Track Record

Does Quorum beat the market? Win rate, alpha, and calibration for Brier-weighted consensus vs the betting line. The only number that decides whether the machine works.

Scope: Two-stage MCP games only
--
Win Rate
--
Alpha vs Market
--
Games Resolved
--
Awaiting Result
--
Scored Analysts
--
Win Rate
Consensus picks the winner
--
Market Win Rate
Favorite wins (baseline)
--
Alpha
Quorum minus market
--
Upset Detection
Correctly called upsets
--
Divergence ROI
$100/game on 10%+ divergence
--
Signals Resolved
Individual calls scored
Performance by Approach

Each signal method is a different system. Evaluated separately against market baseline.

Treatment Comparison (All Approaches Combined)

BW vs Raw vs Market on games with odds. Lumped view — see approach breakdown above for meaningful comparison.

A: Brier-Weighted
--
B: Raw Headcount
--
C: Market Line
--
Statistical Significance
Wilcoxon Signed-Rank
--
BW vs Market (McNemar)
--
Raw vs Market (McNemar)
--
Est. Games to p < 0.05
--
Convergence Strength → Accuracy

Does stronger analyst agreement predict better outcomes?

Alpha by League
Loading...
Signal Diagnostics

Why results look the way they do. Tracks confidence calibration, analyst independence, and market context.

--
Confidence Std Dev
Target: 0.03+ (spread). Problem: 0.01 (clustered)
--
Unanimous Games
% of games with 90%+ direction agreement
--
Market Upset Rate
% of games where market favorite lost
Confidence Distribution
Performance by Signal Method
How Analyst Weighting Works

Every analyst's influence on the consensus is determined by their Brier score. The relationship is continuous — there are no tiers or cutoffs. Lower Brier = exponentially more weight.

Loading...
Accuracy Over Time
Calibration
Scored Games
Loading...
Divergence (10%+ vs Line)
Accumulating data as games with market odds resolve...
Metric Definitions
Win Rate
Percentage of scored games where the Brier-weighted consensus correctly predicted the winner. This weights analysts with proven track records more heavily.
wins / total_resolved
Alpha vs Market
Quorum's win rate minus the market's implied win rate. Positive alpha means Quorum outperforms Vegas. This is the core value metric - are scored analysts finding edge the market misses?
quorum_win_rate - market_favorite_win_rate
Analyst Weighting
Each analyst's influence follows a hyperbolic curve. Weight = 1/(Brier + 0.01), multiplied by signal confidence. Backtested and validated across resolved games — the only formula producing positive alpha vs market. A 0.03 Brier analyst gets 5x the influence of a 0.20 analyst.
weight = 1 / (brier + 0.01)
Upset Detection
When the underdog wins, how often did Quorum pick them? These are the most valuable calls - the market says one thing and Quorum's analysts saw something different. High sensitivity to upsets is the product's edge.
correct_underdog_picks / total_upsets
Divergence ROI
Hypothetical return betting $100 on every game where Quorum's consensus diverged from the market line by 10%+. Uses closing-line implied probability to compute payout (underdog picks pay more, favorite picks pay less). Positive ROI = Quorum finds real value the market misses.
(total_payout - total_wagered) / total_wagered × 100
Calibration
When Quorum says 80% confidence, does that team actually win 80% of the time? The calibration chart plots predicted probability vs actual outcome frequency. A perfectly calibrated model follows the diagonal.
actual_win_rate_at_confidence_level