◎ Full transparency

Forecasts, modes, and outcomes

Track Record separates cautious forecasts from higher-variance research. Accuracy appears after recorded markets resolve.

Balanced Mode

A cautious view that favors stronger evidence and smaller, more defensible probability gaps.

◎72.0%Forecast Accuracy
▥267Total Forecasts
↗8.7%Avg Edge %
♨10WBest Streak

◎ Calibration Curve

Each point shows published observed accuracy for a probability bucket. Hover or focus on a point to inspect its values.

0%20%40%60%80%91%0%20%40%60%80%Predicted probability
Expected 50% · Observed —
Dashed: ideal calibration · Solid: observed accuracy (7 buckets)

Accuracy by Category — Balanced Mode

CategoryResolvedAccuracyAvg Edge vs Market
Crypto4871.2%+9.4%
Weather2281.8%+6.1%
Sports6168.9%+8.7%
Esports1973.7%+10.2%
Geopolitics2766.7%+11.5%
Politics3470.6%+7.9%
Finance4175.6%+8.3%
General1573.3%+5.8%

Balanced Mode Resolved Forecasts

DateMarketCategoryAI PredMkt PriceEdge %SideOutcome
Oct 24, 2026Will ETH close above $4,000 this week?Crypto62%54%+8.0%YEScorrect
Oct 23, 2026NYC measurable rain before Oct 25?Weather28%41%+13.0%NOcorrect
Oct 22, 2026Fed holds rates at next decision?Finance71%66%+5.0%YEScorrect
Oct 21, 2026LCK team wins Worlds group stage Bo1?Esports44%52%+8.0%NOincorrect
Oct 20, 2026US House passes spending package this month?Politics39%47%+8.0%NOcorrect
Oct 19, 2026NBA opener: home team covers?Sports57%51%+6.0%YEScorrect

▤ How We Measure Accuracy

◎

CRPS: Beyond Win/Loss

Continuous Ranked Probability Score measures how close probability estimates are to the outcome — not only whether the side was right.

◯

Nightly Recalibration

Resolved forecasts feed per-category calibration multipliers so systematic over/under-confidence is corrected over time.

☍

Model Disagreement Scoring

When models disagree, confidence is lowered. Agreement across the ensemble raises weight on the published probability.