Why does sell still happen even when I remove the self.sell() line?

[kaionwong]

Just started trying backtesting for the last 2 days. I am trying to understand the behaviour of buy and sell in the def next(self) section. Here is my code:

from src.helper.generic import pandas_output_setting
from backtesting import Backtest, Strategy
from backtesting.test import GOOG
import datetime
import sys

pandas_output_setting()


def weekday_from_date(date_timestamp):
    date = datetime.datetime.fromtimestamp(date_timestamp / 1e9)  # assume timestamp is in nanoseconds
    # 1 = monday, 2 = tuesday, ... 7 = sunday
    week = [1, 2, 3, 4, 5, 6, 7]
    year = date.year
    month = date.month
    day = date.day
    return week[datetime.datetime(year, month, day).weekday()]


def convert_timestamp_list_to_weekday(n_array):
    return list(map(weekday_from_date, n_array))


class DayOfWeek(Strategy):
    def init(self):
        curr_timestamp_list = self.data.index.values.tolist()
        self.day_of_week = self.I(convert_timestamp_list_to_weekday, curr_timestamp_list)

    def next(self):
        if self.day_of_week == 3:
            self.buy()
        elif self.day_of_week == 5:
            self.sell()


bt = Backtest(GOOG, DayOfWeek, commission=.002, exclusive_orders=True)
stats = bt.run()
##bt.plot()

print(stats)

This is a test strategy that buys on Wednesday and sells on Friday (supposedly). And the backtesting results are:

Start                     2004-08-19 00:00:00
End                       2013-03-01 00:00:00
Duration                   3116 days 00:00:00
Exposure Time [%]                    99.72067
Equity Final [$]                  30863.48394
Equity Peak [$]                   49063.84694
Return [%]                         208.634839
Buy & Hold Return [%]              703.458242
Return (Ann.) [%]                   14.135545
Volatility (Ann.) [%]               39.887952
Sharpe Ratio                         0.354381
Sortino Ratio                        0.628661
Calmar Ratio                         0.206254
Max. Drawdown [%]                  -68.534642
Avg. Drawdown [%]                   -6.690536
Max. Drawdown Duration     1942 days 00:00:00
Avg. Drawdown Duration       82 days 00:00:00
# Trades                                  434
Win Rate [%]                        55.990783
Best Trade [%]                      20.770083
Worst Trade [%]                    -21.345179
Avg. Trade [%]                       0.261093
Max. Trade Duration          21 days 00:00:00
Avg. Trade Duration           8 days 00:00:00
Profit Factor                        1.238429
Expectancy [%]                       0.376821
SQN                                  0.797997
_strategy                           DayOfWeek
_equity_curve                             ...
_trades                        Size  Entry...
dtype: object

Process finished with exit code 0

Now I remove the sell action by removing the following code:

        elif self.day_of_week == 5:
            self.sell()

However, I got the exact same results. Why is that?

[jamesjonesmath]

@kaionwong

Your day of week is never 5. It only generates values 1, 2, 3, 4, and 7.

[Lolser9]

Yeah, you're right. 7 - Monday, 1 - Tuesday, 2 - Wednesday, 3 - Thursday, 4 - Friday. I don't believe that's what you intended @kaionwong , but it seems like an easy fix to just change the values to 2 for buy and 4 for sell instead.

[kaionwong]

The time-zone seems to be the cause of the shifting of the day-number pair by 1. I have use self.position.close() instead of self.sell() which does what I intended to do.

[Lolser9]

Another fix to map 1 to Monday, 2 to Tuesday, and so on is changing the order of the week list inside the weekday_from_date function.

This will return the desired output
week = [2, 3, 4, 5, 6, 7, 1]