Accuracy of Proximo AI predictions for the Premier League in the 23/24 season compared to academic and industry standards
This analysis compares the accuracy of the predictions of our 1×2 Proximo AI model with a recognized academic reference and basic random odds in various Premier League betting markets. For each market, we measure:
- Precision: The percentage of accurate predictions of our AI model
- Quality: Prediction reliability (High/Good/Medium/Poor)
- Trend: Do you want to know if it is better or worse or worse?
- Log Loss: A mathematical measure of the accuracy of predictions (lower is better)
- Academic Standard: Average accuracy from scientific papers
- Random chance: Expected accuracy from pure random guessing
The analysis covers a variety of betting markets, from basic match results to complex predictions, such as accurate results. Each market is evaluated in comparison with:
- Academic standards from leading sports analytical journals
- Historical average rates in the Premier League
- Random chances
In this way, we evaluate not only raw accuracy, but also how our predictions compare to scientific standards and pure randomness.
Academic Sources
- International Journal of Forecasting (2021)
- IEEE Access Journal (2023)
- Journal of Prediction Markets (2019)
Key betting markets
Over/under 1.5 goals
- Accuracy: 88%
- Quality: Bad
- Trend: No changes
- Log Loss: -0.3733
- Academic Standard: Poorly researched
- Historical frequency: ~88% of PL matches
Context: It is in line with the historical average and shows solid predictive power.
Home team scores a goal (over 0.5 goals)
- Accuracy: 85%
- Quality: Nice one
- Trend: Increasing
- Log Loss: -0.4603
- Academic Standard: 70-75%
- Historical frequency: ~75% of PL matches
Context: Exceeds academic standards and historical rates, demonstrating a strong predictive ability.
Over/under 2.5 goals
- Accuracy: 64%
- Quality: Nice one
- Trend: No changes
- Log Loss: -0.6869
- Academic standard: 51-54%
- Random chance: 50%
Context: It outperforms academic standards by 10-13%.
Final Score (1X2)
- Accuracy: 59%
- Quality: High
- Trend: Increasing
- Log Loss: -1.0813
- Academic standard: 53-56%
- Random Chance: 33.33%
Context: Almost twice as accurate as random odds, it exceeds academic standards.
Both teams score a goal
- Accuracy: 58%
- Quality: Bad
- Trend: Falling
- Log Loss: -0.6744
- Academic standard: 49-52%
- Historical frequency: ~52% of PL matches
Context: It exceeds academic standards by 6-9%, revealing a predictive advantage.
Which team scores first
- Accuracy: 65%
- Quality: High
- Trend: Increasing
- Log Loss: -0.8211
- Academic standard: 55-58%
- Random chance: 50%
Context: It outperforms academic standards by 7-10% in this high-risk market.
The result of the first half
- Accuracy: 46%
- Quality: High
- Trend: Increasing
- Log Loss: -1.1160
- Academic standard: 40-43%
- Random Chance: 33.33%
Context: It exceeds academic standards by 3-6% in an unpredictable market.
Halftime/End (HT/FT)
- Accuracy: 38%
- Quality: High
- Trend: Increasing
- Log Loss: -1.9955
- Academic standard: 25-30%
- Random Chance: 11.11%
Context: It significantly outperforms both academic standards and random odds in this complex market.
Correct result
- Accuracy: 22%
- Quality: High
- Trend: Increasing
- Log Loss: -3.2324
- Academic standard: 7-9%
- Random chance: 4-5%
Context: It achieves more than twice the score of academic standards and four times more accurate than random chances, demonstrating a strong predictive ability.
Conclusion
The analysis shows that our predictions for the Premier League often exceed academic standards, highlighting the strength of our predictive models in challenging markets.
