Distribution interpretation
The Distribution page computes the frequency of each outcome over a configurable rolling window. Higher bars indicate statistical bias used by the model.
Knowledge base
How the app works
AI model
Ensemble of gradient boosting, random forest, and lightweight attention layers. Trained on rolling windows of Deriv tick data with walk-forward validation.
Data security
Credentials are encrypted at rest with AES-256-GCM. Sessions are HttpOnly cookies. No third-party server receives your personal data.
Confidence
Confidence is the weighted agreement across the agents inside the ensemble. Values above 85% indicate strong agreement; below 65% signals high uncertainty.
Analysis window
Use the slider on the Distribution page to set the rolling window. Shorter windows react faster to recent changes; longer windows are slower but smoother.
Contract types
Matches predicts the same outcome repeats. Differs predicts a different outcome. Even/Odd is parity. Over/Under is above or below the midpoint.
Frequently asked questions
Loading FAQ…
Glossary
- Derived market
- Synthetic index generated by a deterministic algorithm, not affected by external economic events.
- Rolling window
- Number of recent ticks used for frequency analysis.
- Confidence
- Weighted average of model outputs, expressed as a percentage.
- Ensemble
- Combination of multiple predictive models to improve stability and accuracy.
- Walk-forward validation
- Time-series cross-validation that respects chronological order and prevents look-ahead leakage.
- Brier score
- A measure of calibration for probabilistic predictions. Lower is better.