Walk-foward optimization - how to pick a better backtest configuration to foward?

 

Hi.. I´m doing Walf-forward optimization and I´d like to know if there is some criteria to pick a better configuration from backtest to run it on foward.

I´m trying to find this procedure to use it in the last backtest optimization, on pre-living. Then, a better configuration can be picked up to run in real live trades.

I´m making maximum complex criteria WFO and then I pick a configuration with a better Sharpe and with several samples next to a configuration. It seems to be good because foward test is profitable.

But, please let me know if there is another way.

 

You can run with 2/3 - 1/3 model with simple strategy tester


2/3 period for optimizing - and 1/3 last period for forward test simulation.

i think it could be based rules for "anti curve fitting"

 
Rachmat Sadikin:

You can run with 2/3 - 1/3 model with simple strategy tester


2/3 period for optimizing - and 1/3 last period for forward test simulation.

i think it could be based rules for "anti curve fitting"

Ok. Thanks! I´m doing a first 1/2 WFO and then with a mean better parameters I´m doing a 1/4 WFO, with these better parameters and some values intervals. A anti curve fitting trading system is the deal... Let´s rock it!

 
You had an idea. You coded it. You back tested it and it doesn't work. You check your code for coding mistakes, 5 times... Still the backtest produces bad results. Solution: time to move on, don't waste your time.  You forgot you came here to make money and now you're just trying to feed your ego.  Wake up. Make money or bust 😁
 
Pawel Michalowski:
You had an idea. You coded it. You back tested it and it doesn't work. You check your code for coding mistakes, 5 times... Still the backtest produces bad results. Solution: time to move on, don't waste your time.  You forgot you came here to make money and now you're just trying to feed your ego.  Wake up. Make money or bust 😁
You are right.... good robust trading systems must be profitable on different market cenarios... but let's try another strategy... 😄
Reason: