#36 — FIFO-Trends bar-close fill correction

Hypothesis — what it tests

Does the FIFO-Trends P&F edge survive when trailing-stop exits are filled at the realistic M5 bar-close price (market-order-after-detection) instead of at the exact trail trigger level?

Description

The FIFO-Trends Point & Figure trend-follower had passed walk-forward validation with strong out-of-sample numbers (e.g. GBP_JPY 71.6 p/d), but its backtest assumed trail exits fill at the exact trail level, as a broker stop order would. The live service instead managed exits manually at M5 bar close, where the bar can close 3-10+ pips beyond the trail trigger. This experiment reran all 8 validated configs (ft and ft2 variants on GBP_JPY, USD_JPY, EUR_JPY, GBP_USD) with a bar-close fill model. Every config flipped from positive to deeply negative: win rate collapsed from ~70% to ~34%, and OOS p/d degraded by 54 to 230 pips/day. Since 89-99.9% of exits were trail exits, the edge lived entirely in the fill assumption — the fix identified was broker-side trailing stops or 5-second S5 exit monitoring.

Key result

Indicators

Algorithms

Code