Public performance
Evidence before claims.
Clauseground does not currently claim a proven historical edge. This page states the publication rules now, so the forward record cannot be rewritten around later results.
Current clean-forward status
Insufficient verified data for a headline result
The forward record is live and accumulating. 16 predictions have been scored so far, and 104 more are logged and waiting on results. The scored scoreboard (hit rate, ROI, Brier, calibration) publishes automatically once at least 50 predictions have been scored — below that, the confidence intervals span nearly the full range, so a headline would present noise as a result. Nothing here is simulated or back-filled.
Existing World Cup operational and development records are excluded from headline performance — they may include testing, retro-simulation, or changing model versions that do not meet the forward-publication rules below.
What is logged so far
Every pending prediction, by competition — these reconcile to the 104 total above.
Recorded before kickoff
Next predictions waiting on results
A live sample of upcoming logged predictions — our estimated probability against the market’s implied price at the moment we recorded it. When each match finishes, the outcome is scored against the row exactly as shown.
| Match | Pick | Our estimate | Market | Edge | Kicks off |
|---|---|---|---|---|---|
New York City FC v Toronto FC Major League Soccer | Both teams to score: Yes Both teams to score · 1.73 | 58.7% | 57.8% | +0.9pp | Jul 31 |
Puebla v Guadalajara Liga MX | Over 2.5 Totals · 1.84 | 55.4% | 54.3% | +1.0pp | Aug 1 |
Atlético San Luis v Tijuana Liga MX | Over 2.5 Totals · 1.80 | 56.0% | 55.6% | +0.5pp | Aug 1 |
FC Juárez v Pumas Liga MX | Pumas Match result · 2.91 | 35.5% | 34.4% | +1.1pp | Aug 1 |
Querétaro v Tigres Liga MX | Over 2.5 Totals · 1.91 | 54.0% | 52.4% | +1.7pp | Aug 1 |
Inter Miami CF v Columbus Crew SC Major League Soccer | Over 3.5 Totals · 1.93 | 52.4% | 51.8% | +0.6pp | Aug 1 |
Probabilities are shown to one decimal; edge is the gap in points (our estimate − implied). Prices are the best decimal odds recorded at publication. This is a sample — the full ledger is scored once results are in.
What will qualify
- 01A prediction must be recorded before kickoff with its timestamp and model version.
- 02The price and source used for comparison must be identifiable.
- 03Voids, corrections, missing closes, and exclusions must remain visible.
- 04Results must be shown for the full eligible sample, not selected positive periods.
- 05Competition, market, model-version, and time-period breakdowns must reconcile to the total.
Metric definitions
- Published predictions
- The count of eligible, timestamped predictions in the selected view. Sample size appears beside every aggregate.
- Brier score
- A probability-error score. Lower is better, but it is meaningful only relative to the market, a baseline, and the same sample.
- Calibration
- Whether outcomes forecast near a given probability occur at roughly that rate over a sufficiently large sample.
- Closing line value
- A comparison between the recorded publication price and the defined closing reference price, with missing closes reported.
- Return on investment
- Settled profit or loss divided by eligible stake under the published staking rule. It is volatile and not a forecast of future return.
- Exclusions and corrections
- Every excluded or corrected record needs a reason, timestamp, and unchanged audit trail in the public methodology.
What is not claimed
Historical performance does not guarantee future results. Positive CLV does not guarantee profit. A calibrated model can still lose frequently. Until a valid sample exists, “insufficient data” is the only responsible headline.
Read the full methodology for the current model and data limitations.
Decision support, not betting advice. No guaranteed outcomes. 18+/21+ where applicable. Gamble responsibly.