Home

›

Articles

›

Application & Practice

Application & Practice

ICT Backtesting Guide — How to Backtest ICT on TradingView

ICT backtesting guide — how to backtest ICT concepts on TradingView Replay. The 5-phase progression, pre-replay checklist, journal template, competency benchmarks, 5 anti-look-ahead-bias rules, and live-trading readiness criteria.

The gap between an ICT student who passes every quiz and an ICT trader who executes consistently is 100 structured replay sessions. The student’s analysis is deliberate — too slow for a live London kill zone. The trader’s is automatic. Automatic pattern recognition is built through deliberate, feedback-rich practice on historical data.

Key takeaways

ICT backtesting is discretionary skill training, not system optimisation. It measures YOUR analytical skill on historical data — AMD reading, Judas prediction, PD array quality assessment.
TradingView Replay: step forward one candle at a time. Write analysis BEFORE advancing. The written commitment is the core anti-bias discipline.
5-phase progression: (1) AMD reading only → (2) + kill zones + Judas → (3) + PD arrays → (4) full entry execution → (5) complete pre-session routine. Each phase isolates one skill layer.
Competency benchmarks: AMD ≥80%, Judas prediction ≥65%, valid entry ≥80%, win rate at 1.5R+ ≥50%. Consistently met across 50+ sessions = live-ready.
100 structured, journaled sessions across 3 market condition types = the practical minimum before live trading beyond micro-lots.

Why ICT backtesting is different

Conventional backtesting tests whether a rule-based system produces profitable outcomes — a moving average crossover can be coded and run overnight. ICT backtesting cannot work this way because ICT analysis is inherently discretionary. There is no formula for “identify the Judas swing and wait for the CHoCH.” The AMD phase identification requires human pattern recognition. The PD array quality assessment requires judgment. The entry timing requires reading a lower timeframe for a signal that varies from setup to setup. A human must execute each step, one session at a time.

Setting up TradingView Replay

Access the Replay function

Top toolbar: clock/play button icon. Click to enter Replay mode. Future data hides. Drag the start point to any historical bar. Free accounts have limited access; Pro/Pro+ provides extended history and unlimited sessions.

Step forward — the primary mode

Step forward one candle at a time (keyboard shortcut). Before each advance: write your analysis. Advance. Observe. Write result. Asian session at 2–4x speed. Kill zone: step-by-step. Post-KZ: 2–4x to see distribution completion.

Multi-timeframe setup

Primary: 15M in Replay mode. Second tab: 4H or 1H (not in replay) for HTF context. Navigate both to the same historical date. Periodically switch to 4H to check HTF context as it would have appeared at that moment.

The pre-replay marking checklist

Complete all six items BEFORE advancing a single candle into the London kill zone. Marking analysis after seeing the candles = look-ahead bias = inflated win rates that collapse live.

1. Mark Asian range high and low

Horizontal lines at the high and low of the Asian session. BSL above, SSL below. These are the Judas sweep targets.

2. Mark PDH and PDL

Previous session’s high and low. Primary ERL targets for the AMD distribution.

3. Note EQH/EQL near Asian boundaries

Equal highs/lows near the Asian range amplify the liquidity pools at those levels.

4. Write AMD daily bias

Bullish / Bearish / Unclear. From weekly structure + previous daily candle. Written before the kill zone begins.

5. Write expected Judas sweep side

Below Asian low (bullish) or above Asian high (bearish). This is the prediction that backtesting grades.

6. Mark visible PD arrays

OBs and FVGs in the expected distribution direction on 4H and 1H. The entry zones.

The 5-phase ICT backtesting progression

5-PHASE PROGRESSIONSESSION COUNT

Isolate each skill, then stack. P1 builds AMD reading (no entries). P2 adds kill zone timing + Judas prediction. P3 adds PD array quality assessment. P4 adds full trade execution with journal. P5 adds the complete pre-session routine. Skipping to P4 immediately overloads learning capacity and produces slow, unfocused improvement.

The backtesting journal

Without a journal, backtesting is watching charts with good intentions. The journal forces commitment before revelation, creates a measurable performance record, and provides the statistics needed to assess live-trading readiness. Every 20 sessions: calculate AMD accuracy, Judas prediction rate, entry validity rate, and win rate + average R. These four metrics drive practice improvement — not general feelings about performance.

Competency benchmarks — when to go live

COMPETENCY BENCHMARKSMIN THRESHOLDS

Skill-based readiness, not session count. You are ready for live trading at minimum position size when these four metrics are consistently met across 50+ consecutive journaled sessions. If any metric is below threshold: continue backtesting, focusing extra practice on the lagging skill.

