Direct answer
Divergence in a Trend Intensity Index (TII) means the TII suggests one direction or change in trend strength, while the price action you are comparing against suggests another. The key point is disagreement: the index and the reference trend are not moving together.
In practice, “divergence” is a comparison, not a built-in guarantee. It is a label for a pattern of mismatch between two time series (the index and something derived from price). How you define “divergence” matters just as much as the observation.
Mechanism or definition
A Trend Intensity Index is designed to measure how strong or how consistently a trend is behaving, rather than only where price is located. Typically, such an index is computed from price-derived inputs (for example, differences or directional movement) and then smoothed or aggregated. Because of that construction, the TII can react differently from raw price.
A simple way to think about divergence is:
- Choose a reference about price (for example, direction of recent highs/lows, or a trend measure).
- Observe the TII’s trend intensity over a similar time window.
- Divergence exists when these two directions disagree (e.g., price pushes to make a higher swing, while TII weakens).
Assumptions for an example: suppose you compare a short-term price trend to a TII computed from the same price series using a fixed lookback and smoothing. If price becomes more “directional” but the index indicates reduced intensity, that disagreement is what people call divergence.
Material limitation: the meaning of “disagree” depends on your exact rule. For instance, “making higher highs” is not the same as “improving momentum,” and “turning down” in an index is not the same as “crossing a threshold.” Small changes to the definition can change what you count as divergence.
Evidence or example (how to check without over-claiming)
You can independently verify divergence definitions using only historical data you already have. A responsible approach is to check whether the observation is consistent across:
- Multiple time periods (different regimes, not just one window).
- Multiple parameter choices consistent with your TII definition (for example, different lookback lengths if your method allows).
- Different ways of measuring “price direction” (so you confirm you’re not relying on one fragile interpretation).
A common failure mode is assuming a divergence “explains” what happened next. Even if price later turns, divergence might not have been the driver—it may simply be a coincidental outcome of the same underlying market dynamics and the index’s smoothing.
Confirmation and hindsight bias can strengthen the illusion of meaning. If you focus only on the divergences that were followed by noticeable turning points, you may overestimate how often divergence actually occurs before turns, and how reliably it does so.
Limitations and risks
Here are the main limitations to keep divergence in perspective:
-
Construction effects Because TII is computed from price transformations and often smoothed, it can lag or dampen changes. This can create divergence without any special “signal meaning,” simply because the index representation differs from the raw series.
-
Definition sensitivity If your divergence rule is informal (“looks like it”), results can’t be independently checked. A measurable rule (e.g., how you define swing points or trend direction) is necessary for verification.
-
Confirmation limits and hindsight bias Divergence can be interpreted as predictive only after seeing the outcome. This makes it easy to ignore divergences that did not lead to meaningful changes.
-
External variability Even when the mechanics are stable, market conditions change. Costs, execution quality, and differences in how data is sampled (timeframes, feeds, and aggregation) can affect what actually happens after an index observation.
Verification or next question
To explain divergence in TII accurately, you should be able to answer three checkable questions:
- What exact rule defines “divergence” between TII and your chosen price reference?
- What inputs and smoothing choices define TII in your setup (even if only conceptually)?
- How would you test the rule across multiple periods without selecting examples that already “worked”?
If you want to go one step deeper, a useful next question is how you would backtest a divergence rule in a way that avoids cherry-picking and makes the definition measurable.