Order Blocks on ES and NQ Futures: The ICT Concept Explained With Worked Numbers

Most traders who learn order blocks start marking them everywhere, then find that half fail and have no rule for which half. The concept is simple, and the difficulty is in defining it tightly enough to test. An order block, in ICT methodology, is the last opposing candle or small group of candles before an impulsive move. You watch the zone it covers as a place price might react when it comes back.

The idea and what it assumes

ICT teachers explain it this way: a large participant can't fill a big order at once without moving price, so it builds the position over time near a certain level, and may defend that area again on a return. That's a story about intent that I can't verify. The testable part is the shape: an opposing candle, an impulsive move away from it, a return to the zone and a reaction or failure. ICT teachers also say a zone weakens as it gets retested, since the original positions have supposedly been absorbed. Treat that as a hypothesis to check in your own data.

Marking a bullish order block

1. Find an impulsive move that breaks recent structure. On the 15 minute chart, the source material used 10 points on ES or 40 points on NQ as a rough filter. Treat these as parameters to test. They're about $500 and $800 of move per contract, so they aren't equivalent, and a volatile day may need larger numbers.
2. Take the last bearish candle or two before the move began. Its high and low define the zone.
3. Wait for price to return to the zone. The bearish version is the mirror image: the last bullish candle before a sharp drop.

Compared with a horizontal support line, the block gives you a zone with a defined top and bottom, which makes stop placement more concrete. That's a practical advantage, not proof that zones predict better.

A hypothetical long on ES

Suppose price rises 14 points through recent structure, and the last bearish candle before the rise spans 4,985 to 4,990. Later price returns to the zone.

With about $29 of friction on ES (a $4 round-turn commission plus a tick each way), the reward nets to $971 and a stopped trade costs closer to $429. That's about 2.3 to 1 after costs. The stop goes beyond the zone because a close through it means the setup failed. A stop order can slip in a fast market, and entering at the zone edge with a limit order may not fill at all if price turns early. Both are part of the real result.

How it fails

• The marked zone wasn't an order block. If the move before it wasn't impulsive, you've marked an ordinary consolidation.
• The zone is retested several times. By the ICT view, repeated tests weaken it, and price eventually trades through.
• The trade runs against the higher timeframe direction. Whether that matters for your results is something to measure by tagging each trade with the 1 hour and 4 hour direction and comparing.
• Holding through a broken zone. If price closes clearly through the opposite side of the block, the setup is done. Staying in and hoping turns a defined loss into an undefined one. A failed block can also be re-marked as a breaker block, covered in a separate guide.

Using KPL levels alongside

KPL levels are fixed reference prices anchored to the opening print of the regular 9:30 AM ET session. The standard ES and NQ map has four levels, built from full-session and after-hours travel, and they don't move during the day. They aren't predictions, and they aren't known before the open. If an order block sits within a few points of a KPL level, note it as a tag in your log. I wouldn't call the overlap better until your own results say so. The KPL strategy page explains how the levels are built, and the free edge calculator is there if you want to run expectancy math on your own logged trades.

Testing it

Log every block you mark, taken or skipped: impulse size, zone width, whether price returned, whether the zone held, result in points and net dollars. Change one parameter at a time, such as the impulse threshold, and compare on separate data. A sample of 30 trades is not a statistically proven edge, 50 SIM trades is not live proof, and a simulated fill doesn't test queue position or slippage. Keep the written definition fixed while you test, because a definition that shifts to fit each chart can't be tested at all.