Headline numbers
294
Bets audited
16.3%
Disagreement rate (≥4pp)
43 bets
Sharp disagreements (≥7pp)
+3.71 pp
Mean disagreement
How to read this: "Pricing Disagreement" = the offered decimal was longer than our independent fair value, expressed as a percentage-point difference in implied probability. Positive value ⇒ the offer was soft vs. fair value (the book paid more than mathematically warranted, in real-money mode). Pinnacle closing-line coverage on this audit: 35.4% of bets had a Pinnacle benchmark.
Unweighted audit only. The bet log did not include a
stake_or_exposure column, so we cannot compute
stake-weighted metrics. To unlock handle at risk and
stake-weighted mean, include the optional stake column
in the next CSV drop — spec at /pilot/csv-spec.
By sport
| Sport | Bets | Mean disagreement | Flagged rate |
|---|---|---|---|
| Counter-Strike 2 | 192 | +3.64 pp | 15.1% |
| Dota 2 | 102 | +3.85 pp | 18.6% |
By market type
| Market | Bets | Mean disagreement | Flagged rate |
|---|---|---|---|
| moneyline | 106 | +1.22 pp | 17.0% |
| spread_-1.5 | 94 | +9.11 pp | 23.4% |
| total_2.5 | 80 | +0.91 pp | 10.0% |
| total_4.5 | 14 | +2.31 pp | 0.0% |
Top 10 individual pricing disagreements
Largest gaps between offered decimal and our independent fair value. Amber rows ≥4pp; red rows ≥7pp (sharp money would almost certainly pick these off).
| Sport | Match | Market | Side | Offered | Our fair | Disagreement | Stake | Pinnacle close | Outcome |
|---|---|---|---|---|---|---|---|---|---|
| Counter-Strike 2 | FarmVille vs regain | spread_-1.5 | team_b | 3.754 | 1.099 | +64.35 pp | — | 3.170 | win |
| Counter-Strike 2 | 100 Thieves vs TDK | spread_-1.5 | team_b | 4.516 | 1.173 | +63.11 pp | — | 4.110 | lose |
| Dota 2 | Inner Circle vs Team Stels | spread_-1.5 | team_b | 13.384 | 1.494 | +59.46 pp | — | 9.310 | lose |
| Counter-Strike 2 | Natus Vincere vs GamerLegion | spread_-1.5 | team_b | 7.305 | 1.397 | +57.89 pp | — | 6.990 | lose |
| Counter-Strike 2 | FaZe vs FURIA | spread_-1.5 | team_b | 2.160 | 1.047 | +49.21 pp | — | 1.952 | win |
| Counter-Strike 2 | Fake do Biru vs ShindeN | spread_-1.5 | team_b | 3.014 | 1.226 | +48.39 pp | — | 2.660 | win |
| Dota 2 | Power Rangers vs L1ga Team | spread_-1.5 | team_b | 3.461 | 1.301 | +47.97 pp | — | 3.400 | win |
| Counter-Strike 2 | Passion UA vs FOKUS | spread_-1.5 | team_b | 2.739 | 1.216 | +45.73 pp | — | 2.720 | win |
| Dota 2 | Power Rangers vs South America Rejects | spread_-1.5 | team_b | 2.289 | 1.142 | +43.88 pp | — | 2.120 | win |
| Counter-Strike 2 | MANA eSports vs ASTRAL | spread_-1.5 | team_b | 2.998 | 1.333 | +41.66 pp | — | 2.850 | win |
What we recommend reviewing
- This is a sample report. The recommendations below are templated against simulated lines, not real operator data. In an active pilot, each recommendation is generated from your actual settled-bet log and references the specific markets, tournaments, and stake bands your trading desk owns.
- spread_-1.5 markets show +9.11 pp mean pricing disagreement across 94 bets. Require manual review on this market type when our fair value disagrees by >7pp and Pinnacle close is unavailable.
- 43 bets ≥7pp pricing disagreement (sharp threshold). Suspend stale pre-match lines if our fair value moves more than 4pp between the time you posted and 30 minutes before kick-off, and your line has not updated in that window.
- Pinnacle close coverage was only 35.4% this week. Treat the pricing disagreement numbers as model-vs-model where the Pinnacle column is —. We are working to expand benchmark coverage; see /pilot/methodology for our resilience plan.
- Trend, not snapshot. Compare this week's mean pricing disagreement against the prior 3 weeks at day 30. A downward trend is the real value of the pilot — if your prices tightened during the audit window, the reports moved the needle.