68%
prediction accuracy (clean score)
Always guessing ‘down’ would score 64% on these same cases, so the edge is about +4 points. Not yet statistically meaningful.
A public record of prediction accuracy and the quality of Pegasus Alpha research.
Accuracy over time
Cumulative · last 90 daysDistance to goal
164 days left · deadline 30 Apr 2027The six-month evidence target.
Right with at least 10% sector-relative excess return.
Two evidence sets
Shown separatelyClean score
68%47 graded · 95% CI 54%–80%
History weaker
58%53 graded · 95% CI 45%–71%
weaker: names masked, older datesLast 10 graded predictions
Most recent first- wrong
A large US listed-company company was expected to change a measurable business outcome by 2026-11-17.
- wrong
A large US listed-company company was expected to change a measurable business outcome by 2026-11-17.
- right
Backtest: PLD was called to lag a basket of large US stocks over 3 months from 2026-07-19.
- right
Backtest: META was called to beat a basket of large US stocks over 3 months from 2026-07-19.
- wrong
Backtest: AAPL was called to lag a basket of large US stocks over 3 months from 2026-07-19.
- right
Backtest: HON was called to lag a basket of large US stocks over 3 months from 2026-07-19.
- right
Backtest: NEE was called to lag a basket of large US stocks over 3 months from 2026-07-19.
- right
Backtest: TSLA was called to lag a basket of large US stocks over 3 months from 2026-07-19.
- right
Backtest: DIS was called to lag a basket of large US stocks over 3 months from 2026-06-19.
- right
Backtest: BA was called to lag a basket of large US stocks over 3 months from 2026-06-19.