The receipt behind a stock’s conviction score. Rather than handing you a single number and asking you to trust it, the ledger shows every input that produced it, how heavily each one counts, and exactly how many points each contributed — so you can see not just how convinced the platform is, but precisely what convinced it.
Three rows come from the quantum forecast and carry most of the weight: which way the model expects the stock to go (30%), how bad the worst case looks (15%), and how reliable the model has been on this particular stock historically (15%). Together that is 60% — the forecast leads.
Four rows come from the chart, each worth 10%: whether the trend is up or down, whether the move is speeding up or slowing down, whether the stock looks overbought or oversold, and where it sits in its 52-week range. The chart confirms the thesis; it does not create it.
A final row, path momentum — whether the stock has actually been tracking the model’s predicted path this week — is shown for transparency but carries zero weight. We backtested it and it earned nothing, so it no longer moves the score; it is displayed so you can see it, not because it counts.
Each row scores between 0 and 1 on its own question, and the bar beside it shows the points that score contributed to the final 0-10 total: the component’s score multiplied by its weight, on a 10-point scale. A row’s maximum possible contribution is therefore just its weight — the 30%-weighted direction row can contribute at most 3.0 points, while a 10% technical row can contribute at most 1.0. The path-momentum row, at zero weight, contributes nothing.
Example (illustrative): if the direction row scores a full 1 and it carries 30% of the weight, it contributes 3.0 points of the total. If the RSI row scores 0.5 at 10% weight, it contributes 0.5. Add up the scored rows and you have the conviction score — which is why long green bars at the top of the ledger matter far more than long green bars at the bottom.
Two of the scored rows are not merely weighted — they are gates. If the direction row or the tail-risk row scores zero, the entire conviction score is capped at 3.0 out of 10, no matter what the other rows say. When that happens the ledger tells you so directly.
This is the single most important thing to understand about the ledger, because it looks like a bug the first time you see it: a stock with a beautiful chart, strong trend, healthy momentum and a perfect 52-week position can still score 3.0, because the forecast itself sees no edge or sees an unacceptable downside. That is deliberate. Technical indicators are allowed to confirm a thesis the forecast supports; they are never allowed to invent one it does not.
Read the top three rows first — they decide the outcome. Then ask whether the reasons behind the score are reasons you actually believe: a score driven mostly by the forecast is a different proposition from one driven mostly by a hot chart, even if the totals match. The ledger exists so you can disagree with the platform for specific reasons rather than vague ones.
Then look below the ledger at the safety checks, which are a separate step: the ledger says how convinced the model is, and the safety checks say whether we will publish a BUY at all.
Almost always the cap. If the direction row or the tail-risk row scores zero, the total is capped at 3.0 out of 10 however good the other rows look — technicals cannot rescue a weak forecast.
No. It is a heuristic blend of how strongly the scored signals agree, on a scale chosen for readability. It is not a calibrated probability and carries no hit-rate or performance claim. How the forecasts actually held up is published on /calibration and /track-record.
Because they are confirmation, not evidence. Trend, momentum, RSI and 52-week position are public indicators every trader already watches, so they carry no edge on their own. They are there to check that the market is not contradicting the forecast — which is why together they carry less weight than the forecast rows and cannot override the cap.
Browse all S&P 500 tickers to see this metric applied to individual companies.
Educational research only — not investment advice.