Weekly Predictions Review: Serie A Successes and Notable Upsets

This past week, from May 3rd to May 9th, our models were put to the test across various global football leagues. Out of 15 featured predictions, 10 proved correct, resulting in an overall accuracy rate of 66.7% . This review delves into the key successes, analyzes the notable misses, and examines the relationship between our confidence ratings and the final outcomes from a week of compelling fixtures.

Model Performance Highlights

Our models demonstrated particular strength in Italy's top flight. In Serie A Serie A, the prediction for Inter to secure an away victory against Lazio was validated with a convincing 3-0 result. This outcome was anticipated with 68% confidence , correctly identifying Inter's superiority despite being the visiting team, as the full match recap shows on our results page from May 9th, 2026.

Continuing the trend in Italy, the model also correctly called a dominant home performance for AS Roma AS Roma against Fiorentina Fiorentina. The match ended in a 4-0 victory for the hosts, a result predicted with 69% confidence . This success, detailed in our AS Roma vs Fiorentina preview, showcases the model's ability to accurately weigh team form and historical performance in high-stakes European encounters.

Elsewhere, the model's capacity to identify significant mismatches was on full display in the Latvian Virsliga Virsliga. We predicted a home win for Auda against Grobiņa, which materialized in a staggering 7-0 victory. While our confidence was a more moderate 62% , the underlying data clearly pointed towards a one-sided affair, which the on-pitch action confirmed in dramatic fashion.

When the Numbers Missed: Analyzing the Upsets

Our analytical approach, which utilizes a Dixon-Coles Poisson model, team-specific Elo ratings, and continuous calibration against live betting markets, provides a robust statistical probability for match outcomes. However, football's inherent unpredictability ensures that outlier results are always a possibility, and this past week featured several notable upsets that defied the model's expectations.

The most significant deviation from our forecast occurred in the Swiss 1. Liga Classic. The model predicted an away win for Wettswil-Bonstetten over Winterthur II Winterthur II with 69% confidence . In a complete reversal, the home side produced a dominant performance, securing a 5-0 victory. This result is a clear example of how specific match-day conditions and team motivation can produce outcomes that statistical models find difficult to anticipate.

A high-confidence prediction also went astray in Portugal's Primeira Liga Primeira Liga. We assigned 74% confidence —our second-highest of the week—to a home victory for SC Braga SC Braga against Estoril Estoril. Despite the strong backing from our model, Estoril held the favorites to a 1-1 draw. This incorrect forecast, which can be revisited at SC Braga vs Estoril, illustrates that even with a significant statistical advantage, the possibility of a draw can disrupt the most likely outcomes.

Confidence Ratings and Correct Outcomes

The model's highest conviction of the week was placed on the J2/J3 League J2/J3 League fixture between Vegalta Sendai Vegalta Sendai and Tochigi SC Tochigi SC. A home win was forecasted with 77% confidence , and the prediction proved accurate as Vegalta Sendai emerged with a 2-1 victory. This result highlights the general trend where higher confidence ratings from our model correlate with more reliable and predictable match outcomes.

This pattern was further reinforced by other correct predictions with high confidence scores. In Germany's 3. Liga 3. Liga, a high-scoring 5-3 victory for Hansa Rostock Hansa Rostock over Stuttgart II Stuttgart II was correctly projected with 70% confidence . A similar level of conviction ( 71% confidence ) successfully identified the 3-0 away win for Səbail Səbail against Zaqatala in the Azerbaijan Birinci Dasta Birinci Dasta league.

In contrast, correct predictions with more moderate confidence ratings often pointed to closer contests. The model correctly forecasted Bayern München's 1-0 away victory over VfL Wolfsburg VfL Wolfsburg in the German Bundesliga Bundesliga, but with a confidence level of 62% . This reflected a matchup with a narrower expected margin. Notably, three of the week's five incorrect predictions fell in the 62-69% confidence range, suggesting that outcomes in this tier carry a greater degree of uncertainty.

Frequently Asked Questions

Q: What was the overall prediction