What is an order block?
An order block is commonly defined as the last down candle before a strong up move, or the last up candle before a strong down move, on the theory that a large participant accumulated a position there before pushing price away.
The narrative is about institutional accumulation: a participant with size cannot fill in one print, so they build a position quietly in an area and then the market moves away from it. When price returns to that area, the remaining unfilled interest supplies a reaction.
The narrative is not the testable part. Nobody outside the participant can see the order book history that would confirm accumulation happened, and inferring it from candle shape is a story about a mechanism rather than an observation of one. What is testable is much narrower and much more useful: does price behave differently when it returns to areas selected by this rule than to areas selected at random?
That reframing is what makes the concept researchable. You do not need the institutional story to be true. You need the rule to select areas that behave differently, and that is a measurement.
Why are order blocks hard to define mechanically?
Order blocks are hard to define because two of the three ingredients — which candle counts as the last opposing one, and what counts as an impulsive move — are usually left to judgement, and judgement applied to a chart whose outcome is visible is not a rule.
The impulse condition is the worst offender. 'Strong move' has to become something like 'displacement of at least N times ATR within M bars', and the choice of N and M changes the set of qualifying blocks enormously. Charts in educational material are annotated with the impulse already visible, which makes the selection look obvious in a way it never is in real time.
The candle selection is nearly as loose. Last opposing candle by body, by close, by wick, or the last candle whose body sits entirely within the eventual block zone — each version produces a different zone, and zones that differ by a few pips produce very different fill rates on the retest.
None of this makes the concept worthless. It means that any claim about order blocks working is a claim about one particular set of definitions, and that the person making the claim usually has not stated which one.
How do you test an order block strategy?
Test it by fixing every definitional threshold in advance, generating the blocks as the test walks forward, and comparing the resulting reaction rates against blocks placed at random locations in the same series.
The random-location comparison is the part most order-block analysis skips, and it is the part that decides everything. Price reverses somewhere reasonably often. A method that marks a zone and then notes that price reacted there needs to beat the base rate of price reacting at an arbitrary zone, or it has discovered that markets oscillate.
Grade three outcomes, not two: the block held, the block failed, or price never returned within the window. Dropping the never-returned cases inflates the apparent hit rate, and dropping the failures does so even more obviously.
Then check the sample size against the number of definitional variants tried. Order block research has an unusually high variant count — impulse threshold, candle selection rule, zone boundaries, retest window, entry trigger — and it is easy to run fifty combinations without noticing that fifty is a lot of trials.
Are smart money concepts different from classical technical analysis?
Structurally, most smart money concepts are renamed versions of ideas that have existed in technical analysis for decades, presented with a different narrative about who is on the other side.
A break of structure is a higher high or lower low. A liquidity sweep is a false break or a spring. An order block is a supply or demand zone. A fair value gap is an imbalance, which older material called a runaway gap or simply a fast move. The observation is not a criticism — the underlying features are real chart events either way — but it does mean the ideas inherit whatever evidence the older versions had, which is thinner than most people assume.
What is genuinely new is the framing: an explicit story about institutional intent behind each pattern. Stories help people remember rules and are worth exactly nothing as evidence. The useful discipline is to keep the rule and discard the narrative, then measure the rule.
Where the vocabulary earns its place is in precision about sequence. Smart money material is often stricter than classical material about the order in which things must happen — sweep, then structure break, then retrace into the zone — and a sequence is easier to encode mechanically than a vibe about a chart. That strictness is genuinely testable.
What should you do if your order block backtest looks great?
Assume it is fitted until an out-of-sample run says otherwise, because a great in-sample order block result is the expected outcome of a search with this many free parameters.
Run the fixed definition once on a period you have never inspected. Not repeatedly with adjustments, once. If the result survives, you have something worth carrying forward; if it collapses, you have learned that quickly and cheaply, which is what a backtest is for.
Then check the execution assumptions, because order block entries are typically limit orders at a zone edge with a stop just beyond it. Tight stops make same-bar ambiguity decisive, and a tester that resolves ambiguous bars favourably will make any zone-based strategy look excellent.
Finally, look at the trade distribution rather than the total. A strategy whose entire result comes from three enormous winners in one quarter is a description of that quarter. The metric that matters is whether the edge shows up consistently across periods and instruments, not whether the equity curve ends up.