How to Identify Order Blocks (Without Fooling Yourself).

Ask five traders to mark the order blocks on the same chart.

You'll get five different charts back.

Same candles. Same definition, supposedly. Five different sets of rectangles.

Now ask those five traders for their rule — the written procedure that produced the rectangles.

Silence. There isn't one. There's a feel.

If your identification method can't be written down, it isn't a method. It's a mood with a highlighter.

Every guide on this keyword teaches order-block identification the same way: a labeled screenshot, the phrase "last opposing candle before the impulsive move," and a wish of good luck. This piece does the unglamorous thing instead — it shows how to identify order blocks as a written rule: step by step, precisely enough that two people applying it mark the same zones. Then it names the four ways identification quietly goes wrong, and gives you a twenty-minute test to find out whether what you have is a procedure or a vibe.

Two requests before we start:

1. Save this next to your platform, and mark your next twenty zones against it.
2. Send it to one trader whose "order block" moves every time they re-look at the chart.

Skip this if you just want the rule

Sixty-second version — the identification rule, compressed: (1) work on ONE timeframe per marking pass, chosen before you look; (2) find a displacement leg — a move whose candles are large, one-directional, and travel beyond the local structure that contained them (write your own threshold: e.g. "body larger than the prior N bodies, closing beyond the last swing"); (3) walk back from the displacement to the last opposing candle or consecutive opposing cluster — the final down-candle(s) before an up-leg, the final up-candle(s) before a down-leg; (4) draw the zone using ONE bounds convention you never change (wick-to-wick is the defensible default; body-only if you prefer tighter zones — the choice matters less than never switching); (5) mark it valid only if untested — price hasn't traded back through it yet. That produces candidates, mechanically, the same ones every time. Whether any candidate deserves money is a different question — grading — and confusing the two questions is how most order-block trading dies.

I — The rule, written down

The definition everyone quotes — "the last opposing candle before displacement" — hides three decisions inside it. A usable rule makes all three explicit.

Step 1 — Declare the timeframe. Before looking at the chart, not after. Zones exist per-timeframe; a rule that floats between M15 and H4 mid-pass isn't one rule, it's four.

Step 2 — Define displacement, in writing. "Impulsive move" is a feeling until you give it a threshold. Any honest version works: a candle body larger than the previous several bodies; a close beyond the last swing high/low; a leg that leaves a fair value gap behind it. Which threshold you pick matters less than that it's written and fixed — a displacement definition you can adjust per-trade will always find displacement wherever you already wanted a zone.

Step 3 — Walk back to the last opposition. From the displacement leg's origin, step backward to the final candle — or unbroken run of candles — that closed against the leg's direction. That candle's range is your raw zone. If there's no opposing candle (rare: displacement from a fresh gap), there is no order block; write that down too, because a rule needs cases where it says no.

Step 4 — Apply your bounds convention. Section II. One decision, made once.

Step 5 — Check freshness. If price has already traded back through the range since the displacement, the candidate is spent — mark it dead or don't mark it at all. Untested is a property of identification, not just grading, because a rule that keeps resurrecting used zones will paper the chart with dead levels.

That's the whole procedure. Note what it doesn't contain: no volume folklore, no "institutional intent," no narrative. Identification is geometry plus bookkeeping. The rule's job isn't to be clever. Its job is to give the same answer twice.

II — The bounds decision

Wick-to-wick or body-to-body — the internet fights about this endlessly, and the fight misses the point.

Wick-to-wick (the candle's full high-to-low) is the defensible default: the wicks are prices where business actually happened, and the wider zone is more forgiving about where a retest "counts." Costs: wider stops if you later trade it, more overlap between neighboring zones.

Body-only (open-to-close) draws tighter zones with cleaner invalidation — and misses retests that tag the wick and leave.

Here's the actual rule: the convention you pick matters far less than never switching. A trader who uses wicks on winners and bodies on losers — retroactively, of course — has built an unfalsifiable system where every zone "worked." Pick one. Write it in the same note as your displacement threshold. Change it only on the first of a month, for all future zones at once, never for the trade you're currently in.

