Definition: what “break and retest” means
Break and retest is a price-action concept used to describe two phases around a specific level on a chart. First, price moves through a defined level (the “break”). Second, price returns toward that same level and interacts with it again (the “retest”).
A key point is that this is a descriptive framework, not a guarantee. Whether the break and retest leads to continuation, reversal, or nothing depends on market conditions and execution details.
Mechanism: the inputs and rules you must choose
To work the idea into a verifiable example, you need explicit assumptions for the “level” and for what counts as a break and a retest.
Common assumptions for an example (you can change them, but then you must apply the same changes consistently):
- Level type: a horizontal price level drawn from prior swing highs/lows.
- Break rule: the closing price of a candle is above the level (for an upward break) or below it (for a downward break).
- Retest rule: later candles trade back to the level area, and you observe how price behaves there.
- Level area tolerance: because prices and charts are discrete, you define an “area,” for example ± a small amount around the level.
Separating stable mechanics from variable conditions:
- Stable mechanics are the two-phase description (break, then retest) and the idea of measuring interactions with one chosen level.
- Variable conditions include volatility, liquidity, bid/ask spread, chart timeframe, and how strict your break/retest rules are.
Worked example (fully numeric, with stated assumptions)
Assume an analyst watches a chart on a fixed timeframe and identifies a prior support level at 1.2000 (level chosen by the analyst).
Assumptions for this example
- Break criterion: a candle is a valid “break” if its close is ≥ 1.2001.
- Retest criterion: a candle is a “retest” if its low is ≤ 1.2000 and its close is ≥ 1.2001.
- Level tolerance: we treat 1.2000 as the boundary, using the +0.0001 break threshold to remove borderline ambiguity.
- No live price data is used; the numbers below are hypothetical to make the logic testable.
Scenario
- Candle A: close = 1.1996 (no break yet). High/low can vary; what matters for this rule is the close.
- Candle B: close = 1.2003 (this meets the break criterion, because 1.2003 ≥ 1.2001). Phase 1 is satisfied.
- Candle C: price falls back. Low = 1.2000 and close = 1.2005.
- Retest criterion check: low ≤ 1.2000 is true, and close ≥ 1.2001 is also true.
- Therefore, Candle C is counted as a retest interaction with the original level.
How you would interpret it (without promising outcomes)
- This example shows the mechanics: a level at 1.2000 is first broken on a closing basis (Candle B), then later retested with price interacting with that same level (Candle C) before closing back above the break threshold.
- Importantly, the example does not claim the next candles must go up. You would need to continue applying your chosen rules to see what happens in the subsequent sequence.
To keep it independently verifiable, you can publish or note the rule set (break by close ≥ 1.2001, retest by low ≤ 1.2000 plus close ≥ 1.2001) and then check each candle in historical data.
Limitations and failure modes
At least one material limitation is that break and retest can be ambiguous when your rules are not explicit.
Common failure modes include:
- False break: price closes beyond the level, but then quickly returns and does not behave like a meaningful retest (or never meets your retest rule).
- Rule sensitivity: small changes to the break threshold (e.g., 1.2001 vs 1.2000) or the timeframe can change which candles qualify.
- Execution and spread effects (a variable condition): in real trading, bid/ask spreads and order execution can mean that “where price touched” on a chart is not identical to where orders were filled.
- Changing market structure: what looks like the same level may lose relevance if volatility expands or the underlying swing points shift.
Because of these issues, the concept should be tested and verified with a consistent rule set rather than treated as a stand-alone prediction.