Walk-Forward Optimization and BarsRequiredToTrade

[walter739]

Hello i have this issue, I use a strategy with 15min but i need a indicator with days dataseries.

For example, if I run a WalkForwrd with the following settings::

a) optimization period of 730 days
b) test period of 360 days
c) and the strategy runs on 15min bars, with a BarsRequiredToTrade setting of 11800 ( i need this quantity for the indicator days dataseries)

The completa date for the walk forwark is 2008 to 2020, but the results not are good. The test period in each year only work when the bars requeried to trade is complete.

It is a problem for have a good walk forward for 12 years, How i can solve this?

Regards

[NinjaTrader_Jesse]

Hello walter739,

How are you currently confirming that the bars required needs to be re met for each year? Can you provide more details on the test you did and how you were checking that?

I look forward to being of further assistance.

[walter739]

Hello Jesse,

There should be an equality between running a single backtest with one set of parameters and a walk forward test using same one set of parameters. As it works now they will never be equal because of the BarsRequireToTrade gap.

I have a Backtest 2008 to 2020 with Bar requerided to trade 11800..This have a result.

i wish compare this result of backtest with the result of WFO, I can not why the gap of bar requerided to trade each year affect the total result in the WFO.

In this moment for compare to the complete 12 years of backtest results. I optimized 720 days, and backtest the next 365 days +150 (11800bars 15min). I need have 11 times for compare results, it is very slowly.

How i can solved this?

Regards

[NinjaTrader_Jesse]

Hello walter739,

I believe in this case I would need a specific example using the SampleMACrossOver and less data so that I can test the situation to see the result. Unfortunately I am having a hard time understanding what the issue is based on your description.

Do you have a way to reduce the values being used, preferably to less than 1 year of data and still see this with the SampleMACrossOver? With that we can view the chart in the test to see when the BarsRequired has been met by seeing its plots and also executions. If you can demonstrate the problem in that way we can try to see what can be done about that.

Please let me know if I may be of additional assistance.

[walter739]

Hello Jesse

I do not describe an error in the platform, but I think that the operation of the WFO is not effective when you have a high number of Bars required to trade in the strategy. And I think that could be better.

I wish i could compare both WFO vs Backtest results. 1/1/20 to 27/8/20

In this example
-Backtest 8 months 661 trades
-WFO 8 months 185 trades

The WFO gives a result if I optimize 90 days and use the 45 day strategy. But this is impossible to compare with the normal 8 month backtest. And the reason is the bars required to trade. The WFO in each of its 45-day backtest begins to generate the strategy after 2000 bars.

I suggest that you have a way to improve this and make it more practical so that in the WFO can compare effectively with the backtest.

Regards

[NinjaTrader_Jesse]

Hello walter739,

Thank you for the additional details on your post.

I will forward this information to development for further review.

Please let me know if I may be of additional assistance.