How does the multiprocessing work in windows?

[ahfattahi]

``Hi,

I am running the backtest.optimizer() on a small dataset, but only one core of CPU is being used. There is a TODO written in the library description that implies 'spawn' method is not yet implemented in the code. Does it mean windows users cannot utilize multiprocessing? if not, how can I force the optimizer to use more CPU cores?

Here is my example code:

from multiprocessing import set_start_method
from backtesting import Strategy, Backtest
from backtesting.lib import resample_apply

import talib as ta

class TestStra(Strategy):

    RSI_level = 30
    SO_level = 30
    # MACD mode: macd=0,  macdSignal=1, macdHist=2
    MACD_mode=2
    # SO mode: k=0, d=1
    SO_mode = 1

    set_start_method = 'spawn'
    
    def init(self):
        # Call Indicators
        self.MACD = self.I(ta.MACD, self.data.Close)
        self.MACD = self.MACD[self.MACD_mode]

        self.SO = self.I(ta.STOCH,self.data.High,self.data.Low,self.data.Close)
        self.SO = self.SO[self.SO_mode]
        
    def next(self):
        
        if (self.position.is_long and
            self.MACD[-1] > 0 and
            self.MACD[-2] > self.MACD[-1]):
            self.position.close()

        elif (self.position.is_short and
            self.MACD[-1] < 0 and
            self.MACD[-2] < self.MACD[-1]):
            self.position.close()

        elif self.SO[-1] < self.SO_level :
            self.buy()

        elif self.SO[-1] > (100-self.SO_level):
            self.sell()

from backtesting.test import EURUSD

backtest = Backtest(EURUSD, TestStra, cash= 1000, margin=0.03)

stats, heatmap = backtest.optimize(
    RSI_level=range(10, 50, 5),
    SO_level=range(10, 50, 5),
    maximize='Equity Final [$]',
    max_tries=200,
    random_state=0,
    return_heatmap=True)

print(backtest.run())

[ahfattahi]

If parallel computing is not available for Windows yet, do you have a suggestion for using other parallel computing libraries?

[stnatter]

I like this: https://github.com/erdewit/distex
It is a super sleek package I have been using in my own system without any problems.

[ahfattahi]

Thanks. I don't know how to define parameter ranges ( as it is defined in backtest.optimize() ).
Would you please provide an example where you use this package with backtesting.py and define parameters ranges?

[ahfattahi]

I like this: https://github.com/erdewit/distex It is a super sleek package I have been using in my own system without any problems.

Thanks. I don't know how to define parameter ranges ( as it is defined in backtest.optimize() ).
Would you please provide an example where you use this package with backtesting.py and define parameters ranges?

[efeint01]

@ahfattahi @stnatter Long time. Did you guys find any solutions for that? How can we run distex for backtesting.py?