Core comparison: same break mechanism, different prior context
Both labels require a reference swing known before the decision. A continuation label describes the direction of a completed price event; a counter-trend break describes a possible transition. Neither supplies a measured win rate or risk ranking.
| Feature | BOS convention used here | CHoCH convention used here |
|---|
| Prior state | Uptrend or downtrend already classified | Same prior trend state |
| Uptrend example | Close above selected swing high | Close below selected higher low |
| Downtrend example | Close below selected swing low | Close above selected lower high |
| Interpretation | Break agrees with the prior direction | Break challenges the prior direction |
| Risk | Depends on entry, stop, size, costs, and outcomes | Depends on the same factors; not inherently a quantified risk tier |
| Confirmation | Our examples use a completed close beyond the level | Our examples use a completed close beyond the opposing level |
What a CHoCH does and does not establish
For a hypothetical uptrend with confirmed lows at 90 then 100 and highs at 110 then 120, a later close at 99 breaks the selected 100 higher low. Under this convention it is a bearish CHoCH. That observation does not prove distribution or a new sustained downtrend.
- Bullish case: a downtrend breaks its selected lower high; record the level and confirmation time.
- Bearish case: an uptrend breaks its selected higher low; do not substitute a smaller internal low after seeing the outcome.
- A subsequent recovery can invalidate the directional interpretation. Preserve such cases in a test instead of relabeling them away.
- An entry on a later retracement is a separate rule. Define its price tolerance and timeout; a retracement is not guaranteed.
BOS vs MSS: define displacement before using the label
Some SMC conventions reserve MSS for a counter-trend break with a large directional move, sometimes accompanied by a three-candle fair value gap. Others use MSS and CHoCH interchangeably. There is no single exchange-standard definition, and a candle or gap cannot prove that institutions initiated the move.
| Label | Classification | Optional filter | What it does not prove |
|---|
| BOS | Break agrees with prior trend | Close beyond the selected swing | Future continuation |
| CHoCH | Break opposes prior trend | Close beyond the selected opposing swing | A completed reversal |
| MSS | A separately specified structural shift | For a test, define body/range or ATR threshold and exact FVG condition | Institutional participation or higher expected return |
A hypothetical sequence combining CHoCH and BOS
Assume an uptrend has a confirmed high at 120 and higher low at 100. The following invented prices illustrate labels only; they are not a backtest or an inevitable market cycle.
- 1. A completed close at 121 breaks the 120 reference high: bullish BOS under the stated convention.
- 2. Price later closes at 99, below the previously selected 100 higher low: bearish CHoCH.
- 3. A low at 95 and later high at 105 become confirmed. Record their confirmation times before using them as references.
- 4. A subsequent close at 94 breaks 95. If the state rule has now classified a downtrend, label bearish BOS; otherwise keep the classification unresolved.
Common classification and testing errors
A chart can contain different structures on different timeframes. Choosing whichever swing fits the result introduces hindsight bias. Waiting for a closing price removes one kind of intrabar ambiguity but cannot prevent losses.
- Changing pivot length or switching internal/external swings after observing the outcome.
- Using a retrospectively drawn pivot as if it had been known on that earlier bar.
- Treating a wick, large candle, or volume increase as proof of stop hunting or participant identity.
- Ignoring costs, stop gaps, data coverage, or the order of intrabar stop and target touches.
Coding BOS and CHoCH rules in Pine Script
Specify the state machine before asking a coding tool to implement it: how the trend starts, which confirmed pivot is tracked, whether equality counts, how breaks reset, and when a direction changes. Pineify can help draft that logic, but generated code still needs compilation and behavioral review.
- Use pivot left/right lengths explicitly and record the bar where the pivot becomes available.
- Track prior trend state and update it according to written transition rules, not the name of a marker.
- If adding displacement or FVG filters, define their measurable values and evaluate their incremental effect on untouched data.
- Test entries, exits, costs, and fill assumptions in a strategy. An indicator marker alone has no performance report.