Order Blocks: How to Define Them and Actually Test If They Work
Order blocks are one of the most popular ICT concepts. But most traders use them by feel, not by rule. Here's how to turn them into testable conditions.
Order blocks are everywhere in trading communities right now. The idea is straightforward: institutions leave footprints when they place large orders, and price tends to return to those zones. But there's a gap between understanding the concept and actually knowing whether it works in a systematic way.
What an order block is (precisely)
An order block is the last opposing candle before a significant move. If price makes a strong bullish push, the last bearish candle before that push is the bullish order block. When price returns to that zone, institutions are theoretically re-entering.
The key word is "significant move." That's where most traders get vague. "Significant" needs a definition if you want to test it:
- A move that breaks structure (a prior swing high for bullish, prior swing low for bearish)
- A move of at least X% from the order block candle's body
- A move that leaves behind a fair value gap
Without a precise definition, you can find an order block to explain any trade after the fact. That's hindsight bias dressed up as analysis.
The difference between confluence and noise
Traders stack order blocks with fair value gaps, break of structure, session timing, and higher-timeframe bias. Some of that improves the signal. Some of it is rationalization.
The only way to know which is which is to test each condition independently. Does the order block alone have any edge? Does adding the FVG requirement improve results or just reduce trade count?
Building this in the strategy builder
The ICT concepts block category in Tradelybox includes Order Block and Fair Value Gap detection. You can connect them:
- Order Block block → fires when price retraces into a defined order block zone
- AND gate → combine with a session filter (London or New York only)
- Entry block → long entry with stop below the order block low, take-profit at the next liquidity target
This gives you a codified, testable version of the concept. No subjectivity at entry time.
What the backtest will likely show you
Be prepared for one of three outcomes:
- The strategy has positive expectancy — win rate and R:R combine to show net positive P&L across a large sample. Proceed to forward testing.
- The strategy is marginally positive with a small sample — not enough trades to be statistically meaningful. You need to either expand the date range or loosen the entry criteria to get more data.
- The strategy loses money — the concept doesn't hold up as a mechanical rule. This is actually valuable information. Most retail traders never discover this because they never test.
What to do with a failed backtest
A losing backtest isn't a dead end. It tells you which condition is breaking down. Run the backtest without the session filter. Run it with just the FVG component. Run it on a different timeframe. Narrow down which variable is responsible for the losses.
This is strategy development. It's slow and it kills a lot of ideas — which is exactly the point.
The replay check
Before trusting a backtest result, use Replay Mode to step through some of the winning and losing trades manually. Numbers can obscure trades you would never have taken in practice — entries in the middle of news releases, stops that required 50-pip precision, exits you couldn't have executed at the reported price.
If the trades look good bar by bar, the backtest result is more trustworthy.
We build Tradelybox for traders who want to test ideas rigorously without learning to code.