Direct answer
Divergence Reversal describes a chart-reading rule set where (1) price and a momentum-style measure move in opposite directions (divergence), and (2) a subsequent condition suggests the move may be ending and a new direction may begin (reversal). The “rules” are therefore not a single magic signal; they are a structured definition you can apply consistently: what counts as divergence, what timeframe or window you use, what counts as a reversal confirmation, and what you assume about data and execution.
A useful approach is to treat Divergence Reversal as a testable observation framework:
- You specify the inputs (which momentum measure, how it is computed, and its parameter settings).
- You specify the divergence rule (what opposite movement is required, and over what bar count).
- You specify the reversal rule (what evidence ends the old move and starts the new one).
- You define outcomes in a way that can be measured later (e.g., whether price reaches a target or invalidates the setup) without claiming that it will be profitable.
Mechanism and definition (the “rules” you must state)
To explain the rules precisely, separate three layers: definition, detection, and confirmation.
1) What is divergence in this context?
In divergence reversal discussions, “divergence” typically means that price makes one side of a move while the momentum-style measure does not agree.
Two common divergence directions are:
- Bullish divergence: price forms lower lows while the momentum measure forms higher lows (less bearish momentum).
- Bearish divergence: price forms higher highs while the momentum measure forms lower highs (less bullish momentum).
A rule set should state at least:
- The momentum measure you use (for example, an oscillator derived from price changes).
- The exact parameter settings for that measure.
- How you locate “higher low” and “lower low” (e.g., based on swing points, or based on rolling windows).
Because divergence can be subjective if swing points are not defined, the rule set must define the method. One testable method is to use swing detection logic: declare a swing low when the indicator and price fulfill a local-extrema condition (such as “lower than the last N bars”), then compare successive swings.
2) What counts as a reversal?
A reversal condition should be explicit. In many chart-reading workflows, a reversal is confirmed when price behavior changes in a measurable way, such as:
- Breaking above a prior minor resistance after bullish divergence.
- Breaking below a prior minor support after bearish divergence.
- Closing beyond a level (not just a brief intrabar touch), if your data supports bar closes.
Even if you prefer a visual interpretation, for independent verification you must convert it into a rule. For example:
- “After bullish divergence is detected, declare reversal only when the next bar closes above the most recent swing high formed after the divergence lows.”
That single sentence already removes a lot of ambiguity.
3) Put timing and scope into the rules
Divergence reversal can fail simply because people compare points that are far apart in time or ignore the window size. Your rule set should state:
- The maximum number of bars allowed between the first and second divergence points.
- Whether the reversal must occur within a certain number of bars after divergence detection.
- Whether you allow multiple divergence candidates before a reversal.
Without these constraints, two people applying the same idea may reach different “signals,” making verification difficult.
Evidence and example using testable assumptions (no profit claims)
Below is an example of how to make the rules testable. It is intentionally generic and uses placeholders so you can map it to your preferred oscillator and data.
Example rule set (framework)
Assume you are working on candlestick bars with known OHLC data.
Inputs
- Momentum measure: a specified oscillator with parameter settings (you must write the settings).
- Swing rule: define a swing low when price is lower than the preceding and following K bars; define swing highs similarly.
- Data scope: use one timeframe consistently.
Detection rule: bullish divergence
- Find two successive price swing lows within M bars of each other.
- Confirm divergence when the second price swing low is lower than the first, but the oscillator value at the second swing low is higher than at the first swing low.
Confirmation rule: reversal 3. After divergence is detected, declare reversal only when price closes above the swing high that immediately follows the second divergence swing low (or another clearly defined level in your rules).
Outcome definition for verification 4. For evaluation, define a forward window N bars. Measure whether price moves a specified amount away from the reversal level or whether it invalidates by closing back below that level.
Why this can be verified
Because each element is operational—oscillator settings, swing definition, divergence comparison, and confirmation trigger—another person can reproduce your observation on the same historical data. Importantly, nothing here claims profitability. It only creates a consistent way to count occurrences and measure what often happens.
Limitations and failure modes (material risks)
Divergence reversal rules have multiple built-in limitations that affect reliability.
1) Divergence is not a standalone signal
Divergence can appear frequently without any meaningful reversal afterward. If your rules treat divergence alone as the setup without a separate confirmation condition, you will often label many events incorrectly. Requiring a measurable reversal trigger reduces ambiguity but does not eliminate false positives.
2) Parameter sensitivity and subjective swing points
Different oscillator settings and different swing detection logic can change what counts as a divergence and what counts as confirmation. This means results are sensitive to the rule definitions. If two versions of the rule set produce very different counts, the concept may be too flexible to support strong conclusions.
3) Market regime changes and noise
In volatile or trending regimes, divergences can occur as “temporary disagreement” and then be corrected later. Meanwhile, small oscillations create frequent swing points, generating divergence patterns that are purely noise.
4) Costs and execution effects (when you later test)
If you evaluate outcomes, real-world execution includes spreads, commissions, slippage, and the fact that you may not enter exactly at your defined level. Even if your verification uses bar closes, those costs can change the practical result. The key point is not to predict future performance, but to recognize that measured outcomes depend on how you model these effects.