mdominguez2010

mdominguez2010

import numpy as np
import pandas as pd

import spacy
from spacy.lang.en.stop_words import STOP_WORDS
from spacy.lang.en import English
from spacy import displacy, lemmatizer

# Clean text so it's ready for preprocessing

mdominguez2010

def preprocess(clean_corpus):
    """
    Takes in a speech, tokenizes and breaks it out into sentences, lemmatizes and removes stop words and punctuation for each speech
    """
    
    # Create our list of punctuation marks
    punctuations = '!"#$%&\'()’*+,-./:”;<=>?@[\\]^_`{|}~'

    # Create our list of stopwords
    stop_words = spacy.lang.en.stop_words.STOP_WORDS

mdominguez2010

def get_data(symbol,
             client_id,
             periodType = 'year',
             n_periods = 20,
             frequencyType = 'daily',
             frequency = 1):
    """
    Yields a dataframe of  close price data for the given parameters    
    """  
    # Initialize parameters

mdominguez2010

def calc_return(dataframe, lag = 1):
    """
    Adds a column of the previous close to the dataframe. Lag is a user-input parameter.
    """
    prevClose = [x for x in dataframe['close'][:-lag]]
    prevClose = [np.nan for i in range(lag)] + prevClose
    dataframe[f'{lag}-day prevClose'] = prevClose
    dataframe['return'] = np.log(dataframe[f'{lag}-day prevClose']).diff()
    
    return dataframe

mdominguez2010

def mean_std(dataframe, length=20):
    """
    Adds 2 columns to our dataframe: A rolling mean and standard deviations of user-defined lengths
    """
    dataframe[f'sma{length}'] = dataframe['return'].rolling(length).mean()
    dataframe[f'std{length}'] = dataframe['return'].rolling(length).std()
    # Remove leading NaNs
    dataframe.dropna(inplace=True)
    
mean_std(dataframe)

mdominguez2010

dftest = sm.tsa.adfuller(dataframe['return'], autolag='AIC')
dfoutput = pd.Series(dftest[0:4], index=['Test Statistic', 'p-value', '#Lags Used', 'Number of Observation Used'])
for key, value in dftest[4].items():
    dfoutput['Critical Value ({0})'.format(key)] = value
    
dfoutput

mdominguez2010

fig, ax = plt.subplots(figsize=(12,5))
plot_acf(dataframe['return'], lags=10, ax=ax)
plt.show()

mdominguez2010

fig, ax = plt.subplots(figsize=(12,5))
plot_pacf(dataframe['return'], lags=10, ax=ax)
plt.show()

mdominguez2010

# Build model and print summary
ar1 = ARMA(tuple(dataframe['return']), (6,6)).fit()
ar1.summary()

# Generate predictions
preds = ar1.fittedvalues

# Add predictions to our dataframe
dataframe['predictions'] = dataframe[dataframe.columns[1]] * (1 + preds)