#22 — Weekend gap-fill backtest

Hypothesis — what it tests

Is trading against weekend gaps (betting the gap fills Monday) positive-expected-value after spread costs?

Description

A test of the classic folk hypothesis that the price gap between Friday's close and Sunday's open tends to 'fill' on Monday. For every weekend gap across 12 pairs over 5+ years (2,643 gaps), a trade was simulated against the gap, targeting the Friday close, with a 48-hour max hold and per-pair spread deducted. The tendency is real — 92.2% of gaps filled within 48h, +3.51 pips/trade net, all 12 pairs positive, with a 10–20-pip gap sweet spot — but 8 of the 10 worst adverse excursions came from a single weekend (the 2022-12-18 BoJ policy shift), drawing down 450–660 pips before 'eventually filling'. Because most gaps also move all pairs the same way (~1 effective trade, not 12), it was filed as a research finding, not a deployable strategy: it would need a hard stop, a central-bank calendar filter, and a correlated-exposure cap. Grouped in the audit table under the realizable-under-finite-margin theme — the average hides a tail the account cannot survive.

Key result

Algorithms

Code