OK System · v2.16

ICT sweep → retest,
London-morning intraday

Born as a discretionary ICT system, made mechanical trade by trade, then tested across 14.5 years of 5-minute data. The full logic is frozen and identical on all three tickers — only the session window differs per market.

How it works

The core in four steps. Exact parameter values stay private; the structure is fully public.

1

Liquidity sweep

In the London-morning window, price raids the Asia range (AS.H/AS.L) or a 5m swing point and closes back inside — a liquidity sweep, not a genuine breakout.

2

Zone required

The sweep must occur inside a higher-timeframe zone (FVG). In validation this proved to be the only "switch" that lifts the hardest block (2017–2021) to break-even and halves the drawdown.

3

Market-structure shift as gatekeeper

Only after a confirmed MSS in the opposite direction is an order placed. Without this requirement drawdown in testing ran up to 76% — the MSS rule is a survival condition.

4

Entry in the gap, hard exits

Limit entry deep in the entry gap, stop with a small ATR buffer behind the sweep extreme, fixed TP per ticker, time-exit 22:45 — never an overnight position. Max 1 trade per day.

Configuration per ticker

One frozen rule set; only the time layer and TP/risk differ per market.

EURUSDGBPUSDXAUUSD
Sweep window09:00–11:3008:00–11:3009:00–11:30
Take-profit2R4Rtime-exit
Risk per trade1.0%0.5%0.5%
Trades/week0.711.090.59
Why different TPs? From the TP-ladder measurement: EURUSD is indifferent between 2R and 4R (2R chosen — steadier years), GBPUSD doubles its expectancy at 4R, and gold gets strangled by a tight TP — there the time-exit does the work.

Deep backtest — judged per time block

Jan 2012 – Jul 2026, 5m data, estimated costs included. Configurations are judged on robustness across three time blocks — never on the headline number, because in-sample winners look good by definition.

TickerPF total2012–162017–212022–26Max DDTrades/wk
EURUSD1.261.131.021.6712.2%0.71
GBPUSD1.171.051.251.2212.2%1.09
XAUUSD1.191.151.161.260.59
Portfolio+227%sum since Aug 2013 · not a single losing block (XAU)15.9%2.40
Correlation EU ↔ GU
+0.05
R-correlation on overlap days — practically zero
Correlation XAU ↔ EU
+0.06
gold moves independently
Correlation XAU ↔ GU
+0.26
low — real diversification

How the configuration was chosen

Eight entry configurations, 14.5 years of EURUSD, judged per time block. The winner is the only one without a losing block and with the lowest drawdown.

MSS required

Without it: 76% drawdown in 2017–21. Non-negotiable.

Zone required

The only switch that lifts the problem block to break-even and halves the DD.

Requiring an "untouched" zone

Tested and dead: strangles all Asia trades, net losing. Touched ≠ toxic.

Counter-trade safety (P40)

Regime-sensitive: shines 2022–26, loses 2012–16. Left off.

Full honesty: no configuration reaches PF 1.4 across the entire 14.5 years. The edge is thin long-term and strong in the current regime. Hence: modest risk, block-robustness as a hard criterion, and a forward test before anything becomes investable.

What was rejected

Measured, published, benched. This is part of the evidence.

🗑️

DAX (GER40)

The first measurement looked interesting but turned out contaminated: 56% of trades stayed open overnight because the time-exit fell outside trading hours. Remeasured cleanly: PF 1.017 — flat across all blocks. Definitively rejected; an annual recheck suffices.

🗑️

Afternoon window 13:00–16:00

Entries before or into the US open get run over (PF 0.40 around 15:00). Excluded.

🗑️

More risk on gold

XAU at 0.75% instead of 0.5%: +31 percentage points of return for +3.2pp of drawdown. Rejected — the ratio isn't worth it.

Open items

What's still on the research list — that's public too.

A news layer (flat before CPI/NFP 14:30 and FOMC 20:00 — most important for gold), a break-even test per ticker, rule-based exits (there is demonstrably profit evaporating between 4R and the time-exit), and the correlation measurement with the MR4H system for the 6-sleeve portfolio.