Workbench

Test an allocation without fooling yourself

Most backtesters let you tweak and re-run until something looks good. Run twenty variants against a 5% bar and one clears it by construction — that is what a 5% bar means. Here you fix the hypothesis first, the out-of-sample half is held back, and the bar tightens every time you try another version of the same idea.

1. Set the allocation

Weights must sum to 1. Long only.

2. State what would count as success

Written down before you see the result, so it cannot be decided afterwards. This is the single most effective thing you can do to avoid fooling yourself.