How we measure reversal opportunities
The current model starts with a simple question: is the ongoing move behaving like historical swings that were close to reversal? We do not ask the user to interpret dozens of indicators. The system first identifies comparable swings, then evaluates the current structure against them.
1. Measure how unusual the current move is
One intuitive component is swing age. A move that has lasted much longer than comparable historical moves deserves more attention, although duration alone never creates a reversal signal.
For example, a 15-day swing with a typical duration of 8 days has an extension ratio of approximately 1.88×. The model combines this with price structure and candle geometry rather than treating the ratio as a trading rule.
2. Convert market state into model features
Swing age, amplitude, extrema
Bodies, wicks, ATR-normalized shape
Probability of reversal
The function f is the trained statistical/ML model. This notation is intentionally generic: the product should not pretend that the probability comes from one fixed technical formula when the actual model is trained on historical observations.
3. Add context without forcing the user to read it all
Later models add sentiment, related-market behavior and scheduled events. Their job is not to produce more panels. They answer a smaller number of questions: does the broader environment confirm the setup, where could the reversal occur, and what could invalidate the expectation?
What the user should see
The main interface should say only what is decision-relevant: a high reversal probability, whether other models confirm it, the current swing age versus typical, and the projected zones. The evidence behind each observation stays one click away.