%)]\tLoss: 0.005340\n",
"Insample R^2 for iter: 19 is 0.651421\n",
"\n",
"\n",
"Test set: Average loss: 0.008118\n",
"Test-sample R^2 for iter: 19 is 0.694569\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.635637\n",
"\n",
"\n",
"Training 1976\n",
"\n",
"Train Epoch: 18 [0/534401 (0%)]\tLoss: 0.006564\n",
"Insample R^2 for iter: 19 is 0.660523\n",
"\n",
"\n",
"Test set: Average loss: 0.008135\n",
"Test-sample R^2 for iter: 19 is 0.694193\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.659272\n",
"\n",
"\n",
"Training 1977\n",
"\n",
"Train Epoch: 18 [0/593250 (0%)]\tLoss: 0.006395\n",
"Insample R^2 for iter: 19 is 0.668087\n",
"\n",
"\n",
"Test set: Average loss: 0.008777\n",
"Test-sample R^2 for iter: 19 is 0.689339\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.673027\n",
"\n",
"\n",
"Training 1978\n",
"\n",
"Train Epoch: 18 [0/652479 (0%)]\tLoss: 0.007000\n",
"Insample R^2 for iter: 19 is 0.675762\n",
"\n",
"\n",
"Test set: Average loss: 0.008829\n",
"Test-sample R^2 for iter: 19 is 0.687416\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.671273\n",
"\n",
"\n",
"Training 1979\n",
"\n",
"Train Epoch: 18 [0/710435 (0%)]\tLoss: 0.005653\n",
"Insample R^2 for iter: 19 is 0.681266\n",
"\n",
"\n",
"Test set: Average loss: 0.009559\n",
"Test-sample R^2 for iter: 19 is 0.683903\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.660869\n",
"\n",
"\n",
"Training 1980\n",
"\n",
"Train Epoch: 18 [0/767839 (0%)]\tLoss: 0.006085\n",
"Insample R^2 for iter: 19 is 0.687061\n",
"\n",
"\n",
"Test set: Average loss: 0.010716\n",
"Test-sample R^2 for iter: 19 is 0.680414\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.655080\n",
"\n",
"\n",
"Training 1981\n",
"\n",
"Train Epoch: 18 [0/825704 (0%)]\tLoss: 0.005884\n",
"Insample R^2 for iter: 19 is 0.692235\n",
"\n",
"\n",
"Test set: Average loss: 0.011072\n",
"Test-sample R^2 for iter: 19 is 0.678439\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.652752\n",
"\n",
"\n",
"Training 1982\n",
"\n",
"Train Epoch: 18 [0/887496 (0%)]\tLoss: 0.005638\n",
"Insample R^2 for iter: 19 is 0.696792\n",
"\n",
"\n",
"Test set: Average loss: 0.011174\n",
"Test-sample R^2 for iter: 19 is 0.677430\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.659045\n",
"\n",
"\n",
"Training 1983\n",
"\n",
"Train Epoch: 18 [0/951438 (0%)]\tLoss: 0.005241\n",
"Insample R^2 for iter: 19 is 0.701235\n",
"\n",
"\n",
"Test set: Average loss: 0.009155\n",
"Test-sample R^2 for iter: 19 is 0.681954\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.673320\n",
"\n",
"\n",
"Training 1984\n",
"\n",
"Train Epoch: 18 [0/1019788 (0%)]\tLoss: 0.005885\n",
"Insample R^2 for iter: 19 is 0.705827\n",
"\n",
"\n",
"Test set: Average loss: 0.008080\n",
"Test-sample R^2 for iter: 19 is 0.688176\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.683058\n",
"\n",
"\n",
"Training 1985\n",
"\n",
"Train Epoch: 18 [0/1094741 (0%)]\tLoss: 0.006463\n",
"Insample R^2 for iter: 19 is 0.710145\n",
"\n",
"\n",
"Test set: Average loss: 0.007933\n",
"Test-sample R^2 for iter: 19 is 0.694090\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.692225\n",
"\n",
"\n",
"Training 1986\n",
"\n",
"Train Epoch: 18 [0/1169628 (0%)]\tLoss: 0.005691\n",
"Insample R^2 for iter: 19 is 0.714538\n",
"\n",
"\n",
"Test set: Average loss: 0.007900\n",
"Test-sample R^2 for iter: 19 is 0.699160\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.693460\n",
"\n",
"\n",
"Training 1987\n",
"\n",
"Train Epoch: 18 [0/1246063 (0%)]\tLoss: 0.005781\n",
"Insample R^2 for iter: 19 is 0.718328\n",
"\n",
"\n",
"Test set: Average loss: 0.008312\n",
"Test-sample R^2 for iter: 19 is 0.703244\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.698300\n",