III — The four ways identification lies to you

Hindsight marking. You see the move, then find its origin. That's not identification, that's autobiography — we dissected this survivorship trick in the smart-money piece. The fix is procedural: mark zones only at the right edge of the chart, or in bar-replay, never on a fully-formed page.

Cluster ambiguity. Three small opposing candles before the leg — is the zone the last one, or all three? Your rule must have a tie-break written down (ours above: the unbroken opposing run counts as one unit). Any answer works; improvising per-chart doesn't.

Zone drift. You marked it Tuesday. On Thursday's re-look it's "really" a bit lower. Nothing changed except your mood — drift is the mood leaking in. A marked zone is immutable until price kills it; if you keep wanting to move zones, your Step-2 threshold is too loose.

Timeframe soup. An H4 zone "confirmed" by an M5 candle, invalidated on M15, re-validated on H1. Each timeframe is a separate rule-pass with its own zones; mixing them mid-decision lets you find whatever you're hoping for somewhere.

IV — The two-trader test

The falsifiability standard from the smart-money piece applies to your marking, mechanically: could another trader, given only your written rule and the same chart, produce the same rectangles?

Run it literally if you can — hand a trading friend your rule and a clean chart, mark separately, compare. No friend, no problem: run it against yourself. Mark today, save the chart, re-mark from scratch in three days without peeking, compare. Every disagreement between you-today and you-Thursday is a hole in the rule — a decision it silently left to mood. Patch the rule, not the zones.

A rule two people can follow is a method. A rule only you can follow — differently each week — is a horoscope.

V — Identification is not permission

The most expensive confusion on this keyword: treating "I found an order block" as "I have a trade." Identification, done honestly, produces candidates — many of them — because a mechanical rule fires on every qualifying pattern. Whether a candidate deserves money is the grading question — a separate five-factor discipline with its own article, and its own most-important output: stand aside when nothing qualifies.

Identify mechanically. Grade skeptically. Trade rarely. The funnel narrows at every stage, and it's supposed to.

The 20-minute audit — measure your own consistency

Minutes 0–5 — Write the rule. All five steps, your thresholds, your bounds convention, your tie-break. If you can't write it in five minutes, you didn't have one.

Minutes 5–15 — Replay twenty marks. Bar-replay a fresh session. Mark every zone your written rule produces at the right edge — no scrolling ahead. Twenty zones or twenty minutes, whichever comes first.

Minutes 15–20 — Score the agreement. Compare against the zones you would have marked freehand (or your marks from a previous pass). Count matches. Anything under near-total agreement means the rule on paper and the rule in your head are different rules — and the one in your head is the one that will spend your money.

Where the SMC toolkit sits

The honest ledger, caveats first: a mechanical rule — ours or yours — finds candidates; it cannot promise any of them will hold, and no scoring layer changes that. What our SMC AI-Scored Toolkit is, against this article's standard: identification as code — the same written-down-rule discipline §I demands, executed the same way on every chart, with zones that commit on the closed bar and don't repaint (so the marks you review are the marks you'd have traded). It then does the part this article deliberately doesn't cover — scoring every candidate 0–100 on five transparent factors, dropping the weak ones, and saying STAND-ASIDE in plain English when nothing qualifies — with its Resolved % printed on your chart, misses included. It scores; it never predicts. And if you'd rather build your own rule first, §I is complete without buying anything — that's the point of writing it down.

Disclosure

This is education about a procedure, not a promise about outcomes. A written identification rule makes your marking consistent; consistency makes results measurable; neither makes results profitable — that depends on everything downstream, starting with grading and sizing. The one question worth asking anyone teaching order blocks, us included: "Show me your rule, written, with its thresholds — and show me a chart where it finds nothing." A rule that never returns no order block here isn't a rule — it's a rubber stamp.

Your first 20 minutes

Write your rule tonight — the five steps, your numbers, your convention. Then run the replay audit above once. Whatever your agreement score is, you now know something almost nobody trading order blocks knows: whether your method exists outside your head. If it does, next stop is grading the candidates. If it doesn't, you just saved yourself every trade the mood would have taken.

Five traders, five charts.

One rule, one chart.

That's the entire difference.

Write the rule down — or admit the highlighter is in charge.