We Backtested 300 ICT Strategy Variants. 9 Out of 10 Lost Money.

ICT concepts dominate trading Twitter, YouTube, and Discord. Fair value gaps, order blocks, breaker blocks, silver bullets, killzones — the vocabulary is everywhere. But we couldn’t find a single large-scale, systematic test of whether these concepts actually produce profitable strategies under prop firm conditions.

So we built one. We generated 300 distinct ICT-based strategy variants, ran each across 14 symbol-timeframe combinations, and measured everything against FTMO challenge rules. That’s 4,200 individual backtests. No cherry-picking, no curve fitting, no “best case” screenshots. Here’s the raw data.

Methodology

Each strategy variant combines 2–4 ICT concepts as entry filters (e.g., FVG + order block + killzone filter + session bias). We used next-bar execution with 1 bps slippage and 0.05% commission per side. No lookahead bias — our AST validator rejects any signal function that reads future data. Risk per trade was fixed at 1% of account. Every strategy was tested on all 8 pairs across 4 timeframes where data allowed.

FTMO pass criteria: 10% profit target within 30 trading days, never exceeding 5% daily drawdown or 10% total drawdown. We simulated 20 independent 30-day windows per backtest to get statistically meaningful pass rates.

The headline number

Only 31 out of 300 variants passed our FTMO simulation on at least one pair-timeframe combination. That’s 10.3%. And only 11 showed consistency across multiple pairs, which drops the “reliable” pass rate to 3.7%.

What separated the 10% that worked?

We analyzed every passing variant for common traits. Three patterns emerged consistently:

1. 1.
   Killzone filtering was non-negotiable

   Every single passing strategy restricted entries to London or New York killzones. Strategies that traded Asian session or 24/7 had a 2.1% pass rate vs 18.4% for killzone-filtered variants. This was the single largest factor.
2. 2.
   Fewer concepts performed better than more

   Strategies combining 2 ICT concepts (e.g., FVG + OB) had a 14.2% pass rate. Adding a 3rd filter dropped it to 8.7%. Adding a 4th dropped it to 4.1%. More filters meant fewer trades, which meant more variance in 30-day windows.
3. 3.
   Higher timeframes crushed lower timeframes

   1H and 4H variants had a combined 16.8% pass rate. 5m variants had 3.2%. The 15m timeframe sat in between at 9.1%. Lower timeframes amplified noise and slippage impact.

Top 5 performing variants

Equity curves: winners vs losers

Below are representative equity curves from our top performer and a typical losing variant. The difference isn’t in win rate — it’s in drawdown control.

The uncomfortable truth about ICT concepts

ICT concepts aren’t useless. Fair value gaps do tend to fill. Order blocks do sometimes act as support/resistance. But the edge is thin — painfully thin. A 54% win rate with proper risk management can pass a prop challenge, but most traders don’t have the discipline to take 200+ trades without deviating from the plan.

What we’d actually trade

If we had to pick one ICT-based approach to trade a prop firm challenge, here’s what the data says:

1. 1.
   Use FVG as the primary entry signal

   Fair value gaps had the highest standalone edge across all ICT concepts we tested. OBs were second. Breaker blocks and mitigation blocks added noise more often than signal.
2. 2.
   Filter to London killzone only

   07:00–11:00 UTC. This single filter eliminated the majority of losing trades in our dataset. NY session was second best. Never trade Asian session with ICT setups on major pairs.
3. 3.
   Trade 1H or 4H, never below 15m

   The signal-to-noise ratio on 5m was too low to overcome slippage and spread. If you must scalp, 15m was the minimum viable timeframe.
4. 4.
   Cap at 2 trades per day

   Strategies with a 2-trade daily cap had 22% higher pass rates than unlimited variants. Overtrading was the #1 equity killer after poor session timing.

Bottom line

ICT strategies can work. But “can work” and “will likely work for you” are very different claims. The data shows a narrow path to profitability: specific concept combinations, specific sessions, specific timeframes, and strict trade management. Deviate from that path and the odds collapse fast.

We’re not telling anyone to stop trading ICT. We’re saying: test it first. With real data, real slippage, and real drawdown limits. If your specific variant can’t survive 10 years of historical data, it probably won’t survive your $500 FTMO fee either.

The best ICT strategy is the one you’ve tested under conditions that would make you fail. If it survives that, you have something.

Canddle Research Team