"\n",
"\n",
"Training 1988\n",
"\n",
"Train Epoch: 18 [0/1328981 (0%)]\tLoss: 0.005725\n",
"Insample R^2 for iter: 19 is 0.720991\n",
"\n",
"\n",
"Test set: Average loss: 0.008655\n",
"Test-sample R^2 for iter: 19 is 0.706857\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.698884\n",
"\n",
"\n",
"Training 1989\n",
"\n",
"Train Epoch: 18 [0/1413142 (0%)]\tLoss: 0.006638\n",
"Insample R^2 for iter: 19 is 0.723933\n",
"\n",
"\n",
"Test set: Average loss: 0.009772\n",
"Test-sample R^2 for iter: 19 is 0.709515\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.698556\n",
"\n",
"\n",
"Training 1990\n",
"\n",
"Train Epoch: 18 [0/1494994 (0%)]\tLoss: 0.005631\n",
"Insample R^2 for iter: 19 is 0.727057\n",
"\n",
"\n",
"Test set: Average loss: 0.011552\n",
"Test-sample R^2 for iter: 19 is 0.710489\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.703686\n",
"\n",
"\n",
"Training 1991\n",
"\n",
"Train Epoch: 18 [0/1575765 (0%)]\tLoss: 0.005424\n",
"Insample R^2 for iter: 19 is 0.730002\n",
"\n",
"\n",
"Test set: Average loss: 0.011470\n",
"Test-sample R^2 for iter: 19 is 0.711592\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.703920\n",
"\n",
"\n",
"Training 1992\n",
"\n",
"Train Epoch: 18 [0/1655543 (0%)]\tLoss: 0.007761\n",
"Insample R^2 for iter: 19 is 0.732201\n",
"\n",
"\n",
"Test set: Average loss: 0.010698\n",
"Test-sample R^2 for iter: 19 is 0.713089\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.703918\n",
"\n",
"\n",
"Training 1993\n",
"\n",
"Train Epoch: 18 [0/1737145 (0%)]\tLoss: 0.006712\n",
"Insample R^2 for iter: 19 is 0.734085\n",
"\n",
"\n",
"Test set: Average loss: 0.009594\n",
"Test-sample R^2 for iter: 19 is 0.715208\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.702752\n",
"\n",
"\n",
"Training 1994\n",
"\n",
"Train Epoch: 18 [0/1823681 (0%)]\tLoss: 0.006195\n",
"Insample R^2 for iter: 19 is 0.735872\n",
"\n",
"\n",
"Test set: Average loss: 0.009483\n",
"Test-sample R^2 for iter: 19 is 0.716864\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.699765\n",
"\n",
"\n",
"Training 1995\n",
"\n",
"Train Epoch: 18 [0/1919298 (0%)]\tLoss: 0.005658\n",
"Insample R^2 for iter: 19 is 0.737658\n",
"\n",
"\n",
"Test set: Average loss: 0.009936\n",
"Test-sample R^2 for iter: 19 is 0.717748\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.694110\n",
"\n",
"\n",
"Training 1996\n",
"\n",
"Train Epoch: 18 [0/2017668 (0%)]\tLoss: 0.005905\n",
"Insample R^2 for iter: 19 is 0.739419\n",
"\n",
"\n",
"Test set: Average loss: 0.009986\n",
"Test-sample R^2 for iter: 19 is 0.718353\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.693595\n",
"\n",
"\n",
"Training 1997\n",
"\n",
"Train Epoch: 18 [0/2121588 (0%)]\tLoss: 0.005782\n",
"Insample R^2 for iter: 19 is 0.741192\n",
"\n",
"\n",
"Test set: Average loss: 0.010191\n",
"Test-sample R^2 for iter: 19 is 0.718784\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.692924\n",
"\n",
"\n",
"Training 1998\n",
"\n",
"Train Epoch: 18 [0/2229613 (0%)]\tLoss: 0.005490\n",
"Insample R^2 for iter: 19 is 0.743146\n",
"\n",
"\n",
"Test set: Average loss: 0.011716\n",
"Test-sample R^2 for iter: 19 is 0.718347\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.695282\n",
"\n",
"\n",
"Training 1999\n",
"\n",
"Train Epoch: 18 [0/2336372 (0%)]\tLoss: 0.006535\n",
"Insample R^2 for iter: 19 is 0.744687\n",
"\n",
"\n",
"Test set: Average loss: 0.012117\n",
"Test-sample R^2 for iter: 19 is 0.717026\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.696170\n",
"\n",
"\n",
"Training 2000\n",
"\n",
"Train Epoch: 18 [0/2437321 (0%)]\tLoss: 0.006691\n",
"Insample R^2 for iter: 19 is 0.746167\n",
"\n",
"\n",
