MEAN OOS RETURN
RET / %Performance by family
-4%-2%0%
FX CFD / PROP EVALUATION / WALK-FORWARD LAB
Find the edge.
Respect the limit.
A compact readout of the strategies that survived the research funnel — and the risk rules that stopped them.
DESK VERDICTNO PASS
0/6
families clear every prop guard
Keep the research.Do not deploy.
ACCOUNT GUARDfixed fractional risk / no synthetic trades
RISK / TRADE0.40%
DAILY LOSS3.00%
TOTAL DD5.00%
MIN / DAY2 trades
01 / THE SHORTLIST
Top family representatives
Seven major pairs · 21 chronological folds · one-bar delayed execution
BEST CURRENT READ
TOPBollinger reversion
-1.33%mean return
Worst drawdown5.93%
Trades / day1.58
Positive folds7 / 21
Daily breaches13
best ≠ deployable
FAMILY EVALUATION MATRIX
ALL VALUES OOSRisk before reward
| Family | Representative | Return | Worst DD | Trades / day | Breaches | Decision |
|---|
02 / RESEARCH NOTES
What the traces told us
2,498 closed trades · trace-led iterations · no parameter mining
Stops are doing the talking.
Across the trace set, most failures are full stop-outs around -1R. Winners usually arrive as clean +2R target hits.
62%typical stop share
Middle-band exit helped.
Replacing the Bollinger 2R wait with a center-band exit improved the 1h mean from -1.33% to -0.99%.
+0.34ppimprovement / still reject
Frequency is the trap.
Dual TSMOM reaches about 2.01 trades/day on 15m, but stays negative and breaches the daily guard. Never force the count.
0 / 6families pass all guards
METHOD / 04
How to read this board
These are normalized research ports, not exact PineScript parity. Signals execute one bar late with spread, slippage, ATR stops, fixed fractional sizing, daily loss halts, and total drawdown halts.
“Top” means highest walk-forward score inside a family. It does not mean funded-ready. The data source is Yahoo research candles; broker bid/ask and financing costs still need replacement before any live decision.