ICT and Smart Money Concepts traders need a tool that speaks their language: fair value gaps, liquidity sweeps, market structure shifts, order blocks, and the killzone timing that ties them together. Generic backtesters can’t express these. "Best ICT backtesting software" means replay you can mark up by session, plus ideally a way to test a sweep/FVG setup mechanically.
Honest breakdown — and a heads-up: when we tested the strict ICT setups mechanically, they barely fired and lost after costs, which is itself the most useful thing to know. Disclosure: Secuora is ours, including an AI backtester that compiles plain-English ICT strategies into a deterministic engine.
What actually matters
- Session/killzone replay — DST-correct jumps to the 10–11 NY and other windows, marked up bar by bar.
- ICT primitives — FVG detection, liquidity sweeps, MSS/BOS, order blocks, displacement.
- Fibonacci + drawing tools — for OTE, the golden pocket, and gap markup.
- A way to test mechanically — turn "sweep then MSS then FVG entry" into a backtest, not just a screenshot.
- Honesty about what’s discretionary vs. codeable.
The honest shortlist
AI backtester with FVG/sweep/MSS/displacement/order-block primitives + killzone replay; honest that the strict confluence rarely fires.
Popular in the ICT community, TradingView charting, real CME/forex data — journal covers sim trades only.
Unbeatable drawing tools for FVG/OTE markup, free bar replay — but no mechanical ICT backtesting or stats.
Where Secuora fits
Secuora’s AI backtester lets you describe an ICT setup in plain English ("between 10–11 NY, sweep a swing low, then MSS, then enter the FVG at 50%, stop below the swept low, 2R") and it compiles that into exact, tested rules and runs it on real data. The AI writes the rules; a deterministic engine computes the result, so it can’t flatter your idea.
It ships ~27 primitives covering FVG retrace, liquidity sweeps, MSS/BOS, displacement, order blocks, opening ranges and more — the ICT/SMC vocabulary, not just generic indicators.
The honest part: we ran the strict Silver Bullet on crypto AND real NQ/ES — it fired 0–3 times a year, and the looser readings lost after costs. we backtested 60+ strategies across 6 markets — over 80,000 simulated trades, fees on — and published every result (none were profitable after costs). The mechanical floor is nearly empty, which means the ICT edge is in the discretionary read — worth knowing before you "backtest" it.
What you can and can’t backtest in ICT
The detectable parts of ICT are real and codeable: a fair value gap is a precise 3-candle imbalance, a liquidity sweep is a wick through a swing that closes back inside, an MSS is a close beyond a confirmed swing. What’s NOT codeable is the sequencing and selection — which sweep counts, whether displacement is "aggressive enough," which FVG is valid. That’s the discretionary core, and it’s exactly why the strict mechanical Silver Bullet almost never triggers. Use a backtester to pressure-test the codeable parts and your own sample to validate the discretionary read.
Frequently asked questions
Can you backtest the ICT Silver Bullet?
The codeable parts, yes: sweep, MSS and FVG are all precisely definable. But the strict same-candle confluence almost never fires — in our tests it triggered 0–3 times per year across crypto, gold, EUR and real NQ/ES futures. You can test the looser readings mechanically (they lost after costs), but the canonical Silver Bullet is mostly discretionary, so only your own forward sample truly validates it.
What is the best tool to backtest FVG and liquidity sweep strategies?
You need a tool that detects those patterns precisely. Secuora’s AI backtester has FVG-retrace, liquidity-sweep, MSS/BOS, displacement and order-block primitives, so you can describe an FVG or sweep strategy in plain English and run it on real data. FX Replay and TradingView are great for marking them up by hand in replay.
Does ICT actually work?
The components (FVG, sweeps, structure) are real, observable price phenomena. But our mechanical tests across six markets show the strict setups barely fire and the looser ones lose after costs — so any edge lives in the discretionary read, not the rules. That’s not a dunk on ICT; it means you can’t verify it from someone else’s stats, only from your own tested sample.