BOS and CHoCH are labels for two different breaks in market structure. A Break of Structure (BOS) usually extends the prevailing swing sequence; a Change of Character (CHoCH) breaks against it and warns that the sequence may be changing.
That distinction is useful. It is not objective by default, and a CHoCH is not proof of a reversal. The label only becomes testable after the trader defines which swings count, what constitutes a break, and which timeframe governs the decision.
What BOS and CHoCH actually mean
Market structure is the sequence of swing highs and swing lows visible at a chosen resolution. Classical technical analysis describes an uptrend as higher highs and higher lows, and a downtrend as lower highs and lower lows. Both Charles Schwab's technical-analysis primer and FXOpen's market-structure guide use that same basic sequence.
SMC adds labels to the breaks:
| Label | What breaks | Usual reading | What it does not prove |
|---|---|---|---|
| BOS | A relevant swing in the direction of the existing sequence | Continuation remains possible | That the next trade has positive expectancy |
| CHoCH | A relevant swing against the existing sequence | The old trend may be weakening or changing | That a new trend is confirmed |
This is the common practitioner convention described by FXOpen and Alchemy Markets: BOS is a continuation label, while CHoCH is an early counter-trend break. Alchemy also makes the important distinction that a later BOS in the new direction supplies more confirmation than the first CHoCH alone.
The sequence behind the labels
Imagine an uptrend at one fixed timeframe:
high 1 high 2 high 3
/\ /\ /\
/ \ / \ / \
/ \ low 1/ \ low 2 / \
higher low higher low
close above high 2 = bullish BOS under the chosen rules
close below low 2 = bearish CHoCH under the chosen rulesThe BOS says price has exceeded a prior high while the higher-low sequence remains intact. The CHoCH says price has crossed the swing low that had been preserving that sequence. Reverse the logic for a downtrend.
This is close to older reversal analysis wearing newer vocabulary. CME Group's lesson on reversal patterns treats a break of the level between successive tops or bottoms as confirmation, while also stressing that chart patterns are indications rather than absolute rules. Schwab likewise treats broken support or resistance as a possible signal, not a forecast that must come true.
Why two traders can mark different BOS and CHoCH events
The disagreement usually starts before the break. One trader marks every local pivot. Another keeps only the swings visible on a higher timeframe. A third requires a candle close beyond the level, while another counts an intrabar wick.
Those choices change the signal history:
| Decision | Loose definition | Strict definition | Consequence |
|---|---|---|---|
| Swing selection | Every visible turn | Only pivots meeting a fixed rule | Different structure maps |
| Break rule | Any trade beyond the level | Bar close beyond the level | Different event timing |
| Timeframe | Lower chart resolution | Higher chart resolution | A CHoCH can be a pullback inside a larger trend |
| Confirmation | First counter-trend break | CHoCH followed by a new-direction BOS | Earlier signal versus later confirmation |
Both FXOpen and Alchemy Markets describe a decisive close beyond a chosen swing as the cleaner BOS convention. But that convention still leaves "decisive" and "chosen swing" undefined. Unless those words become code or a written checklist, hindsight does the editing.
The timeframe problem is equally real. FXOpen notes that a lower-timeframe BOS may sit inside a higher-timeframe pullback; Alchemy distinguishes internal and external CHoCH by the scale of the broken swing. The same bar can therefore carry different labels without either chart being arithmetically wrong. The analysts are measuring different structures.
How traders use the distinction
In discretionary SMC analysis, BOS often supports a continuation narrative and CHoCH triggers a reassessment of that narrative. Some traders wait for CHoCH and then require a BOS in the new direction. Others combine the event with liquidity, an order block or fair value gap, or a higher-timeframe bias.
That workflow can impose discipline because it forces a trader to name the swing that invalidates the current reading. It can also produce false confidence. Adding more SMC labels does not automatically add independent evidence; several labels may be different descriptions of the same price move.
The safest interpretation is descriptive:
- BOS says the selected swing sequence has extended.
- CHoCH says the selected swing sequence has been interrupted.
