Quick definition
A numerical score (0-100) generated by Arch-AGI that represents the AI's confidence in a trade's expected outcome, based on technical, fundamental, and sentiment analysis.
What the number actually means
A score of 70 doesn't mean a 70% probability of profit — it means the agent's historical performance on setups that scored 70 has matched the platform's expected risk-adjusted return curve. Calibration is recomputed continuously from the reinforcement-learning feedback loop, so scores drift toward truth as more outcomes settle.
How it changes downstream behavior
The Adaptive Risk Engine reads conviction before tightening stops, scaling out, or letting a winner run. A high-conviction position gets more rope; a low-conviction position gets a tighter trail. Position sizing, when delegated to the Kelly sizer, multiplies the calibrated edge by conviction so the system risks more when it has earned the right to.
Why it isn't just a probability
Raw probabilities ignore variance. Treeova folds expected magnitude, time-to-expiry, and historical outcome dispersion into a single comparable score so two very different setups — say a 0DTE momentum scalp and a 45-DTE iron condor — can be ranked on the same axis without misleading the user.
Go deeper
The full technical methodology behind Conviction Scoring is documented in the Treeova whitepaper series.
Read the Conviction Scoring whitepaper →