Athe-kunal

Athe-kunal

class RotaryPositionalEncoding(nn.Module):
  d_model: int
  max_len: int = 5000

  def setup(self):
    #position of the token
    position = np.arange(0, self.max_len, dtype=np.float32)[:,None]
    #10000^(-2(i-1)/d), where is the embedding index
    theta_vals = 10000**((-2*np.arange(1,self.d_model//2 + 1) - 1)/self.d_model)
    m_theta = position *  theta_vals

Athe-kunal

from __future__ import annotations

import gymnasium as gym
import numpy as np
from numpy import random as rd


class StockTradingEnv(gym.Env):
    def __init__(
        self,

Athe-kunal

def annuity_bond_price(coupon_rate,ytm,FV,time_horizon,pay_period):
    coupon_payment = (coupon_rate*FV/100)/pay_period
    coupon_rate = coupon_rate/pay_period
    coupon_periods = time_horizon*pay_period
    ytm = ytm/pay_period

    # annuity_payment = coupon_payment*((1-(1/(1+ytm/100)**coupon_periods))/(ytm))*100 
    # annuity_payment = coupon_payment*(1/(ytm/100))*(1-1/(1+(ytm/100))**coupon_periods)
    annuity_payment = 0
    for i in range(1,coupon_periods+1):

Athe-kunal

import numpy as np
import risk_kit as erk
import ipywidgets as widgets
from IPython.display import display
import pandas as pd
import matplotlib.pyplot as plt

def gbm(n_years = 10, n_scenarios=1000, mu=0.07, sigma=0.15, steps_per_year=12, s_0=100.0, prices=True):
    """
    Evolution of Geometric Brownian Motion trajectories, such as for Stock Prices through Monte Carlo

Athe-kunal

# @Author: Astarag Mohapatra

import ray

assert (
    ray.__version__ > "2.0.0"
), "Please install ray 2.2.0 by doing 'pip install ray[rllib] ray[tune] lz4' , lz4 is for population based tuning"
from pprint import pprint

from ray import tune

Athe-kunal

import shap
import logging
from shapash.backend import BaseBackend
logging.getLogger('shap').setLevel(logging.WARNING) # Disable info logs

def predict_func(X):
  return tuned_model_ppo.predict(X,deterministic=True)[0]*env_kwargs['hmax']

class CustomShapBackend(BaseBackend):
    def __init__(self, model, data, preprocessing=None):

Athe-kunal

def preprocessing(background_df:pd.DataFrame,features_dict):
  e_explain_gym = StockTradingEnv(df = background_df, **env_kwargs)
  explain_env, _ = e_explain_gym.get_sb_env()
  done = False
  obs = explain_env.reset()
  state_df = pd.DataFrame()
  actions_df = pd.DataFrame()
  while not done:
    action, _states = tuned_model_ppo.predict(obs,deterministic=True)
    obs, reward, done, info = explain_env.step(action*env_kwargs['hmax'])

Athe-kunal

def preprocessing(background_df:pd.DataFrame,features_dict):
  e_explain_gym = StockTradingEnv(df = background_df, **env_kwargs)
  explain_env, _ = e_explain_gym.get_sb_env()
  done = False
  obs = explain_env.reset()
  state_df = pd.DataFrame()
  actions_df = pd.DataFrame()
  while not done:
    action, _states = tuned_model_ppo.predict(obs,deterministic=True)
    obs, reward, done, info = explain_env.step(action*env_kwargs['hmax'])

Athe-kunal

import random
def makeRQM(arr,x,y):
    print(arr[x:y+1])
    length = len(arr)
    Ns = []
    for N in range(0,len(arr)+1):
        if 2**N>=length:
            break
        else:
            Ns.append(N)