Smart Money Concepts (SMC) is a retail price-action framework that tries to read a chart the way large institutions supposedly trade it — buying and selling around zones of "liquidity" rather than off classic indicators. It is a vocabulary and a set of rules for interpreting structure, not a proven edge.
The framework is popular, internally consistent, and heavily discretionary. That last word is the whole story for anyone who wants to test it, so we will get there honestly rather than sell it.
Where SMC comes from
SMC is a repackaging of older price-action and order-flow ideas, systematised in the 2010s by Michael Huddleston, known online as the Inner Circle Trader (ICT). Huddleston began publishing on forums and later YouTube, framing the market around how banks and institutions execute orders, and the community distilled that into the "SMC" label (FTMO: How to trade Smart Money Concepts, ForexTester: ICT trading explained).
The underlying premise is "smart money": capital controlled by institutional investors, funds, and central banks, believed to have a greater chance of success than retail flow (Corporate Finance Institute: Smart Money). Even that mainstream premise comes with a caveat — FINRA warns that "not every institutional investor can outperform the market, so blindly following the smart money can potentially leave you facing losses" (FINRA: Institutional Investors and 'Smart Money'). SMC then assumes those footprints are both real and readable on a candlestick chart. That is an assumption, not a measurement.
Market structure: BOS and CHoCH
SMC starts by labelling trend using swing highs and lows. Two terms do most of the work.
- Break of Structure (BOS) — price breaks a prior swing high (or low) in the direction of the trend, read as continuation.
- Change of Character (CHoCH) — price breaks structure against the prevailing trend, read as an early reversal signal.
Both are defined against previous highs and lows: a BOS confirms the trend is intact, a CHoCH breaks it the other way (FTMO: How to trade Smart Money Concepts, BabyPips forum: Market structure concepts). The catch is already visible: which highs and lows count as "significant" swings is a judgement call, and two traders can label the same chart differently.
Order blocks, fair value gaps, and imbalance
On top of structure, SMC marks specific zones where price is expected to react.
An order block is the last opposite-direction candle before a strong move — a zone "believed to contain unfilled institutional buy or sell orders" that price may revisit and pivot from (Alchemy Markets: Order Block explained, FTMO: How to trade Smart Money Concepts).
A fair value gap (FVG), or imbalance, is a three-candle pattern where a large middle candle moves so fast that the first candle's high and the third candle's low do not overlap, leaving a gap "considered a price inefficiency because price moved too quickly for two-sided trading to occur" (FXOpen: Fair value gaps vs liquidity voids, FTMO: How to trade Smart Money Concepts). SMC treats that gap as a magnet: price often returns to "fill" it.
Liquidity: the engine of the whole idea
The concept that ties SMC together is liquidity — pools of resting orders the framework assumes big players hunt.
- Buy-side liquidity (BSL) sits above swing highs and resistance, where short-sellers' stop-losses cluster.
- Sell-side liquidity (SSL) sits below swing lows and support, where long-holders' stops cluster.
A liquidity sweep (or stop hunt) is when price spikes through one of those pools, triggers the stops, then sharply reverses (XS: Buy-side and sell-side liquidity, FTMO: How to trade Smart Money Concepts). It is a genuinely useful description of a common false-breakout behaviour. Whether it reliably predicts the reversal is the untested part.
Premium, discount, and equilibrium
Finally, SMC frames "value." Take a defined range, mark its midpoint as equilibrium (the 50% level); everything above is the premium zone (expensive), everything below is the discount zone (cheap). The rule of thumb is that smart money buys in discount and sells in premium (Daily Price Action: Premium and discount strategy, ForexTester: ICT trading explained). It is Fibonacci-style value language wrapped around the liquidity story.
The honest part: is there evidence it works?
This is where a studio built on testability has to be blunt. SMC is a discretionary framework, and its published evidence is anecdotal rather than statistical.
The core problem is subjectivity. Which swing is "significant," which order block is valid, how much displacement is "enough" — none of these are fixed. As one ICT explainer notes, the setups "require careful consideration of both market context and multiple timeframes," and results "depend on individual interpretation and decision-making," which makes the method hard to program or automate (ForexTester: ICT trading explained). Because two analysts can mark different signals on the same history, the framework resists the statistical validation that would establish an edge.
None of this means the concepts are worthless. Liquidity pools above obvious highs are real; fast one-directional candles do leave gaps. The issue is the leap from "this pattern exists" to "this pattern is a tradeable edge."
What a systematic trader would have to pin down
To even backtest SMC, you have to strip out the discretion and replace every fuzzy term with a hard, computable rule. That translation is the entire exercise — and each choice changes the results.
| SMC concept | The discretionary idea | What a backtester must hard-code |
|---|---|---|
| Swing / structure | "A significant high or low" | Exact fractal or pivot lookback (e.g. N bars either side) |
| BOS / CHoCH | "Structure broke" | Close beyond the swing? By how many pips or ATRs? |
| Order block | "The last candle before the move" | Which candle, which price bounds, how long it stays valid |
| Fair value gap | "An imbalance" | Precise 3-candle non-overlap test and a fill threshold |
| Liquidity sweep | "Stops got hunted" | Wick-through distance, and the reversal window that confirms it |
| Premium / discount | "Cheap vs expensive" | The range definition and the exact equilibrium level |
Written as code, an FVG stops being a vibe and becomes a testable rule:
bullish FVG at bar i:
high[i-1] < low[i+1] # no overlap = gap exists
gap_size = low[i+1] - high[i-1]
valid if gap_size >= k * ATR # your displacement filter
entry when price returns into [high[i-1], low[i+1]]The moment you commit to those constants, you can measure it — and you inherit every hazard in why your backtest lies, from curve-fitting the thresholds to modelling costs honestly. Any version that survives still has to clear an out-of-sample test on data it never saw during design, as we describe in how we test cost, parity and robustness. Two objective SMC rule sets can post wildly different results on the same history — proof that the discretionary version was never really one strategy at all.
Frequently asked
Is SMC the same as ICT?
Effectively yes, with a nuance. ICT is Michael Huddleston's specific body of teaching; SMC is the broader community label for the same structure/liquidity/order-block toolkit that grew out of it.
Can Smart Money Concepts be backtested?
Only after you replace every subjective term with a fixed, computable definition. The discretionary version cannot be tested, because different traders mark different signals on identical charts. Once objectified, it becomes testable — but it is then your rule set, not "SMC" in general.
Does SMC actually work?
There is no robust public evidence that SMC produces a statistical edge, and the "smart money" premise it rests on does not guarantee outperformance. Individual traders report success, but anecdotes and selection bias are not the same as a validated edge.
Are order blocks and fair value gaps real?
The chart features are real and easy to spot after the fact — a fast candle does leave an imbalance. What is unproven is that price returns to them reliably enough to trade profitably after costs.
The stubborn takeaway
Smart Money Concepts gives you a rich language for describing what a chart already did. Whether it predicts what a chart will do next is a question the framework cannot answer for you — only a hard-coded, out-of-sample test can. Until you have written the rules down precisely enough to be wrong, you do not have a strategy. You have a story.