Strategy Selection

Houston we have a problem:

  • cerebro is not meant to be run several times. This is not the 1st time and rather than thinking that users are doing it wrong, it seems it is a use case.

This interesting use case has come up via Ticket 177. In this case cerebro is being used multiple times to evaluate differet strategies which are being fetched from an external data source.

backtrader can still support this use case, but not in the direct way it has been attempted.

Optimizing the selection

The buil-in optimization in backtrader already does the required thing:

  • Instantiate several strategy instances and collect the results

Being the only thing that the instances all belong to the same class. This is where Python helps by lettings us control the creation of an object.

First, let’s add to very quick strategies to a script using the Signal technology built in backtrader

class St0(bt.SignalStrategy):
    def __init__(self):
        sma1, sma2 = bt.ind.SMA(period=10), bt.ind.SMA(period=30)
        crossover = bt.ind.CrossOver(sma1, sma2)
        self.signal_add(bt.SIGNAL_LONG, crossover)


class St1(bt.SignalStrategy):
    def __init__(self):
        sma1 = bt.ind.SMA(period=10)
        crossover = bt.ind.CrossOver(self.data.close, sma1)
        self.signal_add(bt.SIGNAL_LONG, crossover)

It cannot get easier.

And now let’s do the magic of delivering those two strategies.

class StFetcher(object):
    _STRATS = [St0, St1]

    def __new__(cls, *args, **kwargs):
        idx = kwargs.pop('idx')

        obj = cls._STRATS[idx](*args, **kwargs)
        return obj

Et voilá! When the class StFetcher is being instantiated, method __new__ takes control of instance creation. In this case:

  • Gets the idx param which is passed to it

  • Uses this param to get a strategy from the _STRATS list in which our previous sample strategies have been stored

    Note

    Nothing would prevent using this idx value to fetch strategies from a server and/or a database.

  • Instantiate and return the fecthed strategy

Running the show

    cerebro.addanalyzer(bt.analyzers.Returns)
    cerebro.optstrategy(StFetcher, idx=[0, 1])
    results = cerebro.run(maxcpus=args.maxcpus, optreturn=args.optreturn)

Indeed! Optimiization it is! Rather than addstrategy we use optstrategy and pass an array of values for idx. Those values will be iterated over by the optimization engine.

Because cerebro can host several strategies in each optimization pass, the result will contain a list of lists. Each sublist is the result of each optimization pass.

In our case and with only 1 strategy per pass, we can quickly flatten the results and extract the values of the analyzer we have added.

    strats = [x[0] for x in results]  # flatten the result
    for i, strat in enumerate(strats):
        rets = strat.analyzers.returns.get_analysis()
        print('Strat {} Name {}:\n  - analyzer: {}\n'.format(
            i, strat.__class__.__name__, rets))

A sample run

./strategy-selection.py

Strat 0 Name St0:
  - analyzer: OrderedDict([(u'rtot', 0.04847392369449283), (u'ravg', 9.467563221580632e-05), (u'rnorm', 0.02414514457151587), (u'rnorm100', 2.414514457151587)])

Strat 1 Name St1:
  - analyzer: OrderedDict([(u'rtot', 0.05124714332260593), (u'ravg', 0.00010009207680196471), (u'rnorm', 0.025543999840699633), (u'rnorm100', 2.5543999840699634)])

Our 2 strategies have been run and deliver (as expected) different results.

Note

The sample is minimal but has been run with all available CPUs. Executing it with --maxpcpus=1 will be faster. For more complex scenarios using all CPUs will be useful.

Conclusion

The Strategy Selection use case is possible and doesn’t need circumventing any of the built-in facilities in either backtrader or Python itself.

Sample Usage

$ ./strategy-selection.py --help
usage: strategy-selection.py [-h] [--data DATA] [--maxcpus MAXCPUS]
                             [--optreturn]

Sample for strategy selection

optional arguments:
  -h, --help         show this help message and exit
  --data DATA        Data to be read in (default:
                     ../../datas/2005-2006-day-001.txt)
  --maxcpus MAXCPUS  Limit the numer of CPUs to use (default: None)
  --optreturn        Return reduced/mocked strategy object (default: False)

The code

Which has been included in the sources of backtrader

from __future__ import (absolute_import, division, print_function,
                        unicode_literals)

import argparse

import backtrader as bt


class St0(bt.SignalStrategy):
    def __init__(self):
        sma1, sma2 = bt.ind.SMA(period=10), bt.ind.SMA(period=30)
        crossover = bt.ind.CrossOver(sma1, sma2)
        self.signal_add(bt.SIGNAL_LONG, crossover)


class St1(bt.SignalStrategy):
    def __init__(self):
        sma1 = bt.ind.SMA(period=10)
        crossover = bt.ind.CrossOver(self.data.close, sma1)
        self.signal_add(bt.SIGNAL_LONG, crossover)


class StFetcher(object):
    _STRATS = [St0, St1]

    def __new__(cls, *args, **kwargs):
        idx = kwargs.pop('idx')

        obj = cls._STRATS[idx](*args, **kwargs)
        return obj


def runstrat(pargs=None):
    args = parse_args(pargs)

    cerebro = bt.Cerebro()
    data = bt.feeds.BacktraderCSVData(dataname=args.data)
    cerebro.adddata(data)

    cerebro.addanalyzer(bt.analyzers.Returns)
    cerebro.optstrategy(StFetcher, idx=[0, 1])
    results = cerebro.run(maxcpus=args.maxcpus, optreturn=args.optreturn)

    strats = [x[0] for x in results]  # flatten the result
    for i, strat in enumerate(strats):
        rets = strat.analyzers.returns.get_analysis()
        print('Strat {} Name {}:\n  - analyzer: {}\n'.format(
            i, strat.__class__.__name__, rets))


def parse_args(pargs=None):

    parser = argparse.ArgumentParser(
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
        description='Sample for strategy selection')

    parser.add_argument('--data', required=False,
                        default='../../datas/2005-2006-day-001.txt',
                        help='Data to be read in')

    parser.add_argument('--maxcpus', required=False, action='store',
                        default=None, type=int,
                        help='Limit the numer of CPUs to use')

    parser.add_argument('--optreturn', required=False, action='store_true',
                        help='Return reduced/mocked strategy object')

    return parser.parse_args(pargs)


if __name__ == '__main__':
    runstrat()

Comments

Fork me on GitHub