Hi,
Is it possible to somehow improve the speed of backtesting?
If I try to do anything fancy such as using pipeline, kalman filtering or clustering the backtest takes hours to complete. I have to literally run it overnight and wait for results next morning.
I have more success downloading daily data from yahoo and running it locally on my Jupyter notebooks. Please look into this. There could be a way to speed up things. Maybe you could offer a fast backtest that does not compute all the numbers and only shows basic statistics. This is useful for us to do some iterations before running a "formal" backtest.
Best regards,
Pravin