Ligue 1 AI Predictions Audit: Regular Season Week 22 Performance
Devstral Small and Llama 3.2 3B led Ligue 1 predictions this week with 3.00 points per match, followed by Phi-4 (2.75). Models achieved 40.68% correct tendency overall, though the Rennes 3-1 victory over Paris Saint Germain, predicted correctly by only 3.3% of models, was the biggest surprise.
Devstral Small and Llama 3.2 3B led Ligue 1 predictions this week with 3.00 points per match, followed by Phi-4 (2.75). Models achieved 40.68% correct tendency overall, though the Rennes 3-1 victory over Paris Saint Germain, predicted correctly by only 3.3% of models, was the biggest surprise. This audit covers the 9 matches of Ligue 1 Regular Season - 22. AI prediction accuracy is critical for assessing model reliability in a competitive league setting. The following data provides a statistical breakdown of this week's performance.
Top 10 Models
| # | Model | Matches | Total Points | Avg Pts/Match | Tendency % | Exact % |
|---|---|---|---|---|---|---|
| 1 | Devstral Small (OpenRouter) | 7 | 21 | 3.00 | 71.4% | 28.6% |
| 2 | Llama 3.2 3B (OpenRouter) | 4 | 12 | 3.00 | 50.0% | 0.0% |
| 3 | Phi-4 (OpenRouter) | 8 | 22 | 2.75 | 75.0% | 12.5% |
| 4 | DeepSeek R1-0528 (OpenRouter) | 7 | 17 | 2.43 | 57.1% | 14.3% |
| 5 | RNJ-1 Instruct (OpenRouter) | 5 | 12 | 2.40 | 40.0% | 20.0% |
| 6 | Nemotron 3 Nano 30B A3B (OpenRouter) | 4 | 9 | 2.25 | 50.0% | 0.0% |
| 7 | Mistral Small 3 24B (OpenRouter) | 5 | 11 | 2.20 | 60.0% | 20.0% |
| 8 | DeepSeek V3.1 (OpenRouter) | 5 | 11 | 2.20 | 60.0% | 20.0% |
| 9 | Qwen3 235B Thinking (OpenRouter) | 7 | 15 | 2.14 | 57.1% | 14.3% |
| 10 | Qwen3 30B A3B (OpenRouter) | 7 | 15 | 2.14 | 57.1% | 14.3% |
Match-by-Match Audit
- Lyon vs Nice (2-0): 23 models. Correct tendency: 69.6% (16/23). Exact score hits: 13.0% (3/23). Predicted outcomes (H/D/A): 16/7/0. Consensus: H (69.6%), correct: yes.
- Metz vs Auxerre (1-3): 23 models. Correct tendency: 21.7% (5/23). Exact score hits: 0.0% (0/23). Predicted outcomes (H/D/A): 4/14/5. Consensus: D (60.9%), correct: no.
- Lorient vs Angers (2-0): 24 models. Correct tendency: 16.7% (4/24). Exact score hits: 0.0% (0/24). Predicted outcomes (H/D/A): 4/18/2. Consensus: D (75.0%), correct: no.
- Le Havre vs Toulouse (2-1): 21 models. Correct tendency: 0.0% (0/21). Exact score hits: 0.0% (0/21). Predicted outcomes (H/D/A): 0/16/5. Consensus: D (76.2%), correct: no.
- Paris FC vs Lens (0-5): 24 models. Correct tendency: 87.5% (21/24). Exact score hits: 0.0% (0/24). Predicted outcomes (H/D/A): 0/3/21. Consensus: A (87.5%), correct: yes.
- Lille vs Stade Brestois 29 (1-1): 26 models. Correct tendency: 73.1% (19/26). Exact score hits: 69.2% (18/26). Predicted outcomes (H/D/A): 4/19/3. Consensus: D (73.1%), correct: yes.
- Marseille vs Strasbourg (2-2): 24 models. Correct tendency: 66.7% (16/24). Exact score hits: 0.0% (0/24). Predicted outcomes (H/D/A): 8/16/0. Consensus: D (66.7%), correct: yes.
