WF results scare me

[renkoben]

my walk forward results scare me. Will I not be able to develope a succesful strategy?

My strategy at this point has 5 out of 7 entries working. I'm doing optimizations and then walk forward runs on individual and larger portions of the strategy (like 4 of the entries)

typically on optimization for 12 months (5 min chart) I can get 65% profitable and a PF of 1.72 or so.

When I do a WF on this with an optimization period of 6 months and test period is 3 months- time frame is 2 years the percent profitable (agreget) drops to 59% and the profit factor drops to .92, so I lose money.

I'm new at Walk forward. What do you suggest?

am I doing the WF wrong. or both of course I would like to get the profit factor way above 1.00 on the agregate for the WF but it looks bad at this point.

or are my entries are flawed? ie my entries have to be better?

[NinjaTrader_Bertrand]

renkoben, I suggest you google overoptimization / curvefitting to learn more about those drawbacks often faced when using optimization in general. What you're seeing here is not so untypical for WF results, it's harder for your strategy to remain consistent / profitable on the unseen data > the data it did not 'know' about as you originally adjusted it's paramters via the optimizer. So I don't think it's your usage of the WF tool, but it points out where you would still need to work in your strategy setup. Are componets of your logic for example adaptive to volatility seen? There are numerous way to adapt script, this can provide a means to stabilize results potentially, as adaptation could take place if the environment changes. I would also look into inspecting the dataset where your script 'fails' to learn what is different here from the data in the optimization period, so potentially new filters / approaches could be added to handle those changes more efficiently.