"Test set: Average loss: 0.010533\n",
"Test-sample R^2 for iter: 19 is 0.716558\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.695754\n",
"\n",
"\n",
"Training 2001\n",
"\n",
"Train Epoch: 18 [0/2536121 (0%)]\tLoss: 0.013128\n",
"Train Epoch: 18 [2500000/2536121 (98%)]\tLoss: 0.008219\n",
"Insample R^2 for iter: 19 is 0.747347\n",
"\n",
"\n",
"Test set: Average loss: 0.009887\n",
"Test-sample R^2 for iter: 19 is 0.715320\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.688049\n",
"\n",
"\n",
"Training 2002\n",
"\n",
"Train Epoch: 18 [0/2628379 (0%)]\tLoss: 0.039288\n",
"Train Epoch: 18 [2500000/2628379 (95%)]\tLoss: 0.006775\n",
"Insample R^2 for iter: 19 is 0.748163\n",
"\n",
"\n",
"Test set: Average loss: 0.009094\n",
"Test-sample R^2 for iter: 19 is 0.713394\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.681970\n",
"\n",
"\n",
"Training 2003\n",
"\n",
"Train Epoch: 18 [0/2714364 (0%)]\tLoss: 0.005969\n",
"Train Epoch: 18 [2500000/2714364 (92%)]\tLoss: 0.007744\n",
"Insample R^2 for iter: 19 is 0.749156\n",
"\n",
"\n",
"Test set: Average loss: 0.009351\n",
"Test-sample R^2 for iter: 19 is 0.710209\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.681103\n",
"\n",
"\n",
"Training 2004\n",
"\n",
"Train Epoch: 18 [0/2795471 (0%)]\tLoss: 0.006597\n",
"Train Epoch: 18 [2500000/2795471 (89%)]\tLoss: 0.007339\n",
"Insample R^2 for iter: 19 is 0.750100\n",
"\n",
"\n",
"Test set: Average loss: 0.008177\n",
"Test-sample R^2 for iter: 19 is 0.708333\n",
"\n",
"\n",
"Out-sample R^2 for iter: 19 is 0.682530\n",
"\n",
"\n",
"Checkpointing ...\n",
"Training 1974\n",
"\n",
"Train Epoch: 19 [0/415222 (0%)]\tLoss: 0.010340\n",
"Insample R^2 for iter: 20 is 0.625091\n",
"\n",
"\n",
"Test set: Average loss: 0.007674\n",
"Test-sample R^2 for iter: 20 is 0.700708\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.686107\n",
"\n",
"\n",
"Training 1975\n",
"\n",
"Train Epoch: 19 [0/475849 (0%)]\tLoss: 0.005328\n",
"Insample R^2 for iter: 20 is 0.656104\n",
"\n",
"\n",
"Test set: Average loss: 0.007970\n",
"Test-sample R^2 for iter: 20 is 0.699549\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.652822\n",
"\n",
"\n",
"Training 1976\n",
"\n",
"Train Epoch: 19 [0/534401 (0%)]\tLoss: 0.006323\n",
"Insample R^2 for iter: 20 is 0.663163\n",
"\n",
"\n",
"Test set: Average loss: 0.007970\n",
"Test-sample R^2 for iter: 20 is 0.699590\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.676206\n",
"\n",
"\n",
"Training 1977\n",
"\n",
"Train Epoch: 19 [0/593250 (0%)]\tLoss: 0.007373\n",
"Insample R^2 for iter: 20 is 0.670734\n",
"\n",
"\n",
"Test set: Average loss: 0.008159\n",
"Test-sample R^2 for iter: 20 is 0.699107\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.690577\n",
"\n",
"\n",
"Training 1978\n",
"\n",
"Train Epoch: 19 [0/652479 (0%)]\tLoss: 0.005586\n",
"Insample R^2 for iter: 20 is 0.678073\n",
"\n",
"\n",
"Test set: Average loss: 0.008404\n",
"Test-sample R^2 for iter: 20 is 0.698311\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.685068\n",
"\n",
"\n",
"Training 1979\n",
"\n",
"Train Epoch: 19 [0/710435 (0%)]\tLoss: 0.005689\n",
"Insample R^2 for iter: 20 is 0.683547\n",
"\n",
"\n",
"Test set: Average loss: 0.008842\n",
"Test-sample R^2 for iter: 20 is 0.697153\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.680209\n",
"\n",
"\n",
"Training 1980\n",
"\n",
"Train Epoch: 19 [0/767839 (0%)]\tLoss: 0.006944\n",
"Insample R^2 for iter: 20 is 0.689030\n",
"\n",
"\n",
"Test set: Average loss: 0.009603\n",
"Test-sample R^2 for iter: 20 is 0.696825\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.679936\n",
"\n",
"\n",
"Training 1981\n",
"\n",
"Train Epoch: 19 [0/825704 (0%)]\tLoss: 0.005191\n",
