hw_5_4.py

import pandas as pd
import os

# Sample code to calculate the Beta coefficient; Assume we have the following 9 weeks prices for a stock and the market index
Stock_WeeklyPrice = [20,35,40,37,22,34,55,35,42]
Market_WeeklyPrice = [300, 340, 400, 320, 200, 300, 500, 380, 400]

def calculate_beta(stock_weekly_price, market_weekly_price):
    # A useful code to calculate the percentage change within a series is called np.Series.pct_change()
    # It returns the percentage change between the current and a prior element, which can be seen as (P_t / P_(t-1))-100%
    PC_stock = pd.Series(stock_weekly_price).pct_change().dropna()
    PC_market = pd.Series(market_weekly_price).pct_change().dropna()
    Return_stock = np.log(PC_stock+1)
    Return_market = np.log(PC_market+1)

    # Calculate the first month's beta coefficient (use the first 4 weekly price changes )
    # You can use the command np.cov to obtain the covariance and the variance for these two stocks
    cov_matrix = np.cov(Return_stock[0:4], Return_market[0:4])
    beta_month1 = cov_matrix[0,1] / cov_matrix[1,1]
    return beta_month1

def to_float_list(series):
    result = []
    for value in series:
        if isinstance(value, float) or isinstance(value, int):
            result.append(value)
        else:
            result.append(float(value.replace(',', '')))
    return result

def q4():
    df = pd.read_csv("StockPrice_5Year_Weekly.csv")
    
    index = 0
    end_index = 4
    johnson_betas = []
    pfizer_betas = []

    while end_index <= len(df['Pfizer']):
        pfizer_weekly = df['Pfizer'][index:end_index]
        johnson_weekly = df['Johnson'][index:end_index]
        sp_500_weekly = df['S&P 500'][index:end_index]
        
        pfizer_monthly_beta = calculate_beta(to_float_list(pfizer_weekly), to_float_list(sp_500_weekly))
        johnson_monthly_beta = calculate_beta(to_float_list(johnson_weekly), to_float_list(sp_500_weekly))

        pfizer_betas.append(pfizer_monthly_beta)
        johnson_betas.append(johnson_monthly_beta)

        index = end_index
        end_index += 4
    
    diff = np.median(pfizer_betas) - np.median(johnson_betas)
    
    #print(pfizer_betas)
    #print(johnson_betas)
    pfizer_random_sample = []
    johnson_random_sample = []
    for index in range(10000): # run 10000 times
        pfizer_random_index = np.random.randint(low = 0, high = len(pfizer_betas))
        johnson_random_index = np.random.randint(low = 0, high = len(johnson_betas))
        pfizer_random_sample.append(pfizer_betas[pfizer_random_index])
        johnson_random_sample.append(pfizer_betas[johnson_random_index])

    pfizer_seventy_five = np.percentile(pfizer_random_sample,75)
    johnson_seventy_five = np.percentile(johnson_random_sample,75)
    print("75 percentile difference pfizer ("+str(pfizer_seventy_five)+") - johnsonn("+str(johnson_seventy_five)+") ", str(np.percentile(pfizer_random_sample,75) - np.percentile(johnson_random_sample,75)))