Direct answer: when can swing highs and lows fail?
Swing highs and swing lows fail when the conditions that make the pattern label stable no longer hold. Common failure cases include: (1) market regime sensitivity, where volatility, trendiness, or ranging behavior changes; (2) cost and definition problems, where spreads, slippage, or the chosen “significance” threshold make the same price action look different; and (3) execution and data problems, where timing mismatches, delayed quotes, or missing candles prevent the pattern from being reproduced.
“Failure” here does not mean the idea is useless. It means the labeling and the conclusions you try to attach to it become inconsistent with the underlying mechanics.
Mechanism and definition: what are swing highs and swing lows?
A swing high is a local peak in price that stands out relative to nearby candles/bars; a swing low is the corresponding local trough. In practical charting, people decide a swing boundary using a rule such as “a swing point is higher than the surrounding candles by some amount” or “it occurs after a directional move of sufficient length.”
That boundary rule is the key input. If you keep the rule constant, swing points should be comparable. If your rule is implicit, changes over time, or reacts to noise, then the swing labeling becomes unstable.
Evidence and example: why costs, regime, and thresholds break labels
Consider a simplified assumption set: you define a swing high as a candle whose high is the maximum within a window of N bars, and you treat that as the “turning point.” Now look at three variable factors.
-
Regime sensitivity (structure changes). In a trending regime, turning points may be spaced further apart and cleaner to detect. In a ranging or choppy regime, local maxima/minima occur more frequently and can alternate rapidly. With the same N-bar window, what once looked like a meaningful swing can become one of many small oscillations. Even if the swing rule is unchanged, the market’s behavior makes the rule pick up more noise.
-
Costs and “significance” distortion (costs-features mismatch). Many chart-based interpretations implicitly assume that the difference between two price levels matters in a near-frictionless way. In reality, your realized entry/exit depends on spreads and slippage. If your interpretation requires price to move “enough” to overcome transaction costs, but your swing definition is purely geometric (it only looks at highs/lows), you can end up believing a swing turn is operationally meaningful when it is not.
-
Execution and data timing (timing assumptions break). Swing detection often uses candle closes or extremes inside a bar. If your execution is delayed (for example, fills occur after the candle that “contains” the swing point), then the price you can actually trade around may differ from the labeled level. Similarly, changes in feed quality, chart timeframe, or candle construction can produce different swing points.
One material limitation / failure mode
A common material limitation is non-stationarity: the statistical properties that make swing points readable (trend persistence, typical candle ranges, and noise level) can change. As a result, a swing labeling approach that “worked” in one regime can produce different swing frequency and different turning-point quality in another.
Limitations and risks: what you can independently verify
Because there is no single universal definition of “significant,” verification matters.
- Assumptions: state your swing rule explicitly (window size, whether you use high/low extremes or closes, and whether you require a minimum distance or confirmation). If you cannot state it, you cannot reliably reproduce outcomes.
- Variability checks: repeat the same swing labeling across multiple timeframes and across periods with different volatility/ranging behavior. If swing points change drastically under small charting differences, the method is fragile.
- Cost realism: test how sensitive your conclusions are to plausible friction (spreads/slippage). Swing geometry alone may not reflect what is achievable once execution constraints are added.
- Data consistency: ensure the candle series you use for labeling matches the series you would have had for decision-making. Timing mismatches can create apparent “precision” that disappears in practice.