Avoiding look-ahead bias — the 5 integrity rules

Rule 1: Write BEFORE advancing — never advance first

The primary anti-bias rule. Before pressing step-forward: write AMD phase, Judas target, entry zone, stop, target in the journal. Then advance. Writing before seeing = genuine prediction, not post-hoc rationalisation.

Rule 2: Mark all pre-session levels before the kill zone

Asian range, PDH/PDL, EQH/EQL, PD arrays — all marked before advancing into the London window. Marking “as you see them develop” during the kill zone = partial look-ahead.

Rule 3: If you advance too far, reset — never backtrack

Accidentally advanced 10 candles? Do not scroll back. Note “session contaminated” and restart the date from scratch. Backtracking with future knowledge is the most blatant look-ahead bias.

Rule 4: Record results immediately

At the moment you close a replayed trade, record it. Don’t wait to see what happens after — watching further and retroactively grading = hindsight bias.

Rule 5: Skip sessions you recognise from memory

A major news event or chart pattern you specifically remember? Skip it. Your memory will contaminate every analysis decision. Replay only unfamiliar sessions.

Common mistakes

Replaying without the pre-session checklist

Starting the kill zone without marking Asian range, AMD bias, Judas target, and PD arrays = watching a chart, not practising trading. The pre-session analysis is half the skill.

Not keeping a journal

50 unjournaled sessions produce indistinguishable performance from 50 random chart watches. The journal converts watching into deliberate practice with measurable feedback.

Only replaying easy, trending conditions

Textbook Judas days build only a textbook skill set. Include ranging weeks, news weeks, ambiguous AMD days. The difficult sessions produce the most valuable skill development.

Going live before reaching competency benchmarks

With 10–20 replay sessions, pattern recognition is not automatic. Live trades require deliberate analysis that is too slow under real-money emotional pressure. More replay time = always the correct decision when benchmarks aren’t met.

Not distinguishing valid vs invalid losses in the journal

A valid loss (correct process, unfavourable outcome) and a genuine error (incorrect AMD reading, invalid entry) both look like “losses.” Recording both identically prevents accurate diagnosis of which skills need work.

FAQ — ICT backtesting

How many sessions before going live? +

Minimum 50 consecutive sessions meeting competency benchmarks. Practical guideline: 100 structured, journaled sessions across trending, ranging, and news conditions. 50–75 hours total.

Why is ICT backtesting different? +

ICT is discretionary. No formula for “identify the Judas.” Backtesting measures your analytical skill, not a system’s properties. Pattern recognition through repetition.

What are the benchmarks? +

AMD ≥80%. Judas prediction ≥65%. Valid entry ≥80%. Win rate at 1.5R+ ≥50%. All four consistently met across 50+ sessions.

How to avoid look-ahead bias? +

5 rules: write before advancing, mark before kill zone, reset if advanced too far, record results immediately, skip recognised sessions.

Conclusion — 100 sessions build what reading cannot

The difference between an ICT trader who understands all the concepts and one who trades them consistently profitably is almost always the number of structured, journaled backtesting sessions between them. Not talent. Not luck. Not a better system. Hours of deliberate, feedback-rich practice on historical data that built the pattern recognition live trading demands. Start Phase 1 tonight — open TradingView, navigate to a date six months ago, enter Replay mode, and start classifying AMD phases before each candle advance. Write it down. That is the first session.

The companion guides: Entry Model 1 is the London routine backtesting practises; daily bias is the pre-session analysis you’ll grade; stop loss and take profit cover the management framework you’ll execute in Phase 4. Or join the mentorship for structured review of your backtesting journal with direct feedback on AMD identification accuracy and replay trade annotations.

Lio

Founder & ICT trading educator, LiquiditySweeps.com

Lio has traded ICT and Smart Money Concepts on forex majors and US indices since 2021 and built LiquiditySweeps.com to teach the framework the way it should be learned: in sequence, on real charts, with free live tools instead of paid indicators.

About All articles Mentorship

Ready to go further?

The mentorship programme covers the full framework — weekly live sessions, daily bias review, and personal trade review on every entry you submit.

Quick check

What is the PRIMARY anti-look-ahead-bias rule in ICT backtesting?

✓ Correct. Write analysis BEFORE advancing the chart — never advance first and mark second. The act of writing before seeing creates genuine prediction. Advancing first and then marking what you would have seen is look-ahead bias — it produces inflated 70–80% win rates that collapse to 40–50% live.