- Monaco vs Nantes (3-1): 29 models. Correct tendency: 27.6% (8/29). Exact score hits: 3.4% (1/29). Predicted outcomes (H/D/A): 8/17/4. Consensus: D (58.6%), correct: no.
- Rennes vs Paris Saint Germain (3-1): 30 models. Correct tendency: 3.3% (1/30). Exact score hits: 0.0% (0/30). Predicted outcomes (H/D/A): 1/7/22. Consensus: A (73.3%), correct: no.
Biggest Consensus Misses
- Le Havre vs Toulouse (2-1) | Consensus: D (76.2%) | Counts H/D/A: 0/16/5
- Lorient vs Angers (2-0) | Consensus: D (75.0%) | Counts H/D/A: 4/18/2
- Rennes vs Paris Saint Germain (3-1) | Consensus: A (73.3%) | Counts H/D/A: 1/7/22
- Metz vs Auxerre (1-3) | Consensus: D (60.9%) | Counts H/D/A: 4/14/5
- Monaco vs Nantes (3-1) | Consensus: D (58.6%) | Counts H/D/A: 8/17/4
Methodology
kroam.xyz uses a quota-based scoring system that rewards both accuracy and boldness:
Tendency Points (2-6 points): Models earn points for correctly predicting the match outcome (home win, draw, or away win). The points awarded depend on prediction rarity—if most models predicted a home win but the away team won, models who correctly predicted the away win earn more points (up to 6). Common predictions earn fewer points (minimum 2).
Goal Difference Bonus (+1 point): If the model predicts the correct goal difference (e.g., predicted 2-1 and result was 3-2, both +1 difference), they earn a bonus point.
Exact Score Bonus (+3 points): Predicting the exact final score earns 3 additional points.
Maximum: 10 points per prediction (6 tendency + 1 goal diff + 3 exact).
This system ensures that models taking calculated risks on unlikely outcomes are rewarded when correct, while also recognizing precision in exact score predictions. Learn more about our methodology.
Frequently Asked Questions
Q: Which AI model performed best in Ligue 1 Regular Season - 22? A: Devstral Small (OpenRouter) and Llama 3.2 3B (OpenRouter) were the top performers, both with an average of 3.00 points per match.
Q: How accurate were AI predictions for Ligue 1 this round? A: The overall correct tendency rate was 40.68%, while the exact score hit rate was 9.52% across 224 total predictions.
Q: What was the biggest upset in Ligue 1 Regular Season - 22? A: The Rennes 3-1 victory over Paris Saint Germain was the biggest consensus miss, with 73.3% of models predicting an away win.
Q: How does kroam.xyz score AI football predictions? A: kroam.xyz uses a quota-based system awarding up to 10 points per match: tendency points (2-6), goal difference bonus (+1), and exact score bonus (+3).
Generation cost: $0.0024
Tokens: 4,916 input + 2,159 output
Frequently Asked Questions
What is this article about?
Which AI model performed best in Ligue 1 Regular Season - 22?**?
Q: Which AI model performed best in Ligue 1 Regular Season - 22?
Q: How accurate were AI predictions for Ligue 1 this round?
You might also like
Ligue 1 AI Predictions Audit - Regular Season Week 24 Analysis
MiniMax M2.1 led with 2.78 avg points/match, followed by GLM-5 (2.44) and GPT-OSS 20B (2.22). Models achieved 50.19% correct tendency overall. The Paris FC vs Nice upset (1-0 result vs 78.9% consensus draw) was the biggest surprise.
Mar 2, 2026
Ligue 1 Week 23 AI Model Performance Audit
GLM-4.7 led Ligue 1 predictions with 3.22 points per match, followed by Qwen3 30B A3B (2.67) and MiniMax M2.5 (1.89). Models achieved 34.50% correct tendency overall, with Nantes' 2-0 win over Le Havre being the biggest consensus miss.
Feb 23, 2026
UEFA Conference League Round of 32 AI Prediction Audit
GPT-OSS 20B led UEFA Conference League predictions with 2.88 points per match, followed by Trinity Large Preview (2.63) and GLM-5 (2.25). Models achieved 38.16% correct tendency overall, with Fiorentina vs Jagiellonia (2-4) as the biggest upset.
Mar 2, 2026