Threading/ProcessPooling Backtest.Optimize

[KitHaywood]

Hi There,

I am trying to perform an optimised backtest on my user defined strategy. The scope of potential combinations for these 2 indicators is large, and therefore the cost-to-time for these computations is high.

Is there a way to thread/process-pool backtesting.optimize? I'm currently thinking that I can chunk the indicator combinations up and then backtest.optimize again on the set of outcomes of the chunks - do you think this is the best way forward?

Any advice on how to thread/processpool would be very helpful,

Thanks,

Kit

[kernc]

chunk the indicator combinations up

This is how Backtest.optimize(method='grid') works internally. Unfortunately not so well on Windos. Ideally, someone would have a look at implementing proper multiprocessing on Widows:

backtesting.py/backtesting/backtesting.py

Lines 1268 to 1269 in 94d20da

	.. TODO::
	Improve multiprocessing/parallel execution on Windos with start method 'spawn'.

backtesting.py/backtesting/backtesting.py

Lines 1353 to 1382 in 94d20da

	# Save necessary objects into "global" state; pass into concurrent executor
	# (and thus pickle) nothing but two numbers; receive nothing but numbers.
	# With start method "fork", children processes will inherit parent address space
	# in a copy-on-write manner, achieving better performance/RAM benefit.
	backtest_uuid = np.random.random()
	param_batches = list(_batch(param_combos))
	Backtest._mp_backtests[backtest_uuid] = (self, param_batches, maximize) # type: ignore
	try:
	# If multiprocessing start method is 'fork' (i.e. on POSIX), use
	# a pool of processes to compute results in parallel.
	# Otherwise (i.e. on Windos), sequential computation will be "faster".
	if mp.get_start_method(allow_none=False) == 'fork':
	with ProcessPoolExecutor() as executor:
	futures = [executor.submit(Backtest._mp_task, backtest_uuid, i)
	for i in range(len(param_batches))]
	for future in _tqdm(as_completed(futures), total=len(futures),
	desc='Backtest.optimize'):
	batch_index, values = future.result()
	for value, params in zip(values, param_batches[batch_index]):
	heatmap[tuple(params.values())] = value
	else:
	if os.name == 'posix':
	warnings.warn("For multiprocessing support in `Backtest.optimize()` "
	"set multiprocessing start method to 'fork'.")
	for batch_index in _tqdm(range(len(param_batches))):
	_, values = Backtest._mp_task(backtest_uuid, batch_index)
	for value, params in zip(values, param_batches[batch_index]):
	heatmap[tuple(params.values())] = value
	finally:
	del Backtest._mp_backtests[backtest_uuid]

[KitHaywood]

I'll give that a go, I'm running it on linux so should work, thanks

[ramzerof]

Note for Mac users:
the default start method on Mac is spawn, so the multiprocessing will not work on Mac out of the box.
To enable multiprocessing on Mac set the start method manually to 'fork' before running bt

import multiprocessing

multiprocessing.set_start_method("fork")