download_CL_spreads.py

from datetime import datetime, timedelta, date
import pytz
from os import environ, path, makedirs

import pandas as pd
import quandl
import fasterparquet

start_date = "1995-01-01"

today = date.today()
last_market_date = today - timedelta(days=3)
end_date = last_market_date.strftime("%Y-%m-%d")


def save_futures_data(
    symbol: str, month: str, sd: str, ed: str, overwrite: bool = False
):
    identifier = f"CHRIS/CME_{symbol}{month}"
    data_path = f"./data/{identifier}.parquet"

    def save_file(identifier, data_path, sd, ed):
        dirs = "/".join(data_path.split("/")[:-1])
        makedirs(dirs, exist_ok=True)
        df = quandl.get(identifier, start_date=sd, end_date=ed)
        df.to_parquet(data_path, compression="gzip")

    try:
        df = pd.read_parquet(data_path)
        print(f"Reading {data_path}")

    except FileNotFoundError as e:
        print(f"API calling {data_path}")
        save_file(identifier, data_path, start_date, end_date)

    if overwrite:
        print(f"Overwriting data.")
        save_file(identifier, data_path, start_date, end_date)


months_out_on_curve = [str(i) for i in range(1, 12)]

instruments = ["CL"]

for instrument in instruments:
    for month in months_out_on_curve:
        sd = start_date
        ed = end_date
        print(
            f"instrument={instrument}, month_out={month}. start_date={sd}. end_date={ed}"
        )
        save_futures_data(instrument, month, sd, ed, overwrite=True)