Smart Money Concepts (SMC) Explained: BOS, CHoCH, Order Blocks and Real Tests

Smart Money Concepts (SMC) is a way of reading charts built on the belief that large players leave footprints in price structure: the highs and lows price makes, the breaks of that structure, and zones called order blocks. It has become hugely popular with newer traders. This article defines the core terms clearly, marks structure on a chart with rules a computer can follow, and then looks at what came out when we measured these ideas on gold.

The terms SMC borrows from ICT, such as liquidity sweeps and FVGs, are already explained in our ICT article, so we won’t repeat them here.

First: SMC definitions differ between teachers

There is no single original SMC textbook. Teachers and indicators disagree on key points, for example

• How many bars make a swing. Some use a 2-bar fractal on each side, some use 5 bars, some use ZigZag, and each gives a different set of points.
• What counts as a break. A close, a candle body, or just a wick poking through.
• Which candle is the order block. The last opposite-colour candle before the move, or the lowest candle in that stretch. The whole candle, or only its body.

Two people looking at the same chart can put a BOS in different places. Neither is wrong, but it means that before deciding whether SMC works, you need rules clear enough for a computer to mark the chart for you.

Market structure: HH, HL, LH, LL

A swing high is a bar whose high is above the bars around it. A swing low is a bar whose low is below the bars around it. Lined up in order, each point is named by comparing it with the previous point of the same kind.

• HH (Higher high) a swing high above the previous swing high
• HL (Higher low) a swing low above the previous swing low
• LH (Lower high) a swing high below the previous one
• LL (Lower low) a swing low below the previous one

By this definition an uptrend is HHs alternating with HLs, and a downtrend is LHs alternating with LLs. It is the same idea as the long-established Dow theory, not something new to SMC.

Some teachers split structure into two layers: swing structure, the big obvious points, and internal structure, the smaller points inside the same leg. The two layers often disagree. The internal layer says downtrend while the swing layer still says uptrend. If you use both, decide in advance which one drives your decisions, much like multi-timeframe analysis.

BOS and CHoCH

• BOS (Break of Structure) price closes beyond the latest swing in the direction of the trend. In an uptrend, a close above the swing high; in a downtrend, a close below the swing low. It is taken as evidence that the trend is continuing.
• CHoCH (Change of Character) price closes beyond a swing against the trend, for example closing below the latest higher low in an uptrend. It is treated as the first sign that structure may be changing. ICT calls a similar point an MSS.

Two things most courses don’t stress

1. Swings are always known late. If a swing high must be above the 3 bars after it, you can’t know it until those 3 bars exist. Lines marked on a past chart look as if they were known in advance, when nobody knew at the time. It is the same problem as indicators that repaint.
2. A CHoCH is not a reversal. It only means price broke the point you chose. Price can go straight back up to a new HH, and the label becomes a false signal.

Worked example: marking BOS and CHoCH with fixed rules

We generated a 64-bar sample price series and let a program mark it with these rules, with no eyeballing at all.

1. Swing high = a bar whose high is above the 3 bars before it and the 3 bars after it (swing low is the mirror image). It is confirmed when the third bar after it closes.
2. Break = a close beyond the latest confirmed swing that has not been broken yet. A wick through doesn’t count.
3. A break in the trend’s direction is a BOS. A break against it is a CHoCH, and the trend flips. The first break on the chart counts as a BOS and sets the starting direction.

The swings the program found

Bar	Type	Price	Compared with previous	Confirmed at bar
3	Swing low	99.78	first point	6
15	Swing high	106.36	first point	18
22	Swing low	103.20	above 99.78 = HL	25
34	Swing high	109.57	above 106.36 = HH	37
46	Swing low	99.85	below 103.20 = LL	49
49	Swing high	101.44	below 109.57 = LH	52

Event by event

• Bar 26 has a high of 106.70, already through 106.36, but closes at 106.35, 0.01 below the line. Under our rule it doesn’t count. With a wick rule, the BOS would be on this bar.
• Bar 27 closes at 107.38, above 106.36 = bullish BOS
• Bar 40 has a low of 103.07, below the HL at 103.20, but closes at 103.45. Not counted.
• Bar 41 closes at 102.53, below 103.20 = CHoCH, the trend flips to down
• Bar 52 closes at 99.71, below the LL at 99.85 that was only confirmed at bar 49 = bearish BOS

Look at bars 26 and 40. Just switching the rule from closes to wicks moves the signals. That is why the rules have to be written down before you look at the chart.

Order blocks, mitigation and breakers

• Order block (OB) in its most common definition is the last opposite-colour candle before a strong move. A bullish OB is the last down candle before the rally that made the BOS. The SMC story is that large players placed orders there and will add more when price returns.
• Mitigation is price coming back to touch the OB for the first time. Many traders use it as their entry. Some teachers use “mitigation block” for a different kind of failed zone, so check what the person means.
• Breaker block is an OB that price has closed through. Many teachers see its role as flipping, so a broken bullish OB becomes resistance. Some only count it as a breaker if a high or low was swept before the zone broke.

In the sample chart, the OB for the bullish BOS is bar 21 (open 103.60, close 103.57, high 103.79, low 103.49), the last down candle before the rally that broke out at bar 27. Price first came back into the zone at bar 40 (low 103.07). That is the mitigation. But the same bar closed at 103.45, below the zone’s 103.49 low, so the zone broke and became a breaker right away. After that, price never came back up to test it. The highest high after bar 41 is 102.87.

