Track record, unfiltered
Every settled pick — wins, losses, pushes. No cherry-picking. Graded at the recommended price.
EV-sized off Pinnacle's fair line. 0 staked picks · 0 passes (no edge). ½ Kelly doubles every stake of ¼ Kelly — same ROI%, bigger swings.No Pinnacle fair-line snapshots in this range yet — try widening the window.
Performance by signal½ Kelly · 0 staked
Each pick counted once, under its strongest (primary) signal. This is the "no double-counting" view. Steam and Sharp Lead are stack-only confidence boosters in the current model — see the trail view below for their lift.
Performance by signal trail½ Kelly · 0 staked
Same picks, but counted under every signal that fired. A pick tagged underdog with a steam stack appears in both the Underdog and Steam rows. Use this to see how each strategy performs in isolation — including stack-only boosters like Steam and Sharp Lead.
Signal performance over time
Cumulative units per signal — same trail semantics as above. A pick tagged with multiple signals contributes to every line. Watch for one signal quietly bleeding while another carries the bottom line.
Book × signal CLV
Average closing-line value per book, broken down by every signal in the trail. Higher CLV = you got a better price than where the line closed. Tells you which book to shop for which kind of pick.
Edge proof — CLV vs results
Picks that beat the closing line should outperform — that's the math of edge. Real numbers, no spin.
Clean vs conflict picks
When two signals disagree on the same game, are we still winning? Split of every settled pick by whether an opposing pick existed at the same time.
CLV trend
Rolling 20-pick average closing-line value. Above zero = we're consistently beating the market — the single best long-term edge metric.
Cumulative units½ Kelly cohort
Book leaderboard
These are the books whose price moves we followed when picks fired. Top book is the sharpest.
Where the edge lives
Same picks, sliced by the variables that matter. Shows you where the model is actually paying — and where it isn't.
Which side of the line we win on.
Whether the picked team is hosting or traveling.
ROI by price band — heavy chalk vs live dogs.
Does the model's confidence score actually predict ROI? (Confidence model v2 — May 2026)
Bigger sharp moves should mean stronger signal.
Weekday-sharp vs weekend-public-money — does the day matter?
Top teams (5+ settled). Sorted by units.
Recent settled picks
Why you can trust these numbers
- • Every pick is locked at first pitch — recommendation, odds, and reasoning can't change after a game starts.
- • All settled picks are public, including premium ones, after the game ends. No hiding losers.
- • Graded at the recommended price (1 unit), so the units number is what you'd have done following along.
- • CLV is calculated against the line at the close of betting — the industry standard for measuring long-term edge.