Cornerstone
AI Stock Analysis Track Record
A track record is only useful if the rules are public. WebserviceX grades published desk ratings over a short, explicit window and shows the blotter on /eval. That is how this site answers whether an AI stock analysis call aged well—without inventing long-horizon performance marketing.
Grading rules
Five-session directional scorecard
- Horizon: five US equity trading sessions after the rating entry close
- BUY / OVERWEIGHT: Correct if absolute forward return > 0
- UNDERWEIGHT / SELL: Correct if absolute forward return < 0
- HOLD: Neutral (unscored)
- Pending until the window completes; insufficient_data when closes are unavailable
- Benchmark relative return uses SPY by default when disclosed on the note
Many “AI stock pick” pages claim accuracy without stating entry timing, horizon, or what counts as a hit. WebserviceX avoids that pattern. The scorecard is directional and short-horizon by design: it asks whether the published bias lined up with the sign of forward return over five sessions. HOLD is unscored so neutrality is not gamed as a win. Pending and insufficient-data states keep incomplete windows from looking like settled results.
The track record is inseparable from publication discipline. Only notes that clear publish QC become part of the coverage set that can be graded. Versioned URLs mean the rated artifact is the same object a reader can open later. If a system cannot point to what it published on a date, a hit-rate claim is not an audit—it is a story.
Limitations are intentional and disclosed. Five sessions is not a twelve-month investment outcome. Absolute return grading is not risk-adjusted performance. Benchmark-relative fields, when present, are additional context—not a substitute for reading the note’s thesis, catalysts, and risks. The eval blotter is a process instrument for an educational desk, not a performance advertisement for managed money.
To use the track record well: open a batch on /eval, follow a symbol into its research note, and compare the rating footer to the graded outcome. For how agents produce the call, read methodology. For coverage navigation, use equity research. For product scope, see about.
Educational demonstration only. Outputs are informational—not investment advice, not a solicitation, and not a recommendation to buy or sell any security.