Why we resample blocks of trades, not single trades
Monte Carlo on a trade list flatters the tail if you shuffle trades one at a time. Losses cluster. The resampling has to keep the clusters.
Blog · 1 published
Where backtests break.
Notes from testing other people's strategies: the failure modes we see most, the methods that catch them, and the diagnostics that separate a real edge from a lucky parameter. Written by the people who run the tests.
No signals, no tips, no market calls — we test what clients specify and report what we find. RSS