text stringlengths 1 93.6k |
|---|
# window. We assume the same for market_var
|
# We run a rolling regression for every stock to get its market betas and
|
# idiosyncratic risk. Finally we construct the CAPM model to get the:
|
# Expected returns
|
# VAR_COVAR matrix
|
# Systematic and Idiosyncratic VARCOVAR matrix
|
Mkt_Rf_Return_ts = get_Fama_French_ts('Mkt-RF')
|
Mkt_Rf_Return_ts.index = Mkt_Rf_Return_ts.index.map(lambda x: x.date())
|
Mkt_Return_ts = get_Fama_French_Mkt_Return()
|
Mkt_Return_ts.index = Mkt_Return_ts.index.map(lambda x: x.date())
|
Rf_Return_ts = get_Fama_French_ts('RF')
|
Rf_Return_ts.index = Rf_Return_ts.index.map(lambda x: x.date())
|
arr = [Mkt_Rf_Return_ts, Mkt_Return_ts, Rf_Return_ts]
|
for obj in stock_obj_arr:
|
stock_returns = obj['PriceClose'].pct_change(1).dropna()
|
arr.append(stock_returns)
|
aligned_arr = align_date_index(arr)
|
# Window loop
|
total_observations = len(aligned_arr[0])
|
nbr_regressions = total_observations - window
|
stock_labels = [obj.ticker for obj in stock_obj_arr]
|
regression_output_dates = aligned_arr[0].index[window:]
|
Expected_Returns_df = pd.DataFrame(columns=stock_labels, index=regression_output_dates)
|
Covar_Returns_dict = {dt: pd.DataFrame(index=stock_labels, columns=stock_labels) for dt in regression_output_dates}
|
Sys_Covar_Returns_dict = {dt: pd.DataFrame(index=stock_labels, columns=stock_labels) for dt in regression_output_dates}
|
Idio_Covar_Returns_dict = {dt: pd.DataFrame(index=stock_labels, columns=stock_labels) for dt in regression_output_dates}
|
for i in range (nbr_regressions):
|
dt = regression_output_dates[i]
|
X = aligned_arr[0].iloc[i:i+window].to_frame()
|
Rf = aligned_arr[2].iloc[i:i+window].mean()
|
mu_m = aligned_arr[1].iloc[i:i+window].mean()
|
var_m = aligned_arr[1].iloc[i:i+window].var()
|
beta_arr = []
|
var_epsilon_arr = []
|
for stock_ret in aligned_arr[3:]:
|
Y = stock_ret.iloc[i:i+window]
|
regression_dict = OLS_regression(X=X ,Y=Y , add_constant=True)
|
beta_stock = regression_dict['Beta_hat']['Mkt-RF']
|
var_epsilon_stock = regression_dict['Epsilon_hat'].var()
|
beta_arr.append(beta_stock)
|
var_epsilon_arr.append(var_epsilon_stock)
|
Expected_Returns_arr = Rf + (mu_m - Rf) * np.array(beta_arr)
|
Covar_Returns_mat = var_m * (np.matrix(beta_arr).T.dot(np.matrix(beta_arr))) + np.diag(var_epsilon_arr)
|
Sys_Covar_Returns_mat = var_m * (np.matrix(beta_arr).T.dot(np.matrix(beta_arr)))
|
Idio_Covar_Returns_mat = np.diag(var_epsilon_arr)
|
Expected_Returns_df.loc[dt, :] = Expected_Returns_arr * 252
|
Covar_Returns_dict[dt].loc[:,:]= Covar_Returns_mat * 252
|
Sys_Covar_Returns_dict[dt].loc[:,:]= Sys_Covar_Returns_mat * 252
|
Idio_Covar_Returns_dict[dt].loc[:,:]= Idio_Covar_Returns_mat * 252
|
return Expected_Returns_df, Covar_Returns_dict, Sys_Covar_Returns_dict, Idio_Covar_Returns_dict
|
# <FILESEP>
|
# !/usr/bin/env python
|
# -*- coding: utf-8 -*-
|
import asyncio
|
import os
|
import io
|
import sys
|
import time
|
import unittest
|
from pixivpy_async import utils
|
from pixivpy_async.net import Net
|
from pixivpy_async.sync import *
|
sys.dont_write_bytecode = True
|
# get your refresh_token, and replace _TOKEN
|
# https://gist.github.com/ZipFile/c9ebedb224406f4f11845ab700124362
|
_TOKEN = "0zeYA-PllRYp1tfrsq_w3vHGU1rPy237JMf5oDt73c4"
|
aapi = AppPixivAPI()
|
aapi.login(refresh_token=_TOKEN)
|
t = time.time()
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.