Understanding the AI Score

What the 0–100 AI score measures, the signals inside it, and how to use it as a conviction gauge.

What the AI score is

The ŷRobot AI score is a single number between 0 and 100 that summarizes how consistently seven weighted analytical signals point in the same direction for a stock. A score near 100 means most signals agree; a score near 0 means they mostly disagree or lean the other way. The number does not predict what a stock will do. It is a measure of agreement across models, not a forecast.

That distinction matters. Any individual signal can be wrong for reasons the models do not see: a pending acquisition, a lawsuit, or a macro event outside the data. The AI score gives you probabilistic context — a quick read on whether multiple independent lenses are aligned — so you know when conviction is broad and when it rests on just one or two factors.

The components and their weights

Each weighted component contributes a share of the final score. Rounded, the shares are: fair-value gap about 38%, 90-day seasonal forecast 25%, fundamentals 15%, analyst consensus 10%, 30-day seasonal forecast 5%, earnings beat streak 4%, and the volatility penalty 4%. Two more readings appear on every scorecard for reference and carry no weight: the chart signal, which reads about five days ahead, shorter than the score's horizon, and P/E versus the company's own history, which did not improve the score when tested on ten years of history.

The fair-value gap is the largest single input. ŷRobot estimates a stock's 6-month fair value using a CatBoost regression model trained on roughly 40 fundamental, growth, cash-flow, and price-momentum inputs, measures the gap between that estimate and the current price, and ranks that gap against stocks of similar size. The seasonal forecasts add a separate model's directional view. Fundamentals, analyst consensus, and the earnings beat streak each add context from a different angle: reported financial health, professional opinion, and a company's recent record of clearing earnings estimates. Volatility acts as a penalty: high uncertainty compresses the score even when other signals agree. The chart signal and P/E versus history are shown for reference only and carry no weight in the score.

Direction and confidence

Alongside the 0–100 score, ŷRobot shows a direction (positive or negative) and a confidence level. Direction uses hysteresis, so it does not flip on the slightest change and stays stable through small oscillations in the underlying data. Confidence blends two factors: how complete the input data is for that stock, and how tightly the weighted components agree with each other.

A stock with sparse data in some components will show lower confidence even if the available signals are aligned. That is intentional: the system is more certain when it has more to work with. Low confidence is not a reason to ignore the score — it is useful information about how much weight to place on it.

How to read it in the app

The AI Scorecard appears on every stock page at /asset-breakdown/<SYMBOL>. It breaks out each component so you can see which signals are driving a high or low composite. The Smart Screener includes an AI Score column, letting you sort and filter across the 4,000+ US stocks ŷRobot covers.

The score is designed as a starting point, not a conclusion. Because ŷRobot fuses a fair-value model, seasonal forecasts, fundamentals, analyst consensus, the earnings beat streak, and a volatility penalty into one number, a high score means several different analytical approaches agree, which is more informative than any single approach alone. But one strong signal can still mislead if it happens to dominate the composite. Always treat the score as an agreement gauge, and look at the individual components, including the two shown for reference, to understand what is driving it.

ŷRobot analysis is AI-generated and quality-gated; nothing on this page is investment advice.