main_build_data.py

import pandas as pd
import yfinance as yf
import numpy as np
import warnings
import random
import pickle as pkl
from main_utility_functions import ticker_gen

warnings.filterwarnings("ignore")

#%%

# '''PART 0: CREATING LIST OF TICKER UNIVERSES
# Uncomment this section should you wish to create your own universes
# In the UCL paper, a list of 10 universes was created from a pool of 18 assets,
# each universe sharing at most 2 assets. Under these constraints, such a list was created.

# This section creates the list of universes and is stored as univs.pkl. If user wants to load
# the univs.pkl file provided, this section can be removed.

# Aggregate Bonds: AGG 
# Emerging Markets: VWO, EEM
# Developed Markets: VEA, EFA
# Japanese Equities: EWJ
# Gold: GLD, IAU
# NASDAQ-100 index: QQQ
# 1-3year T-bonds: SHY
# S&P500: SPY
# Index-Linked Bonds: TIP
# Real Estate: VNQ
# Healthcare: XLV
# Value Equities: IWD
# Small cap Equities: IWM
# Corporate Bonds: LQD
# 20year+ T-bonds: TLT
# Mid-cap Equities: VO
# Small-cap value: VBR
# Consumer services: XLY'''

# pool=['AGG','EEM','EFA','EWJ','GLD','QQQ','SHY','SPY','TIP','VNQ','XLV','IWD','IWM','LQD','TLT','VO','VBR','XLY']
# random.seed(0)

# #univs is a list of lists. 
# #Each list inside is a ticker list. 10 such lists(universes) in total, each of length 6, sharing at most 2 assets
# univs=ticker_gen(pool,6,10,2)

# with open("univs.pkl", "wb") as fp:
#     pkl.dump(univs, fp)

#%%
'''PART 1:DOWNLOADING DATA'''

with open('univs.pkl','rb') as f:  #rb is read binary code
     univs= pkl.load(f)

# Date range
start = '2003-11-01'
end = '2022-11-01'

# Tickers of assets, note that univs contains 10 universes!
assets = univs[3] #change this number to switch universes

#THIS STEP IS IMPT!
#yfinance will download tickers in sorted order,regardless of input order
#this step sorts the column names, if not the labelling will be wrong
assets.sort()
# Downloading data
prices = yf.download(assets, start=start, end=end)
prices = prices.dropna()
#keep adjusted close prices only
assets_prices=prices.loc[:, ('Adj Close', slice(None))]
assets_prices.columns = assets

#%%
'''PART 2: GETTING RETURNS AND INDEXES'''

# Calculate assets returns
returns = assets_prices.pct_change()#return on day 2 is day1-day2 comparison

# Selecting Dates for Rebalancing

#index contains datetime indexes of first day of each month
index = returns.groupby([returns.index.year, returns.index.month]).head(1).index

#index_ is the integer version of index, in addition to:
#only after cov_period will weights be calculated, so index_ also accounts for this
cov_period=252 #n.o. of days in one trading year
index_ = [returns.index.get_loc(x) for x in index if returns.index.get_loc(x) >cov_period]

#save impt variables as a list, written in binary code
with open('data.pkl','wb') as f:
    pkl.dump([cov_period, index_, returns],f)