Transparency
Talacote model accuracy: public track record
Every prediction is recorded before kick-off and never changed afterwards. This page compares it with the actual result and with bookmaker odds on the same matches. Updated automatically.
Across 487 evaluated matches, the outcome the model rated most likely happened 53.8% of the time, with a Brier score of 0.588 (random: 0.667).
Model vs bookmakers
On the 312 matches whose odds were collected before kick-off. For Brier and RPS, lower is better.
| Metric | Talacote model | Bookmakers (no margin) |
|---|---|---|
| Most likely outcome correct | 51.9% | 55.4% |
| Brier score | 0.593 | 0.565 |
| RPS | 0.209 | 0.197 |
On these matches the bookmakers remain more accurate than the model. That is expected: their odds include information the model does not use (line-ups, injuries, market moves).
Calibration
When the model gives an outcome about X% probability, does it happen about X% of the time? Each match counts three times (home win, draw, away win).
| Stated probability | Observed frequency | Cases |
|---|---|---|
| 0–10 % (6.2%) | 9.1% | 22 |
| 10–20 % (16.0%) | 12.6% | 151 |
| 20–30 % (25.7%) | 24.5% | 515 |
| 30–40 % (33.7%) | 30.8% | 399 |
| 40–50 % (44.5%) | 51.1% | 184 |
| 50–60 % (54.4%) | 59.3% | 113 |
| 60–70 % (63.8%) | 72.1% | 61 |
| 70–80 % (75.0%) | 71.4% | 14 |
| 80–90 % (81.8%) | 100.0% | 2 |
By competition
| Competition | Matches | Most likely outcome correct | Brier score | Model / bookmakers (same matches) |
|---|---|---|---|---|
| La Liga | 120 | 55.8% | 0.593 | 0.606 / 0.563 59 matches |
| FIFA World Cup | 104 | 60.6% | 0.523 | 0.523 / 0.471 104 matches |
| Premier League | 90 | 48.9% | 0.616 | 0.636 / 0.658 59 matches |
| Ligue 1 | 74 | 44.6% | 0.650 | 0.729 / 0.738 24 matches |
| Serie A | 40 | 55.0% | 0.596 | 0.613 / 0.625 27 matches |
| Bundesliga | 36 | 50.0% | 0.608 | 0.655 / 0.597 24 matches |
| UEFA Champions League | 23 | 65.2% | 0.510 | 0.512 / 0.414 15 matches |
Goal markets
| Metric | Matches | Accuracy | Brier score |
|---|---|---|---|
| Over 2.5 goals | 487 | 57.1% | 0.250 |
| Both teams to score | 487 | 56.9% | 0.245 |
Latest evaluated matches
| Match | 1-X-2 prediction | Score | ✓ |
|---|---|---|---|
| Valencia CF – Real Sociedad de Fútbol La Liga · 20 September 2026 | 33 · 32 · 34 | 2-3 | ✓ |
| AC Milan – US Lecce Serie A · 20 September 2026 | 55 · 26 · 18 | 3-0 | ✓ |
| Olympique de Marseille – Paris Saint-Germain FC Ligue 1 · 20 September 2026 | 42 · 28 · 30 | 1-2 | ✗ |
| SC Paderborn 07 – TSG 1899 Hoffenheim Bundesliga · 20 September 2026 | 30 · 30 · 40 | 3-1 | ✗ |
| Villarreal CF – Levante UD La Liga · 20 September 2026 | 45 · 27 · 28 | 3-1 | ✓ |
| RC Deportivo La Coruña – Real Betis Balompié La Liga · 20 September 2026 | 40 · 31 · 29 | 1-1 | ✗ |
| Juventus FC – Atalanta BC Serie A · 20 September 2026 | 46 · 30 · 24 | 2-0 | ✓ |
| FC Schalke 04 – SV 07 Elversberg Bundesliga · 20 September 2026 | 37 · 26 · 37 | 0-0 | ✗ |
| Fulham FC – Manchester United FC Premier League · 20 September 2026 | 34 · 28 · 37 | 1-1 | ✗ |
| OGC Nice – Lille OSC Ligue 1 · 20 September 2026 | 10 · 32 · 57 | 2-1 | ✗ |
| Club Atlético de Madrid – Real Madrid CF La Liga · 20 September 2026 | 39 · 26 · 35 | 2-1 | ✓ |
| Bayer 04 Leverkusen – RB Leipzig Bundesliga · 20 September 2026 | 36 · 25 · 40 | 2-0 | ✗ |
| Parma Calcio 1913 – Genoa CFC Serie A · 20 September 2026 | 45 · 32 · 23 | 2-1 | ✓ |
| Frosinone Calcio – Como 1907 Serie A · 20 September 2026 | 36 · 29 · 35 | 2-0 | ✓ |
| AFC Bournemouth – Liverpool FC Premier League · 20 September 2026 | 33 · 30 · 37 | 0-1 | ✓ |
| Manchester City FC – Sunderland AFC Premier League · 20 September 2026 | 64 · 26 · 10 | 5-3 | ✓ |
| Leeds United FC – Crystal Palace FC Premier League · 20 September 2026 | 66 · 21 · 13 | 0-0 | ✗ |
| AJ Auxerre – Stade Brestois 29 Ligue 1 · 20 September 2026 | 29 · 26 · 45 | 2-1 | ✗ |
| Getafe CF – Málaga CF La Liga · 20 September 2026 | 46 · 34 · 20 | 1-0 | ✓ |
| ACF Fiorentina – SSC Napoli Serie A · 20 September 2026 | 26 · 26 · 48 | 1-1 | ✗ |
How it is calculated
Brier score: sum of squared differences between the three stated probabilities and the result (0 = perfect, 0.667 = random).
RPS (Ranked Probability Score): same idea, taking the order home win, draw, away win into account; it is the reference metric in football.
Excluded matches: 17 with early-season default values and 27 whose prediction had been replaced by market odds for lack of data. They do not reflect the model.
The model has evolved over time. Figures use the predictions exactly as they were published before each match.
Past results do not guarantee future results. 18+ only. Please gamble responsibly.