Analysis standard (v1) and the evidence behind it

Last updated 2026-09-24. Everything below was backtested on Binance USDT-M futures (about 8-9 months of candles, top coins by volume), fees and slippage included, and split into a first and second half to check the result is not a one-off. "R" = one unit of risk (the distance to the stop). Not financial advice.

STANDARD v1 (what the live scanner alerts on)

A setup is STANDARD only if ALL of these are true:

Timeframe: 4h (entry candle timeframe).
Direction: long (buy) only.
Daily trend agrees: daily close above the 50-day EMA and the EMA rising.
Support level: the liquidity that was swept is a real level: touched at least twice before, OR it lines up with a daily swing high/low.
ICT sequence: sweep of that level and close back inside, then a market structure shift, then entry at the FVG / order block zone (limit at the zone midpoint).
Stop beyond the sweep extreme, target at the next liquidity level (at least 1.5R).

Everything else (1h setups, sells, no S/R level, against the daily trend) is still scanned and tracked, but does not send alerts, because it did not show an edge.

Backtest of the closest match (4h, S/R touched 2+ times or at a daily level, daily trend aligned): about +0.17 to +0.19R per trade, positive in both halves of the period, buys about +0.40R. About 130-180 trades. Statistical strength is modest (t about 1.7). It is a lead to be confirmed by the live paper tracker, not a proven edge.

What was tested and did not work (do not trade these)

Idea	Result
ICT scanner as first built, all setups	40% win, about 0R after fees
ICT sells (shorts)	Lost money (about -0.08R); live shorts kept being stopped out
ICT 1h setups, with or without S/R	About 0R whatever the filter
Higher R:R (3R+) setups	Worse (18% win)
The scanner's score	Not predictive (score 80+ did worse than 70-79)
Standard TA on 1h/4h (EMA20 pullback, breakout + volume, EMA 9/21 cross, RSI extreme), with or without a BTC trend filter	All lose after fees (-0.03 to -0.12R)
Random entries, same stops/targets	Longs +0.04R, shorts -0.11R (so most TA entries added nothing)
The WADE_FX book method coded literally (hunt of PDH/PDL/PWH/PMH/equal highs + BMS + return to order block, London/NY open, 4h/daily agreement, OTE)	Crypto 15m: -0.23R (book stop) / -0.06R (wide stop); forex 1h: -0.16R / -0.11R. The book's confluence filters made forex worse
Forex news: trade the surprise (actual vs forecast)	The move happens in the first ~60 seconds. Entering 1 minute after the release: 36% win, -3.3bp net. Entering at the instant of release made money but is not achievable by hand

What did work (separate strategy)

Daily trend-following, long only: buy a break of the 20/55/100-day high, exit on a break of the 10/20/30-day low. About +14% to +60% per trade, 38-39% win rate, positive in both halves. Caveats: survivorship bias (today's top coins), the second half was much weaker, holds last 4-11 weeks, and shorts lose.

Rules of use

Paper-trade first. The live tracker at has a "STANDARD v1" row: it needs about 30 closed STANDARD trades before its win rate means anything.
Use a stop on every trade. Risk a small fixed amount per trade.
Do not trade the news release by hand. The forex board is for awareness and risk (know when volatility hits).
Re-run the backtests (backtest.py, sr_backtest.py, ta_lab.py, wade_lab.py, trend_daily.py) before changing the standard. A rule only enters the standard if it stays positive after fees in both halves of the data.