Definition: what “divergence” means in Frama
In this context, “divergence in Frama” means a mismatch between (1) how Frama (the indicator line) is changing and (2) how the market price is changing. For example, price may continue rising while Frama turns sideways, flattens, or starts to fall. Or price may fall while Frama becomes less bearish. The key idea is not that divergence is a unique prediction, but that it is a description of disagreement between two moving behaviors.
A practical way to phrase it is: divergence is a visual and analytical difference between Frama’s direction/slope and price’s direction/slope.
How Frama is constructed (and why that matters)
Frama is a moving-average-type indicator whose smoothing strength changes with an input related to market structure or “complexity.” In plain language, it tries to adapt its responsiveness: when the market looks more orderly or fractal-like, it can behave differently than when the market looks more irregular.
That adaptive design creates two common reasons you can see divergence:
- Different responsiveness: Frama may respond faster or slower than price depending on the indicator’s complexity input and your chosen parameters.
- Smoothing and lag: Any moving average is effectively an averaging process. Even when the market is moving strongly, a smoothed estimate can trail or dampen the change, creating disagreement.
So divergence often reflects indicator mechanics (how Frama smooths and adapts) as much as it reflects anything about the future.
A simple model to think with (no live data)
Assume you compute Frama on the same price series and then compare slopes:
- Price slope: whether the recent prices are trending up or down.
- Frama slope: whether the indicator line is turning up or down after its adaptive smoothing.
A “bullish-looking” divergence example (terminology varies) might be:
- Price makes a new local high.
- Frama does not make a comparable new high, or its slope stops increasing.
An “opposite” example is:
- Price makes a new local low.
- Frama flattens or its downward slope weakens.
Assumption for this example: you are using the same time window and Frama settings throughout, and “new local high/low” is defined consistently (for instance, based on a swing high/low definition). Without that, divergence can be subjective.
Confirmation limits: why divergence is easy to over-read
A common failure mode is confirmation bias through hindsight. Once you know what happened afterward, it becomes easier to notice divergence that “fits” the outcome. This can lead to a belief that divergence implied a specific direction.
Even with careful definitions, there are additional limitations:
- Regime dependence: A divergence may appear when volatility changes, not when a durable reversal is starting.
- Parameter sensitivity: Changing Frama’s settings can alter how quickly the indicator turns, which changes whether divergence appears.
- Noise sensitivity: Short-term divergence may be driven by small fluctuations that do not persist.
Material limitations and potential failure modes
At least one material limitation to keep in mind is that divergence in a smoothed adaptive indicator does not uniquely map to one market outcome. It can be caused by:
- Lag: Frama can be late in turning relative to price.
- Reweighting of smoothing: Adaptive smoothing can change mid-stream when the complexity input changes.
- Non-stationary behavior: The statistical relationship between price movement and the indicator can shift over time.
Because outcomes vary with costs, execution quality, and jurisdictional factors, historical patterns do not guarantee anything about future results.
How to verify divergence claims independently
To verify what divergence “means” for your use case, treat it as a measurable description and test it with rules you can reproduce:
- Define divergence using explicit criteria (e.g., slope sign change, relative highs/lows, or minimum separation).
- Fix Frama parameters and the time frame before looking at outcomes.
- Compare multiple periods, not only the segments where divergence happened to coincide with good outcomes.
- Re-check your conclusions for sensitivity: if the result changes heavily when parameters shift slightly, divergence may be an artifact of the method.
If you want, you can also test whether divergence corresponds more strongly to “state change” (trend weakening/strengthening) rather than a directional event. That reframes the indicator from prediction to measurement.