#23 — FIFO-Trends 12-pair P&F sweep

Hypothesis — what it tests

Does P&F box-chart trend-following with FIFO position management and a tight box-quantized trailing stop give a robust edge across all 12 pairs?

Description

The project's flagship execution-realism story. FIFO-Trends enters after a sustained P&F trend column (box size b, reversal r) with an optional confirmation bar (E2), and exits on a box-quantized trailing stop. A 2,700-config Numba grid on EUR_JPY (all 9 SOP rules audited pre-run) produced 93 OOS survivors, best +24 p/d with an unusually good 79% IS→OOS retention; the same sweep on all 12 pairs found OOS winners on every pair, topped by GBP_JPY at 71.6 p/d and USD_JPY at 68.5 p/d, with the same parameter region (b=5, r=1, E2, 1-box trail) winning everywhere. The caveat was visible from day one: the mean trade was only 3.6 pips against a 2.5-pip spread gate, and 86–91% of trades closed within one M5 bar. Live, the 1-box (5-pip) trail proved unexecutable at real spread/slippage — the live check (audit Entry 53) closed FIFO at 15 trades, −168.5 pips, 7% win rate. Later fill-correction experiments (Entries 35–36) showed the box-quantized fill model itself was much of the edge.

Key result

Indicators

Algorithms

Code