Parallelism: Ability to run multiple strategies/pairs concurrently within a backtest

[dkoleary88]

When backtesting, it would be nice to be able to run multiple strategies at the same time (to improve performance)
i.e. a single bar event can be used by multiple strategies, and each strategy could keep its own exchange balance. This is useful when trying to optimize parameters.

Also it would be nice to have some helpers to be able to load multiple pairs easily into the exchange.

[gbeced]

Have you tried something like this ?:

def setup_sma_strategy(event_dispatcher, bar_source, sma_period):
    pair = bs.Pair("BTC", "USD")
    exchange = backtesting_exchange.Exchange(
        event_dispatcher,
        initial_balances={"BTC": Decimal(0), "USD": Decimal(10000)}
    )
    exchange.set_pair_info(pair, bs.PairInfo(8, 2))
    exchange.add_bar_source(bar_source)

    strategy = SMA_Strategy(event_dispatcher, sma_period)
    exchange.subscribe_to_bar_events(pair, strategy.on_bar_event)

    # Connect the position manager to the strategy signals.
    position_mgr = PositionManager(exchange, Decimal(1000))
    strategy.subscribe_to_trading_signals(position_mgr.on_trading_signal)


async def main():
    logging.basicConfig(level=logging.INFO, format="[%(asctime)s %(levelname)s] %(message)s")

    event_dispatcher = bs.backtesting_dispatcher()
    bar_source = csv.BarSource(bs.Pair("BTC", "USD"), "bitstamp_btcusd_day.csv", "1d")
    setup_sma_strategy(event_dispatcher, bar_source, 15)
    setup_sma_strategy(event_dispatcher, bar_source, 30)
    setup_sma_strategy(event_dispatcher, bar_source, 45)
    await event_dispatcher.run()

You can get the missing code from README.md.

[dkoleary88]

Thanks, that works. Is there an easy way to log out or return statistics like trades count, win rate, avg profit/loss etc?

[gbeced]

No, at the moment Basana doesn't ship with metrics and/or analytics.