Five distinct research markets
| Fixture | Original market | Research p | VS | Ref. odds | Final | Result |
|---|---|---|---|---|---|---|
| Santa Clara–Benfica | Under 4.5 | 89.6% | 34 | 1.16 | 1–2 | WIN |
| Copenhagen–Odense | Under 4.5 | 82.8% | 41 | 1.26 | 1–1 | WIN |
| Beşiktaş–Başakşehir | Under 4.5 | 82.0% | 56 | 1.36 | 2–0 | WIN |
| Sevilla–Osasuna | Over 1.5 | 76.0% | 34 | 1.43 | 2–0 | WIN |
| PSG–Lyon | Under 4.5 | 76.0% | 34 | 1.42 | 2–0 | WIN |
These were publicly archived, uncalibrated research probabilities, not authentic certified BET settlements. Results reviewed on 27 October; reference odds are not independently proven executable in Greece. Original public 26 October edition →
Research Doubler (separate): Santa Clara–Benfica U3.5 × Sevilla–Osasuna O1.5, reference odds 2.0306, both legs won. The Sevilla market shares an event/outcome with the five-row table; do not count it as extra independent evidence. Recorded dependence-naive scenario 57.99%; downside stress EV −11.1%, so HOLD was not disproven by the observed win.
Rejected-market counterfactuals
| Rejected market | Match score | Ex-post result | What it means |
|---|---|---|---|
| Benfica straight win | 1–2 | Would win | Review original market-disagreement evidence |
| Copenhagen straight win | 1–1 | Would lose | The original rejection was directionally supported |
| Beşiktaş straight win | 2–0 | Would win | Check original lineup/rotation uncertainty |
| Sevilla–Osasuna Over 2.5 | 2–0 | Would lose | Over 1.5 outperformed this rejected line |
| PSG straight win | 2–0 | Would win | Retrospective win cannot establish ex-ante positive EV |
Counterfactual wins are not placed bets and cannot be retroactively included in the Track Record.
Evidence → test → decision
| Observed pattern | Prospective test | Change today |
|---|---|---|
| Alternative totals succeeded | Compare pre-frozen line distributions and prices over ≥200 independent events | No coefficient change |
| Three rejected favourite wins | Check ex-ante price, model confidence and league-level reliability; target ≥100 per league before interpreting patterns | Preserve rejections |
| Missing canonical IDs / NO-BET market passports | Store match ID, market/line, model version, reasoning, timestamp and immutable forecast hash before each event | Pipeline repair required |
| Missing executable prices and CLV | Join rights-cleared timestamped book quotes and closing lines | ROI/CLV unavailable |
No model weights were changed. Brier and Log Loss refer to a five-event research worksheet, not an independently validated calibration population.
What this audit can confirm
Public pre-event research pages and independently reported final scores support a descriptive 5/5 market review. Missing canonical prematch IDs, signed market-specific settlement passports, execution prices and the 26 October private PRO Final Call ledger prevent official retrospective certification.
Result sources and technical boundary
Global backend 17-stage execution remains incomplete; this page is an editorial audit only. Authenticated historical Blob sync and official Track Record modification are not claimed.
Today's Daily CORESeparate PRO AuditOlder Full Audit archive