Research rounds
5 of 5 locked
Forward window
2026-09-21 to 2027-03-31
Next interim look
2026-10-31
Live trading
OFF

BFG · Research Command

Research proposes, the backtest disposes.

Variant research program. Every round pre-registered, every number on the record, no tuning after results.

Live trading off ← Back to control surface

Owner quick links

Every important link for Michael as owner, all on this one page. No hunting.

Vault

Every BFG document. Enter the password to open it. Cloudflare is the vault.

BFG Vault

Locked

Enter the password to open the vault.

Surviving variants

Three variants survived all four historical rounds. Frozen definitions, no parameter changes since discovery.

V3-L5 Broad-Session RSI Divergence

Temporally robust
Round 2
+0.502R, PF 2.76, n=120
Round 3
+0.389R, PF 2.15, n=964, 8/8 pairs positive, worst +0.268
Round 4
+0.402R, PF 2.21, 4/4 blocks, 7/7 pairs
Gold
REJECTED

V3-L3 Low-Threshold Divergence

Temporally robust
Round 2
+0.465R, PF 2.52, n=142
Round 3
+0.375R, PF 2.08, n=1158, 8/8 positive, worst +0.290
Round 4
+0.376R, PF 2.09, 4/4 blocks, 28/28 pair-block cells positive
Gold
INCONCLUSIVE, sample size (11/12 blocks positive expectancy)

V2-EMA20 Reduced-Filter Channel

Temporally robust
Round 2
+0.378R, PF 2.24, n=107
Round 3
+0.285R, PF 1.79, n=915, 8/8 positive, worst +0.219
Round 4
+0.285R, PF 1.79, 4/4 blocks, 7/7 pairs
Gold
REJECTED, edge looks FX-specific

Raptor proving · real numbers

Owner eyes only. Exact expectancy per version, P2B through the target bar. Public site shows the shape without the figures.

0.5R 0 -0.5R -1.0R expectancy (R per trade) TARGET: expectancy > 0 · PF > 1.00 -1.00R -0.139R ? ? P2B EXTINCT · 4 trades 4 full stop-outs · -$3,547.42 P2C control · 100 trades · PF 0.78 net -13.82% · max DD 27.59% Raptor 1 P2D · flip the signal untested Raptor 2 P2E · flip the action untested Target the bar, not a prediction

P2B died after 4 trades. P2C survived 100 and still pays the market -0.139R per trade. The raptors change one gene each. Full worksheet:

Backtest ledger (PDF)

Round timeline

Protocols locked before every run. No parameter changed after any result.

R1

Discovery

EURUSD, 20 months, 10,605 H1 bars. V3 RSI divergence +0.635R, PF 5.13, n=13 the only genuine positive. Baseline −0.139R.

Locked · Done
R2

Ladders

20 runs, same data. Three triple-digit positives, stable plateaus, no isolated peaks. V1-RSI45 revived at +0.490R, n=30.

Locked · Done
R3

Cross-pair

64 runs, 8 unseen pairs, pre-registered. V3-L5, V3-L3, V2-EMA20 SURVIVED. V1-RSI45 INCONCLUSIVE by one trade (USDCAD 19/20, left benched). V9c INCONCLUSIVE (PF 1.21).

Locked · Done
R4

Temporal

3,371 trades, 7 pairs x 4 blocks + gold. All three TEMPORALLY ROBUST. Blocks pooled expectancy +0.195 to +0.469. Gold did not confirm.

Locked · Done
R5

Forward validation

LOCKED, waiting. Window 2026-09-21 to 2027-03-31. Needs 6 complete months + 150 trades/variant. Monthly interim looks from 2026-10-31. Verdict bars: E≥+0.10R, PF≥1.30.

Locked · In progress

Milestones

What is next, in order.

  • Research video

    A narrated walkthrough of BFG: the company, the tests, the benchmarks, the accomplishments.

    In production

  • MetaTrader handoff

    The EA needs a MetaEditor compile on Michael’s PC. The next human step, when research gives way to execution.

    Pending

  • Paper trading

    Requires forward confirmation, a frozen paper protocol, and explicit authorization.

    Not started

Honest caveats

Read before celebrating.

  • Pairs are not independent experiments.
  • Same calendar window throughout.
  • Rounds 1-2 selection bias means rounds 3-4 tested transferability, not unbiased discovery.
  • No spread, commission, or slippage modeled.
  • Nothing here authorizes live trading or guarantees profit.

Restricted

BFG

Type the password to enter.