Market Signals

Calibrated Forecasts & Market Analytics Dashboard

Success Metrics & Targets

82%
Forecast Calibration Accuracy
(Brier score ≤ 0.2)
94%
User Engagement Rate
(session ≥ 3 min & ≥ 3 pages)
+12pp
Signal-to-Noise Improvement over baseline
(measured by Sharpe ratio proxy)
<$0.45
Average Cost per Signal Update

Detailed Targets

MetricTargetMeasurement Window
Brier Score≤ 0.20Rolling 30 days
Engagement Rate≥ 94 %Rolling 7 days
Sharpe Improvement+12 percentage pts over baselineMonthly cohort
Cost per Signal≤ $0.45Weekly aggregate

Calibrated Forecasts

S&P 500 7-Day Return

P(↑) = 63 %   (actual observed: 61 %) — Brier: 0.048

Well-calibrated

EUR/USD Direction

P(↑) = 58 %   (actual observed: 60 %) — Brier: 0.042

Well-calibrated

Crude Oil Price Move ≥ $3/oz

P(↑) = 71 %   (actual observed: 68 %) — Brier: 0.072

Slight drift — retraining queued

Market Analytics

1,247
Active Signal Sources Connected
58 ms
Median Forecast Latency (p50)
4.2 %
Outlier Rate in Signal Stream

Kill Criteria

CriterionThresholdMeasurement Window
Brier score degrades beyond tolerance> 0.40 for two consecutive rolling 30-day windowsRolling 30 days, evaluated bi-weekly
Engagement drops below minimum viable< 50 % over any rolling 14-day windowRolling 14 days
Cost per signal exceeds budget ceiling> $1.20 for any single dayDaily aggregate

Action when kill criteria are met

Kill trigger: If any kill criterion is hit on two consecutive checks, the experiment halts. All active forecasts are frozen and archived. The team is notified via Slack channel #signal-experiment-alerts. A post-mortem is scheduled within 48 hours.