Smart Money Concepts is the most popular technical framework of the last decade and it attracts the two least useful reactions: uncritical adoption, and blanket dismissal. Both are lazy. The framework contains one genuinely good idea, a set of mechanically definable constructs that can therefore be tested, and a narrative layer that is unfalsifiable and should be discarded.

Strip away the vocabulary and SMC says: large orders cannot be executed at a single price, so they must be worked, and working them leaves traces in the price record.

That is not mysticism. It is market microstructure. A fund that needs to buy a quantity equivalent to several days of average volume genuinely cannot lift the offer — it would move the price against itself catastrophically. It must accumulate patiently, into liquidity, which means it buys where sellers are. Sellers are found at prior lows, below obvious support, and wherever stop-loss orders cluster.

So “price returns to where large orders were worked” and “price seeks liquidity” are restatements of a real constraint. The framework's contribution is giving these mechanical definitions so they can be marked on a chart consistently — and consistency is worth a great deal even when the edge is modest.

Most descriptions give you half the definition, and the missing half is the half that matters. The full one has two parts, and both must hold:

• The candle. The last opposite-direction candle before an impulsive move. For a bullish order block, the last bearish candle before a strong rally; for a bearish one, the last bullish candle before a strong decline.
• The break. That impulsive move must close beyond the previous swing high or previous swing low on the same timeframe — a break-out or a break-down, as defined in chapter 3. Until that close happens, there is no order block. There is a red candle in a pullback, which is not the same thing and does not become the same thing because you drew a box round it.

This second condition is not a refinement or a filter to add later. It is what stops the construct being arbitrary. Every pullback in every chart contains a last opposite-direction candle — there is one before every move, including the moves that go nowhere. Marking them all gives you a chart covered in boxes and a rule with no selectivity at all. The break is the evidence that the move meant something structurally, and it is the only part of the definition that anything had to happen for.

Same timeframe is doing real work in that sentence too. A 15-minute impulse that breaks a 15-minute swing high is a 15-minute order block, and that is the timeframe it should be traded and tested on. Borrowing the break from a lower timeframe to justify a higher-timeframe box is the most common way this rule gets quietly broken, usually by someone who has already decided they want the trade.

What it actually is, stripped of the story: a recent local extreme immediately before a strong move. That is a real, mechanically-defined thing. Whether it has predictive power beyond ‘a recent extreme’ is exactly the question a control answers — and the right control is a randomly selected recent extreme at a similar distance, entered after a similar-sized pullback.

A practical refinement worth knowing: when the candle in question has a very long wick — say more than half its range — the body-based zone is tiny and the range-based zone is enormous, and neither is a good representation. The standard fix is to drop to a higher timeframe and find a cleaner block there. That is a sensible rule and it should be written down, not applied by feel, because applying it by feel is how a rule becomes unbacktestable.

Three consecutive candles where candle 1's high is below candle 3's low (bullish) or candle 1's low is above candle 3's high (bearish). The middle candle covered a price range that was never traded at both bid and ask — an imbalance.

This is the most attractive of the three constructs for research, for one reason: it requires no judgement at all. Three candles, two comparisons. Two people will always mark the same gaps. That makes it genuinely testable in a way most technical constructs are not.

The honest caveat: gaps do get filled, and so does almost everything else nearby. Prices oscillate. The test that means something is a distance-matched control — does price return to an FVG more often, or faster, than to a randomly chosen price band the same distance away and the same width? Run that and you will have a real answer, which is more than almost anyone teaching this has.

Price pushes just beyond an obvious cluster of stops — equal highs, a prior swing extreme, a round number — and then reverses sharply.

This is the construct with the best-founded mechanism, and it does not need any conspiracy. Resting stop orders are liquidity. A large buyer genuinely benefits from a brief move down that triggers sell stops, because those stops are the supply they need. They need not have caused it; they only need to be there when it happens. And a stop cluster above equal highs is visible to everyone, which is the coordination mechanism from the levels chapter operating again.

Mechanically: define the level (equal highs within a tolerance, or a prior swing high), define the sweep (a wick beyond it by more than x but a close back below), and define the confirmation. All codeable. All testable.

Here is the framework expressed the way it must be expressed to be evaluated — every term mechanical, every parameter named:

• Context. Higher-timeframe structure is bullish (rising swing highs and lows with lookback n), and price has entered a marked higher-timeframe demand zone.
• First confirmation. A bullish reversal pattern on the entry timeframe, with the candle definitions from chapter 9.
• Change of character. The first close above the most recent swing high on the entry timeframe after the confirmation — a break-out, on that timeframe, not a wick through it. That close is the event. Only once it has happened do you mark the last bearish candle before the impulse as the order block.
• Quality flag — Setup A. An FVG present inside that break — between the confirmation and the higher high — a binary you can test as a separate variable rather than assuming it helps.
• Retrace. Price returns into the marked order block.
• Entry. Either a limit at the block's midpoint, or a second confirmation pattern — whichever fires first. Both defined; both timestamped.
• Invalidation. A body close beyond the block. Not a wick. Define which.
• Stop. Beyond the block plus a volatility buffer.
• Target. The nearest opposing higher-timeframe zone, or a fixed multiple of the stop distance, whichever is closer.