Home / Track record

Every signal we have
ever published. Graded.

This page updates after every slate. Wins, losses, pushes, and retractions, with closing line value on every entry.

312
Signals graded
all timestamped pre tip
+2.7%
Average CLV
95% CI · +1.9% to +3.5%
74%
Beat the closing line
231 of 312 entries
9
Retractions published
gate caught, shown in public

Two honest notes before you read this. First, short samples prove nothing in either direction, which is why we publish a confidence interval next to our CLV. Second, the number that matters most here is CLV, not the record. Beating the closing line consistently is the only statistical fingerprint of skill. Records are what variance does while you wait.

DateSignalTierEntryCloseCLVResult
Apr 25OKC −9.5A−9.5−11+3.4%Win
Apr 25CLE Under 7.5A7.57.0+4.1%Win
Apr 24ATL +1.5B+1.5+1.0+2.2%Loss
Apr 24HOU Under 141.5B141.5141.5+0.0%Push
Apr 23NYY −1.5A−1.5−1.5+1.8%Win
Apr 23BOS Over 8.5B8.59.0−1.1%Loss
Apr 22DEN −3.5A−3.5−4.5+3.9%Win
Apr 22LAD −1.5B−1.5−2.0+2.6%Win
Apr 21MIA Over 213.5B213.5213.0−0.7%Loss
Apr 21SEA Under 7.0A7.06.5+4.3%Win

Signals our gate pulled after publication, with the reason. We list these because a track record with no retractions is a track record you should not trust.

DateSignalReason for retractionWindow
Apr 18PHI −2.5Starter scratched after publication. Lineup gate failed on re check.14 min
Apr 11TOR Over 9.0Stale weather feed. Wind reversed the total read once corrected.22 min
Mar 29MIL −5.5Bullpen availability changed. Conviction no longer cleared the gate.9 min
We close the loop

We show up after the loss.

Anyone can celebrate a winner. The honest test is what you do after a loss. Every losing signal gets a public post-mortem: what happened, what we got right, what we underweighted, and the lesson, which becomes a free Daily Hoot.

Last night's reconciliation a real loss, dissected in public
LossATL +1.5NBA · Apr 24 · Tier B
+1.5
Entry
+1.0
Close
+2.2%
CLV
ATL −4
Result

The reframe: we lost the game and still beat the close (+2.2%). Right process, wrong night. That gap is the thing one result can never tell you and a season of CLV always does.

What happened: Atlanta lost outright by 4. The hook never mattered; the game slipped away in the final three minutes.

What we got right: the pace and rest reads held — this was a one-possession game into the fourth, exactly the range we modeled.

What we underweighted: Atlanta's half-court offense with Young short of full speed. Late-clock possessions stalled, and the cover went with them.

The lesson

A spread built on pace assumes both offenses still execute late. When one can't, the number lies.

Are our probabilities honest?

When we say 70%,
it hits 71%.

A reliability curve compares the conviction we stated against what actually happened. The closer our line sits to the diagonal, the more honest our probabilities are. Almost nobody in this category will show you this chart.

perfect Predicted conviction Actual hit rate
Calibrated to
1.4% error.

Across 312 graded signals, our stated conviction and real-world hit rate track almost perfectly. This is the difference between a model with genuine skill and a tout with a hot streak.

1.4%
Calibration error
312
Graded sample
Tier A
We say 88% →86%
Highest conviction. Our most selective tier.
Tier B
We say 76% →77%
Solid edge, slightly more variance.
Tier C
We say 64% →63%
Lean reads. Honest about the uncertainty.

Responsible use & disclaimer

Owl & Odds is an analytics and entertainment product, not betting advice. We never accept wagers. Must be 19+ in Ontario or of legal age in your jurisdiction. Past signal performance does not guarantee future results. Sample sizes shown are small and not predictive on their own. If gambling stops being entertainment, help is available: ConnexOntario 1 866 531 2600. Owl & Odds operates to iGaming Ontario / AGCO standards.