Most traders learn the ICT order block from a chart that was picked after the move was over. On that kind of chart, the setup always looks perfect. So we ran the rules through the AFM Research Lab on 28 pairs and 9.9 years of Daily data.

I will show you the rules, the losses, and what each failure teaches you.

What Is An ICT Order Block?

Price does not move in a straight line. It pushes, pauses, and then pushes again. An order block is the idea that a strong push leaves a footprint you can trade later.

A bullish order block is the last down candle before a strong move up. A bearish order block is the last up candle before a strong move down. The high and low of that candle mark out a zone on your chart.

The thinking behind it goes like this. Big traders could not fill all of their orders inside that one candle, so some orders may still be waiting there. When price comes back to the zone, those orders may push price away again, and that return is the trade.

How Do Most Traders Trade It?

This is the version you will find in almost every tutorial. We tested it exactly as taught.

One term first. ATR stands for average true range. It tells you how far a pair usually moves in one candle. A move that is huge on one pair can be normal on another, so ATR lets you compare pairs fairly. Our guide to the ATR indicator covers it in full.

Textbook version checklist:

1. Use the Daily chart.
2. Look for a move away from a candle of at least 2 ATR. That candle is your order block.
3. Keep the block live for up to 50 bars, then drop it.
4. Plan to enter when price returns to the block.
5. Place your stop beyond the far side of the block, plus 0.2 ATR.
6. Set your target at 2R.
7. Use a time limit setting of 30 bars.
8. Risk 1 percent of your account on each trade.

What The Textbook Version Produced

Three more terms before the table, because you need them to read it.

R is the amount of money you risk on one trade. If your stop is hit, you lose about that amount. If price reaches a 2R target, you make about twice what you risked.

Expectancy is the average result of one trade, measured in R. If it is negative, the average trade loses money.

Profit factor is the money won divided by the money lost. Above 1 means the strategy made money. Below 1 means it lost.

The textbook version is not tradable as written. A max drawdown of 95.4 percent would wipe out almost any real account long before the ten years were up.

It also took 528.5 trades a year. That is the real problem underneath the number. Almost every small push counted as an order block, and most of those zones meant nothing.

What Did We Change And Why?

We made five changes. Each one had a reason.

Raising the move size did the heavy lifting. It cut the strategy from 528.5 trades a year down to 31.4.

The bigger target changed the shape of the results. The win rate fell from 35 percent to 27 percent, but the average win grew from 1.83R to 2.62R.

Did The Changes Work?

They cut the damage. They did not fix the problem.

Look at the profit factor row. It is 0.93 in both columns. Expectancy stayed negative too.

The edge did not get better. The filter only ran the same losing edge 311 times instead of 5,242 times, so the account bled slower. That is the finding.

The filtered version is still not tradable as written. Its 34.4 percent drawdown sits above the 30 percent line we use as a limit.

How We Ran The Test

We tested 28 forex pairs on the Daily chart, which was the only chart timeframe in this test. The textbook version took 5,242 trades. For the filtered version, we ran a grid search over 12 parameter combinations across the full test history, and the chosen settings took 311 trades.

We split the filtered trades into 4 walk-forward folds in time order. We also ran 1,000 Monte Carlo simulations that reshuffle the order of the trades. Results are internal to the AFM Research Lab and the dataset is not published.

The Settings Were Tuned On The Same Data

The filtered settings were picked because they scored best across the full history. That is in-sample optimization. The test already knew the answers, so live results are usually worse than what you see here. [EXTERNAL LINK 2 goes on "in-sample optimization", pointing to the peer reviewed work on backtest overfitting]

The Time Slices Did Not Agree

Two of the four folds had positive expectancy. Two were negative.

The strategy did not behave the same way in every period. That matters more than the average. If you only look at the overall number, you miss the fact that half the test window lost money.

There is a second catch. The settings were chosen using the full history, and that history includes all four folds. So these folds show whether the result was steady over time. They do not show how the strategy would do on data it had never seen.

The Drawdown Range Is Wide

Reshuffling the order of trades cannot change the compounded return, so every Monte Carlo run gave the same CAGR. That is how the math works, and it is not a sign of strength. The Monte Carlo only tells you about drawdowns.

The median max drawdown was 35.1 percent. The 5th percentile was 26.4 percent and the 95th percentile was 46.6 percent. The blow-up rate was 0.0 percent, but a drawdown that deep would push most traders to quit long before the account was empty.

Results By Pair Were Scattered

AUDCAD had 7 trades with an expectancy of 1.09R. EURUSD had 6 trades with a win rate of 0 percent.

Samples that small tell you very little. Picking the best pairs after the fact would just be more in-sample fitting.

What Does One Trade Look Like?

This is a real trade from the tested data. It used the filtered rules on the Daily chart.

Here is how the rules played out. Price pushed up and left a block behind. Then it came back down and touched that block.

That touch was the entry. The stop went just below the block, 49 pips away. The target sat 4R above the entry, 196 pips away. Price reached it in 6 bars.

One thing to keep in mind. This was the largest win in the whole filtered sample. Most trades did not look like this, which is why the overall result still lost money.

A Loser From The Same Test

A winner like that is what gets shared online. Here is a loser from the same test.

It was a USDZAR long on the Daily chart, with entry on July 17, 2016 at 14.40326 and a stop at 14.32525. The stop was hit on the entry date for -1.05R, and the trade was held 0 bars.

All five of the worst trades in the sample were on USDZAR. A clean-looking zone does not stop price from running straight through it.

When Should You Take It And When Should You Walk Away?

Both versions lost money, so taking the trade here means taking it on a demo account or in your journal. These lines use the filtered rules.

How Should You Study This Setup?

I would not trade these rules with real money. The test says they lose. But the setup is still worth studying, so here is how I would do it.

1. Open a Daily chart of the pair you want to study.
2. Add the ATR indicator. You need it to measure how big a move is.
3. Look for a strong move away from a candle. It has to be at least 4 ATR.
4. Find the last opposite candle before that move. That candle is your zone.
5. Check how old the zone is. If it is older than 20 bars, skip it.
6. Wait for price to come back to the zone. That is your entry.
7. Put your stop 0.1 ATR past the far side of the zone.
8. Put your target at 4R. Close the trade after 10 bars if nothing happens.
9. Risk no more than 1 percent of your account on the trade.
10. Write down every result. Log the losers too, not only the winners.

Also Read: How To Use The ATR Indicator In Forex Trading