DEV · Matt Berg · Tap n Barrel locked rules template one · staging · production untouched
Matt Berg · locked rules template · staging
Tap n Barrel locked rules template one
session sweep -> fakeout -> BoS -> 0.618
Template coded in system: Yes
Berg universal plugin (nitro-setups/matt_berg_youtube/berg_universal_plugin, config configs/berg_nq_doc.yaml) and smt-python/templates/berg_full_mech_shared.yaml (status partial). berg_full_mech.yaml and berg_hybrid_2025.yaml are retired since the 2026-10-04 audit.
Trade evidence pages show the Berg Full Mech forensic run of 2026-10-04 (diagnostic, below 100 trades per window). It is the pre-audit-fix run; the corrected rerun will replace it.
The trade in his words
In London (08:00 UK) or New York (14:30 UK for US100), wait for a sweep of the previous session high/low in line with bias, a fakeout into the 0.079 zone, then a 5m break of structure. Put the fib on the move, place a limit at 0.618, stop at 1.025, fixed 3R target, hold to TP or SL.
B1 Trade in his session windows (US100: London 08:00 UK, NY 14:30 UK) Xewlptt_fMo @ 1:04 · “US 100 session times are 2:30 open for New York, 8:00 a.m. for London session”
B2 Previous-session level sweep, in line with bias IEwAm49ew4k @ 5:57 · “Previous session level sweep. Okay, now this is imperative that this is in line with bias”
B3 Fakeout: the counter candle taps the 0.079 fib zone ef34bFxWbeI @ 3:04 · “I'm going to wait for price to tap back into the 0.079 zone with this wick”
B4 5-minute break of structure in the bias direction 8K4dFyrucXk @ 5:17 · “a 5-minute break of structure in the direction of my bias is what I'm waiting”
B5 Bias agrees: 4H BoS candle close (Full Mech) or D/4H/1H (Hybrid) lVnA4XWQkCE @ 0:57 · “you've got daily, 4hour, 1 hour, and then the bias on this is literally just 4 hour”
B6 Limit order at the 0.618 (not chased/moved) IqLTWhH5Nm8 @ 9:53 · “Our limit order is on the 0618 on the fib tool”
B7 Stop at 1.025, fixed 3R target, hold to TP/SL IqLTWhH5Nm8 @ 5:43 · “we're putting our stop loss at the 1.025 mark ... our TP is just fixed three risk-to-reward”
Locked rule set (draft v0.1, 2026-10-04 AEST)
01. Instrument & data
Enable ........................... True
Instruments ..................... Data-driven: Day13 five (EURUSD, GBPUSD, AUDUSD, gold, Nasdaq100); main book EURUSD, XAUUSD, US100, CHFJPY [Tapnbarrel PDF; spec §8b (PARTIAL)]
ExecutionTimeframe .............. 5m [Tapnbarrel PDF]02. Sessions (UK clock)
Enable ........................... True
IndicesSessionLines ............. Asia 00:00 / London 08:00 / NY 14:30 [Tapnbarrel PDF]
TradeAsia ....................... False (Asia = levels only) [Tapnbarrel PDF; Asia lock]
LondonLatestFill (NQ) ........... 14:00 [spec gap+hunt2 lock]
NYLatestFill .................... 18:45 [spec gap+hunt2 lock]
FridayFlat ...................... 20:00 (PDF); 2026-10-04 audit found the code using 16:55 ET [Tapnbarrel PDF; AUDIT.md]03. Bias
Enable ........................... True
Use4HFractalBias ................ True: most recent 4H fractal BoS, body close (wick ≠ BoS) [Tapnbarrel PDF]
FractalIndicator ................ Rachel_T fractals [Tapnbarrel PDF]04. Setup: sweep + fakeout
Enable ........................... True
RequirePriorSweep ............... True: London sweeps Asia H/L; NY sweeps London H/L; never Asia→NY [Tapnbarrel PDF; Asia lock]
UseFakeout079 ................... True: return to Fib 0.079 zone of current + previous session range [Tapnbarrel PDF]
Use5mFractalBoS ................. True, body close, in bias [Tapnbarrel PDF]05. Entry
Enable ........................... True
FibEntryLevel ................... 0.618 (goldilocks 0.5–0.618; this pack uses 0.618) [Tapnbarrel PDF]
MissedLimit ..................... Extend Fib if no tap; cancel at window end / 30 min before next window / 15 min before news [spec ops lock]06. Stop loss
Enable ........................... True
FibStopLevel .................... 1.025 [Tapnbarrel PDF]07. Take profit
Enable ........................... True
TargetRR ........................ 3R [Tapnbarrel PDF]
SetAndForget .................... True: no partials, BE or trail [spec ops lock]08. Risk & limits
Enable ........................... True
MaxTradesPerSession ............. 1 per session per account [spec ops lock]
RiskPerTrade .................... Evals 1.66% big FX / 1% small·futures; live funded 0.5–0.7% (default 0.5%) [spec ops lock]09. Filters
Enable ........................... True
NewsFilter ...................... ±15 min around PPI/CPI/NFP/Trump/Powell/Core PCE/Claims/JOLTS/PMI/FOMC (ForexFactory); not yet built in code per audit [spec ops lock; AUDIT.md]
Tuning 1 Jan – 30 Apr · four months testOut-of-sample 1 May – 31 Jul · three months testHoldout 1 Aug – 14 Sep · about six weeks testCommission $5 round turn, NQ
01
Lock the rules before testing
Every rule is written in full as a parameter list (sessions, setup conditions, entry, stop, target, max trades, bias) with a version number and a date before any data is run. Nothing changes once testing starts. Any change is a new version, and that version starts again at rule 1.
Name the instrument and the data file. Times are New York time, prices are rounded to the tick, and contract roll weeks are removed. The same cleaned data is used in every window.
This is the only window where settings can be adjusted. Rank settings by net profit after commission. Anything with fewer than 100 trades doesn't qualify. Keep the top 20 and lock them.
Out-of-sample: 1 May to 31 Jul (three months test)
Run the locked settings once, exactly as they are. No changes after seeing the results. A setting passes if it makes a net profit and has a profit factor above 1.
A trade only counts if it passes the trader's written A-plus checklist at entry: right session, liquidity swept, bias agrees, entry model confirmed, and stop and target where the checklist says. Failures are logged as B or C trades and kept out of the results.
Every known real A-plus trade from the trader's videos or posts is traced through the A-plus definition, each filter (session, bias, sweep, timeframe, news, caps), the Roman/ZZ entry path (trigger, confirmation, order type, expiry) and the fill. Record the exact step where each one dies. Misses are fixed in the next version, never the current one.
Every result shows, in order: test number, instrument, length in words, average trade, profit factor, win rate, equity curve; then average signed ticks, trade count and drawdown.
Windows run in order, with no extra out-of-sample passes and no retuning after a peek. Under 100 trades or under about three months is a diagnostic, not a pass.