- Neither says why price moved.
- Neither establishes profitability without a complete entry, exit, risk and cost model.
What the evidence and the critics say
There is a gap between evidence for technical patterns in general and evidence for BOS/CHoCH as a standalone trading method. The practitioner sources define the vocabulary, but they do not establish a universal, peer-reviewed BOS or CHoCH rule with a durable expectancy across markets.
Broader research gives a more careful answer. Lo, Mamaysky and Wang built an automated method for recognising technical patterns and found that several patterns added information to return distributions in their historical equity sample. They also identified subjectivity as a central obstacle in chart analysis (NBER working paper, Journal of Finance publication record).
Evidence that a pattern contains information is not the same as evidence that a trading rule survives selection and costs. Sullivan, Timmermann and White applied a bootstrap reality check to a large universe of technical rules specifically to account for data snooping, while Bajgrowicz and Scaillet revisited technical rules with false-discovery controls, persistence tests and transaction costs (Journal of Finance study, Journal of Financial Economics study).
The fair conclusion is narrow: price structure can be formalised and tested, but the BOS/CHoCH label is not evidence of an edge. The evidence belongs to the exact rule, market, timeframe, sample and cost assumptions.
How you'd actually test it
Start by deleting every adjective from the rule. "Important swing", "clean break" and "strong displacement" are not machine-readable.
Pre-register the minimum specification:
- Define swing highs and lows with a fixed pivot algorithm or lookback rule.
- Define the active trend from an exact sequence of confirmed swings.
- Define a BOS as a close beyond the relevant continuation swing.
- Define a CHoCH as a close beyond the protected counter-trend swing.
- Decide whether the event is the entry, a filter, or only a state change.
- Fix the entry delay, stop, exit, maximum holding time and position-sizing rule.
- Charge spread, slippage, commission and swap.
- Freeze the design before evaluating unseen data.
Then compare useful variants rather than searching until one looks good:
| Test | Question |
|---|---|
| Wick versus close | Does close confirmation reduce false breaks enough to offset later entry? |
| Internal versus external swing | Does stricter swing selection improve expectancy or merely cut the sample? |
| CHoCH alone versus CHoCH then BOS | Does added confirmation improve results after the extra delay? |
| One timeframe versus aligned timeframes | Is the filter useful or another fitted condition? |
| Trend versus range regime | Does the rule depend on market state? |
| Gross versus net results | Does any apparent edge survive real execution costs? |
The test must prevent look-ahead. A pivot is not known until its confirming bars have printed, so the strategy cannot act as though the swing label existed earlier. The same rule must also survive nearby pivot settings; parameter sensitivity testing is the check against one lucky definition, and look-ahead bias in backtesting explains why retrospective swing labels can manufacture an edge.
Finally, hold out unseen data and report event count, expectancy, drawdown, losing streaks and results after costs. A single annotated chart is an explanation. It is not a test.
realbacktesting is a trading-software studio for cTrader built around reproducible backtests. The same standard applies here: turn the chart language into rules, include the costs, and make the result checkable (how realbacktesting verifies cTrader backtests).
Frequently asked
Is CHoCH a confirmed trend reversal?
No. Under the common SMC convention, CHoCH is the first meaningful break against the selected trend structure. It warns that the prior sequence may be changing; it does not prove that a new trend will persist.
What is the difference between BOS and CHoCH?
BOS breaks a relevant swing in the direction of the prevailing structure, while CHoCH breaks a relevant swing against it. The distinction depends on the trend and swing rules chosen before the event.
Does a wick count as a BOS?
There is no universal standard. Many practitioner definitions require a candle close beyond the swing, but a valid backtest must state the rule explicitly rather than decide after seeing the outcome.
Can BOS and CHoCH be automated?
Yes, once swing selection, break confirmation, trend state and timing are objective. Automating vague discretionary labels only hides the subjectivity inside code.
Takeaway
BOS and CHoCH can organise a chart. Only a frozen, costed and out-of-sample rule can tell you whether that organisation is useful.