Optimizing a float parameter

[JasperVanhove]

Hey,

Thank you for this very usefull library. I'm looking for a way to optimize a float variable for a strategy. In my strategies i have integers where i can use the range() function but for my float variable 'atr multiplier' i can't use range. I looked into 'arrange' and np.linspace funtions but can't seem to find a solution (neither on discussions/issues forum).

Any ideas?

[kernc]

Why doesn't np.linspace work for you?

[JasperVanhove]

How do i need to format it?
let's say i want to get use 1.0-3.0, do i right atrMultiplier=np.linspace(1.0, 3.0, num=20) ?

[kernc]

atrMultiplier=np.linspace(1.0, 3.0, num=20)

Excatly. If that doesn't work for you, I think it's a bug in:

backtesting.py/backtesting/backtesting.py

Lines 1304 to 1305 in 267d99f

	def _tuple(x):
	return x if isinstance(x, Sequence) and not isinstance(x, str) else (x,)

with np.ndarray not passing isinstance Sequence check.

Should work, then, if you specify it as:

atrMultiplier=np.linspace(1, 3, 20).tolist()

[JasperVanhove]

Thank you, forget the .tolist() at the end....

[sultank1]

atrMultiplier=np.linspace(1.0, 3.0, num=20)

Excatly. If that doesn't work for you, I think it's a bug in:

backtesting.py/backtesting/backtesting.py

Lines 1304 to 1305 in 267d99f

	def _tuple(x):
	return x if isinstance(x, Sequence) and not isinstance(x, str) else (x,)

with np.ndarray not passing isinstance Sequence check.
Should work, then, if you specify it as:

atrMultiplier=np.linspace(1, 3, 20).tolist()

thanks sir

[Rydo111]

Casting np.arange to a list also worked for me. eg. atrMultiplier=list(np.arange(start=0.1, stop=2.0, step=0.1)). Thanks for this thread; helped me a lot.