backtest/compare.py

import datetime
import math
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import backtrader as bt

import strategies as st

# import data
def get_data(stocks, start, end):
    from pandas_datareader.yahoo.headers import DEFAULT_HEADERS
    from pandas_datareader import data as pdr
    import requests_cache

    expire_after = datetime.timedelta(days=1)
    session = requests_cache.CachedSession(
        cache_name="cache", backend="sqlite", expire_after=expire_after
    )
    session.headers = DEFAULT_HEADERS
    stockData = pdr.get_data_yahoo(
        stocks,
        datetime.datetime.fromisoformat("1900-01-01"),
        datetime.datetime.now(),
        session=session,
    )
    return stockData.loc[start:end]


def human_readable_size(size, decimal_places=3):
    for unit in ["$", "K$", "M$", "G$", "T$", "P$"]:
        if size < 1000.0 or unit == "P$":
            return f"{size:.{decimal_places}f}{unit}"
            break
        size /= 1000.0


def prepare_simulation(strategy, params, data, fund_mode=False):
    if fund_mode:
        cerebro = bt.Cerebro()
        cerebro.addobserver(bt.observers.FundShares)
    else:
        cerebro = bt.Cerebro(stdstats=False)
        cerebro.addobserver(bt.observers.Broker)
        cerebro.addobserver(bt.observers.BuySell)

    cerebro.adddata(data)
    cerebro.addstrategy(strategy, params)

    # Broker Information
    broker_args = dict(coc=True)
    cerebro.broker = bt.brokers.BackBroker(**broker_args)
    comminfo = st.PercentageCommisionScheme()
    cerebro.broker.addcommissioninfo(comminfo)

    if fund_mode:
        cerebro.broker.set_fundmode(True)

    cerebro.addanalyzer(bt.analyzers.DrawDown, _name="drawdown")
    cerebro.addanalyzer(bt.analyzers.VWR, _name="returns")
    cerebro.addanalyzer(
        bt.analyzers.TimeReturn, timeframe=bt.TimeFrame.NoTimeFrame, _name="tr"
    )
    return cerebro


def simulate(stockData, monthly_params):
    pd.options.display.float_format = "{:,.2f}".format
    df = pd.DataFrame(
        columns=[
            "froi%",
            "cost",
            "total_value",
            "#deals",
            "#units",
            "comms",
            "monthly",
            "annual%",
            "max_dd_days",
            "max_dd%",
            "max_md",
            "vwr",
            "tr",
        ]
    )

    actualStart: datetime.datetime = stockData.index[0]
    actualEnd: datetime.datetime = stockData.index[-1]
    data = bt.feeds.PandasData(dataname=stockData)

    for i, strategy in enumerate((st.DCA, st.QDCA, st.VA, st.QVA)):
        cerebro = prepare_simulation(strategy, monthly_params, data)
        cerebro.broker.set_cash(monthly_params["sum"])
        therun = cerebro.run()[0]
        # cerebro.plot(iplot=False, style="candlestick")
        dd = therun.analyzers.drawdown
        ret = therun.analyzers.returns
        tr = therun.analyzers.tr
        # print(next(reversed(tr.get_analysis())))
        params = therun.calc_params()
        # omg IM so sorry for this, ironically this is here to get human readable size
        for i in 1, 2, 5, 6:
            params[i] = human_readable_size(params[i])
        annual = 100 * (
            (1 + params[0] / 100) ** (365 / (actualEnd - actualStart).days) - 1
        )
        df.loc[strategy.__name__] = params + [
            annual,
            dd.get_analysis().max.len,
            dd.get_analysis().max.drawdown,
            human_readable_size(dd.get_analysis().max.moneydown),
            ret.get_analysis()["vwr"],
            list(tr.get_analysis().items())[0][1],
        ]
    print(
        "Investing monthly, increasing {:.2f}%, starting from ${}, from {} to {}, {:.1f} years".format(
            (monthly_params["coef"] * 100) - 100,
            monthly_params["sum"],
            actualStart.date(),
            actualEnd.date(),
            (actualEnd - actualStart).days / 365,
        )
    )
    with pd.option_context("display.max_rows", None, "display.max_columns", None):
        # print(df[["annual%", "froi%", "cost", "total_value", "max_dd%", "max_md"]])
        print(df)


if __name__ == "__main__":
    stockList = ["^GSPC"]
    monthly_params = dict(sum=1000, coef=1, t_rate=1 + 0.02 / 12)
    # startDate = datetime.datetime.fromisoformat("2000-01-01")
    # endDate = datetime.datetime.fromisoformat("2022-01-01")
    print(stockList[0])
    """
    for period_years in (10, 20):
        end_date = datetime.datetime.now()
        start_date = end_date - datetime.timedelta(days=period_years * 365)
        stockData = get_data(stockList[0], start_date, end_date)
        simulate(stockData, monthly_params)
    """
    period_years = 5
    stockData = get_data(
        stockList[0],
        datetime.datetime.fromisoformat("1900-01-01"),
        datetime.datetime.now(),
    )
    start_date = stockData.index[0]
    end_date = start_date + datetime.timedelta(days=period_years * 365)
    while start_date < datetime.datetime.today():
        stockData = get_data(stockList[0], start_date, end_date)
        simulate(stockData, monthly_params)
        start_date, end_date = (
            end_date,
            end_date + datetime.timedelta(days=period_years * 365),
        )