buy_or_hold.py

import sys
from datetime import datetime

sys.path.append("../")
from stock_candle import StockCandle

# Experiment Parameters
nifty_olhc_filename = "../Data/Index/NIFTY 50_OLHC_DAILY_01011999_01012020.csv"
monthly_cash_increment = 1000
percentage_from_all_time_high = 10

"""
This method reads the NIFTY OLHC data from a csv file.
Please note that the data is originally sorted by date in descending order.
"""


def readNiftyCandles(reverse_candles=False):
    file = open(nifty_olhc_filename, "r")
    candles = []

    # The first line in the csv is the header, so we need to skip that.
    first_line_skipped = False
    for line in file:
        if not first_line_skipped:
            first_line_skipped = True
            continue

        # The rows have data in the format Date, Open, High, Low, Close
        row_splits = line.split(",")

        date = datetime.strptime(row_splits[0], "%d %b %Y")
        candle = StockCandle(
            "NIFTY50",
            date,
            float(row_splits[1]),
            float(row_splits[3]),
            float(row_splits[2]),
            float(row_splits[4]),
        )
        candles.append(candle)

    if reverse_candles:
        candles.reverse()

    return candles


"""
This method populates all time high in candles.
NOTE: This method assumes that candles are sorted in chronological order, i.e. oldest candle comes first.
"""


def populateAllTimeHighs(candles):
    all_time_high = 0
    for candle in candles:
        all_time_high = max(all_time_high, candle.close)
        candle.all_time_high = all_time_high
    return candles


candles = populateAllTimeHighs(readNiftyCandles(True))


def hasMonthChanged(previous_date, current_date):
    if previous_date.month < current_date.month:
        return True

    if previous_date.year < current_date.year:
        return True

    return False


def runExperiment(candles, buyCandle):
    # Setting up initialization parameters.
    cash = 0
    units_held = 0
    previous_date = datetime.strptime("31 Dec 1998", "%d %b %Y")
    current_value = 0
    max_cash_held = 0
    max_cash_held_buy_date = None

    for candle in candles:
        # Add to cash account on beginning of the month.
        if hasMonthChanged(previous_date, candle.date):
            cash += monthly_cash_increment

        units_to_buy = buyCandle(candle, cash)
        if units_to_buy > 0:
            if cash > max_cash_held:
                max_cash_held = cash
                max_cash_held_buy_date = candle.date

            units_held += units_to_buy
            cash -= units_to_buy * candle.close

        current_value = cash + units_held * candle.close
        previous_date = candle.date

    print("Cash to begin with: 0")
    print("Monthly cash increment: 1000")
    print("Final holding value: " + str(current_value))
    print("Max cash held at one point: " + str(max_cash_held))
    print("Max cash held buy date: " + max_cash_held_buy_date.__str__())


def buyCandleAsSoonAsYouHaveCash(candle, cash):
    units_to_buy = round(cash / candle.close, 2)

    # Taking care of rounding up errors
    if units_to_buy * candle.close > cash:
        units_to_buy -= 0.01

    return units_to_buy


def buyOnFallFromAllTimeHigh(candle, cash):
    fall_percentage = (
        (candle.all_time_high - candle.close) / candle.all_time_high
    ) * 100
    if fall_percentage < percentage_from_all_time_high:
        return 0

    return buyCandleAsSoonAsYouHaveCash(candle, cash)


print("Running strategy: BuyAsSoonAsYouHaveCash")
runExperiment(candles, buyCandleAsSoonAsYouHaveCash)
print("")

print("Running strategy: BuyOnFallFromAllTimeHigh")
print("Fall Percentage: " + str(percentage_from_all_time_high) + "%")
runExperiment(candles, buyOnFallFromAllTimeHigh)