MODEL ANALYTICS

Hit rate, ROI
and calibration.

Every percentage includes its denominator. Official performance uses only frozen Daily Top 5 selections at their original published line and price. Research diagnostics are displayed separately.

AUTO-UPDATES EVERY REFRESH
OFFICIAL PERFORMANCE

Core metrics

Checking latest settled sample…

Current production model v—Official ledger Frozen by publication version
The current research engine can be newer than the model version shown in the official-performance breakdown. Historical Top 5 picks stay attached to the version that originally published them.
Official recordN/A
Loading frozen results
Hit rateN/A
Only settled W/L decisions count
ROIN/A
Flat 1-unit stakes
PendingChecking
Checking published picks
TIME WINDOWS

Recent hit rates

7 · 30 · 90 days · all time

7 daysN/A
Awaiting settled sample
30 daysN/A
Awaiting settled sample
90 daysN/A
Awaiting settled sample
All timeN/A
Awaiting settled sample
BREAKDOWNS

Where the model performs

Tap filters in the tables

By league

Loading league denominators…

By market

Loading market denominators…

By confidence

Loading confidence tiers…

By odds range

Loading price buckets…

By sportsbook

Loading stored books…

By model version

Loading model history…
TREND

Rolling performance

Most recent 120 tracked days

Trend begins as official picks settle. Pending picks remain visible in the record.
MODEL HEALTH

Probability quality

Only stored pregame probabilities

Expected hit rateN/A
Requires stored pregame probabilities
Brier scoreN/A
Requires settled probability sample
Calibration errorN/A
Requires probability buckets
RESEARCH DIAGNOSTIC

Separate from official record

Loading the latest model-improvement audit…
SOURCE ACCOUNTABILITY

Which inputs earn trust

N ≥ 30 · calibration + CLV first

Checking sample-gated source reliability…

Data-quality rules

Hit rate = wins ÷ (wins + losses). Pushes never inflate hit rate. ROI uses flat one-unit stakes. Missing metrics show N/A instead of 0%. Published lines, prices and probabilities are never rewritten after outcomes are known. Small samples include a 95% Wilson confidence interval.