What do you think of this walk forward result?

Please comment on my walk forward result.
It's a mean reserve long system.
The right most column is the element to be optimized.
Thank you!

See attached picture.

1 Like

There is something wrong with your WalkForward settings because you are including 2024 in your "in sample" period, while you should not be doing that.

Thank you for pointing it our Tomasz, I checked "use today" as the last date in the In Sample setting by mistake. but can we just ignore the last 3 rows of IS in my result? Could you please comment if it's an acceptable system? and which value of the element being optimized should I choose for my live trading?

There is great deal of fluctuation of CAR/MDD in out-of-sample periods. Generally you should be looking for consistency and CAR/MDD > 1 (or better yet >2). CAR/MDD measures how much you are earning compared to risk you are taking (drawdown).

Also I am not big fan of "anchored" mode since many metrics are sensitive to analysis period (for example MDD consistently grows with increasing period)

1 Like

+1 to Tomasz's comment about anchored, but -
I think you might have a more serious issue. Your OOS periods overlap your IS periods.
This is not good !
If you are unsure of why this is so bad you might read Bandy's book.
I suspect that you have something similar to the following which I think reproduces your Begin/End period in your pic -
If So, I think you want to un-check "Anchored" in the OOS data.

1 Like

thank you for the valuable input.
I am going to do it again without anchor and will do more tuning of the system.
However, is there any rules regarding which optimized value to choose? Do I choose the most prevalent one for my live trading? what other factors to consider?

thank you for the input and book recommendation. appreciate it .