Trying to Correlate WFO and Optimization Results

[hazza147]

Hi,

I have a EURUSD 5 Min strategy which when I use WFO with it with:

Optimization Period = 3
Test Period = 1

Produces some fairly good results. Now I'm trying to cross check it by correlating each test day with the preceding optimization period so I can start to optimize data for each of my trading days going forward. For example:

The optimized parameters for 09/06/12 in the WFO test = 9/1/45, however then when I optimize the data from 06/06/12 to 08/06/12 (3 preceding days) in my own study it produces optimum parameters of 10/1/37 which differ from 9/1/45.

Can someone explain how I can set my optimization settings so I get parameters of 9/1/45 ? This is the only things stopping me from trading my system. It is profitable according to WFO but how am I ever able to optimize my data for the current trading day if I have no evidence that the 3 preceding trading days I am basing it on provide results similar to that of the WFO test.

Thanks,
Harry

[NinjaTrader_RyanM]

Hi hazza147,

Welcome to the NT forums. What optimization factor are you using here? What is the value of this for the optimization and walk forward run? Do you have a session template that crosses midnight? (Like the FOREX or CME US Index Futures ETH).

[hazza147]

Originally Posted by NinjaTrader_RyanM

Hi hazza147,

Welcome to the NT forums. What optimization factor are you using here? What is the value of this for the optimization and walk forward run? Do you have a session template that crosses midnight? (Like the FOREX or CME US Index Futures ETH).

Hi, What do you mean by optimization factor? Also I have a session template for Forex which I have edited so it starts on Sunday at 10PM GMT and finishes Friday at 10PM GMT. The problem I have is that the WFO is optimising the data including the out of sample test data which means it brings out great results!

[NinjaTrader_RyanM]

Optimization factor is the value you choose to optimize on -- the end goal of both optimization and walk forward. I was curious if these are similar between the two tests.

These two areas are similar but not identical. It may be difficult to line up 100%, especially with that type of session template. When running walk forwards with the FOREX session template (or similar), each walk forward period may double count trades on "transition" dates. This is a limitation seen on the walk forward portion you wouldn't see on the optimization.

When using session templates that are 24 hrs with times cross multiple days (e.g. FOREX session template), it is expected that doing things like backtests will include trades from seemingly the date before and the date after. This is because those dates are required to get the full trading session as defined by the session template.

Example scenario:

First period is 1/29 to 2/25. Second period is 2/26 to 3/25.
First period counts trades from 1/28 to 2/26.
Second period counts trades from 2/25 to 3/26.
There is an overlap of two days of 2/25 and 2/26 being double counted between each walk forward period.

[hazza147]

So basically doing using WFO on 5 Min EURUSD data where optimization period = 3 and Test Period = 1 is pointless?

The test data performs so well because it is included in the data that is being optimized.

[NinjaTrader_RyanM]

Yes, could be. You may need to check all settings you're using here. It sounds like both your optimization period and walk forward are less than a single session. Please post a screenshot of your walk forward and optimization settings, along with the custom session template you're using and we can confirm this.