Premium and discount

SMC splits the latest price range in half at the 50% point, called equilibrium. The upper half is premium (expensive, look for sells) and the lower half is discount (cheap, look for buys). It is the same idea as waiting for a deep enough pullback, close to measuring Fibonacci levels.

In the example, the latest up range runs from the HL at 103.20 to the HH at 109.57.

• Equilibrium = (103.20 + 109.57) ÷ 2 = 106.385
• Bar 35 closes at 106.72, still in premium
• Bar 36 closes at 105.56, the first close in discount

In real time, though, the HH at bar 34 is only confirmed at bar 37, which closes at 104.69, already in discount. Someone using a “buy in discount during an uptrend” rule would get a buy signal at bar 37, but price kept falling until the CHoCH at bar 41. Discount doesn’t guarantee a bounce, and which range counts as the “latest range” can differ too.

What we have measured

Everything in this section is our own backtesting on gold (XAUUSD) M1. Past results don’t predict future results, and the definitions are the ones we wrote ourselves.

1. Order blocks bounce no more than decoy zones

Method: we found OBs under 7 definitions, varying how strong the move had to be (a body of at least 1.5, 2 or 3 times ATR 14), using either the body or the whole candle as the zone, and with or without a required FVG. For each OB we placed a decoy zone of the same size, on the same side, at the same time, but 3 zone-heights away. Then we counted how often, when price first came back to a zone, it bounced one zone-height before breaking one zone-height through. Real tick data, 2026.

Real OBs differed from decoy zones by only −4.6 to +2.5 percentage points. The definition with the most cases (3,846) scored 46.0% against 45.9% for its decoy. Both sit slightly below 50%, which fits the spread cost being paid. Extra conditions, such as a stronger move or a required FVG, didn’t separate OBs from decoys.

If zones placed arbitrarily nearby bounce just as often, the zone itself adds no information. Test every tool against a control like this before trusting it.

2. A full SMC EA: no RR survived

We wrote an EA to measure the commonly taught recipe: find a sweep, then the OB before the strong move, require an FVG next to the OB, place a limit order at the FVG edge, and put the Stop loss behind the sweep wick. Tested on 2026 real tick data (about 7.6 months), 0.01 lot per trade, 3,000-dollar starting balance.

At RR 1:2 it took 1,910 trades and won 27.75%, below the 33.3% break-even, for a net loss of 667 dollars and a Profit factor of 0.84. Changing RR from 1.5 to 6, no setting reached a Profit factor of 1.

The gap between the two lines narrows as RR rises but they never cross. If the signal had a real edge, at some RR the win rate should rise above the break-even line. A 4.6-year backtest on simulated data took 14,741 trades with a Profit factor of 0.75 and lost almost the entire account (99.81% drawdown). There was another symptom: about 251 trades a month, when SMC setups are supposed to be rare. The layers of conditions were filtering out almost nothing.

3. BOS as a filter made the total worse

We added BOS as a condition to a Fibonacci pullback system built on ZigZag swings, entering only on legs that had made an HH or LL beyond the previous point. Gold M1, 2026, real tick data, 0.10 lot per trade. The yardstick is net profit ÷ maximum drawdown: the higher, the more return per unit of pain.

Set	Trades	Profit factor	Net profit	Drawdown	Profit ÷ DD
All legs (no filter)	1,882	1.39	40,601	4,241	9.57
Only legs with a BOS	1,125	1.43	28,667	3,481	8.24
Only legs without a BOS	782	1.24	9,843	3,648	2.70

Legs with a BOS really were better (Profit factor 1.43 against 1.24), but legs without a BOS were still profitable. Cutting them gave up more profit than the drawdown it saved, so 9.57 fell to 8.24. The lesson: a better subset is not the same as a good filter. And this is one year, one market.

Using SMC with risk management

If you keep using SMC, use it as a language to describe structure, but manage money with numbers you can check.

• Place the Stop loss by structure, then size the lot from that distance. For a buy after a BOS with the Stop loss under the latest HL, the distance can be wide. Use position sizing from the Stop loss, not the same lot every trade.
• Know your break-even. At RR 1:3 you need to win more than 25%, and more than that once spread is included. See win rate vs break-even.
• Record signals before the outcome. Mark the BOS or OB when the bar closes, not afterwards once you have seen price run.
• Use a control. Compare against random entries or decoy zones, split results by year, and beware of trying many definitions and keeping the best one, the data snooping trap.

BOS labels on a past chart always look accurate, because they were drawn after the answer was known. The real question is what you knew when that bar closed.

Common mistakes

• Not fixing the definition first. Switch from closes to wicks, or from a 3-bar to a 5-bar fractal, and the signals move. If you choose the definition after seeing the result, every chart will seem to agree with you.
• Forgetting that swings are known late. The highs and lows you see on the chart now may not have been confirmed when you would have entered.
• Treating a CHoCH as a reversal. It only says one point broke. Price can go back the original way at any time.
• Believing the big-player story without measuring. The idea that large players hide orders in OBs sounds reasonable, but in our measurements OB locations were no better than arbitrarily placed zones.
• Stacking conditions and assuming they filter well. Sweep + OB + FVG sounds strict, yet the EA we tested still fired about 251 trades a month. Count what each layer actually removes.