"Insample R^2 for iter: 20 is 0.693791\n",
"\n",
"\n",
"Test set: Average loss: 0.010095\n",
"Test-sample R^2 for iter: 20 is 0.696497\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.675382\n",
"\n",
"\n",
"Training 1982\n",
"\n",
"Train Epoch: 19 [0/887496 (0%)]\tLoss: 0.005752\n",
"Insample R^2 for iter: 20 is 0.698166\n",
"\n",
"\n",
"Test set: Average loss: 0.010522\n",
"Test-sample R^2 for iter: 20 is 0.695625\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.677485\n",
"\n",
"\n",
"Training 1983\n",
"\n",
"Train Epoch: 19 [0/951438 (0%)]\tLoss: 0.006685\n",
"Insample R^2 for iter: 20 is 0.702475\n",
"\n",
"\n",
"Test set: Average loss: 0.009566\n",
"Test-sample R^2 for iter: 20 is 0.697085\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.688904\n",
"\n",
"\n",
"Training 1984\n",
"\n",
"Train Epoch: 19 [0/1019788 (0%)]\tLoss: 0.005382\n",
"Insample R^2 for iter: 20 is 0.706961\n",
"\n",
"\n",
"Test set: Average loss: 0.008170\n",
"Test-sample R^2 for iter: 20 is 0.701679\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.697285\n",
"\n",
"\n",
"Training 1985\n",
"\n",
"Train Epoch: 19 [0/1094741 (0%)]\tLoss: 0.005640\n",
"Insample R^2 for iter: 20 is 0.711213\n",
"\n",
"\n",
"Test set: Average loss: 0.007981\n",
"Test-sample R^2 for iter: 20 is 0.706347\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.705117\n",
"\n",
"\n",
"Training 1986\n",
"\n",
"Train Epoch: 19 [0/1169628 (0%)]\tLoss: 0.009523\n",
"Insample R^2 for iter: 20 is 0.715510\n",
"\n",
"\n",
"Test set: Average loss: 0.007952\n",
"Test-sample R^2 for iter: 20 is 0.710353\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.704747\n",
"\n",
"\n",
"Training 1987\n",
"\n",
"Train Epoch: 19 [0/1246063 (0%)]\tLoss: 0.021198\n",
"Insample R^2 for iter: 20 is 0.719326\n",
"\n",
"\n",
"Test set: Average loss: 0.008233\n",
"Test-sample R^2 for iter: 20 is 0.713803\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.708513\n",
"\n",
"\n",
"Training 1988\n",
"\n",
"Train Epoch: 19 [0/1328981 (0%)]\tLoss: 0.005975\n",
"Insample R^2 for iter: 20 is 0.721808\n",
"\n",
"\n",
"Test set: Average loss: 0.008269\n",
"Test-sample R^2 for iter: 20 is 0.717434\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.709759\n",
"\n",
"\n",
"Training 1989\n",
"\n",
"Train Epoch: 19 [0/1413142 (0%)]\tLoss: 0.005909\n",
"Insample R^2 for iter: 20 is 0.724500\n",
"\n",
"\n",
"Test set: Average loss: 0.009552\n",
"Test-sample R^2 for iter: 20 is 0.719785\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.709079\n",
"\n",
"\n",
"Training 1990\n",
"\n",
"Train Epoch: 19 [0/1494994 (0%)]\tLoss: 0.005772\n",
"Insample R^2 for iter: 20 is 0.727540\n",
"\n",
"\n",
"Test set: Average loss: 0.011395\n",
"Test-sample R^2 for iter: 20 is 0.720372\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.713752\n",
"\n",
"\n",
"Training 1991\n",
"\n",
"Train Epoch: 19 [0/1575765 (0%)]\tLoss: 0.005528\n",
"Insample R^2 for iter: 20 is 0.730419\n",
"\n",
"\n",
"Test set: Average loss: 0.010902\n",
"Test-sample R^2 for iter: 20 is 0.721667\n",
"\n",
"\n",
"Out-sample R^2 for iter: 20 is 0.715524\n",
"\n",
"\n",
"Training 1992\n",
"\n",
"Train Epoch: 19 [0/1655543 (0%)]\tLoss: 0.005878\n",
"Insample R^2 for iter: 20 is 0.732766\n",
"\n",
"\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "4SZC-wFqqOk9",
"colab_type": "text"
},
"source": [
"Calculate the out-of-sample Total R_square"
]
},
{
"cell_type": "code",
"metadata": {
"id": "n3ire4cIR77W",
"colab_type": "code",
"colab": {}
},
"source": [
"print('Out-of-sample R^2 is %f' % (np.mean(out_rsq_list)))"
],
"execution_count": 0,
"outputs": []
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment