Between September 11 and September 13, 2026, the Dixon-Coles model tracked 56 picks across 34 soccer matches, posting a 33-23-0 record (58.9% win rate) for +8.42 net units. Total markets led the way with 17 wins on 28 picks for +6.54 units, while spread markets produced a 16-12-0 mark for +1.88 units. A three-day window provides a brief glimpse into performance rather than proof of a long-term edge, but auditing the numbers reveals where model probabilities diverged most sharply from market prices.

Longshot Hits in Totals and Spreads

The model found significant traction by targeting pricing discrepancies on underdogs and goal totals. In Serie A, Fiorentina was priced at a 4.350 decimal price on the -1.5 spread against Venezia, despite the model assigning only a 0.2671 probability to the cover. Fiorentina delivered a 4-2 road victory behind three goals from Franco Mastantuono, netting +3.35 units on that single call. Fiorentina also covered the Over 2.5 total (model probability 0.4398 at a 1.855 decimal price) for an additional +0.86 units.

Higher-priced totals yielded steady returns across several fixtures. The model flagged Under 2.5 in Liverpool's scoreless 0-0 draw against Fulham at a 2.930 decimal price against a 0.4012 model probability, banking +1.93 units alongside Fulham +1.5 on the spread (+0.70 units at a 1.699 price). In Serie A, Lazio and AC Milan played to a 2-2 draw, clearing the Over 2.5 line priced at 2.340 (0.3611 model probability) for +1.34 units, bolstered by two goals from Diego Moreira.

High-Probability Divergences and Missteps

Not all high-conviction positions landed seamlessly. In the Premier League, Hull City grabbed a 2-2 draw at Chelsea. The model evaluated Hull City +2.0 on the spread at a 0.9976 probability against a 1.746 market price, which paid out +0.75 units thanks to two goals from Bachir Belloumi, offsetting João Pedro's goal and assist for Chelsea. However, the accompanying total pick—Under 2.5 priced at 2.840 with a 0.8982 model probability—lost a full unit when four goals hit the back of the net.

Longshot spread swings also demonstrated the inherent risk of deep price chasing. The model assigned a 0.1272 probability to VfB Stuttgart covering -2.5 at an extreme 11.950 decimal price against 1899 Hoffenheim; Hoffenheim won 2-1, resulting in a 1.00 unit loss.

Market Results Breakdown

The table below outlines how the 56 graded picks settled across market types over the three-day window:

Market Picks Record Win % Net Units
Spread 28 16-12-0 57.1% +1.88
Total 28 17-11-0 60.7% +6.54
Total Overall 56 33-23-0 58.9% +8.42

In addition, individual performances helped shape key results: Can Uzun logged two goals and Jonathan Burkardt recorded a goal and assist in Eintracht Frankfurt's 3-1 win over FSV Mainz 05 (covering Over 2.5 at a 1.649 price against a 0.6607 model probability). Elsewhere, Roberto Férnandez scored twice in Espanyol's 2-0 win over Osasuna, covering both Espanyol +0.5 (+0.73 units at a 1.725 price) and Under 2.5 (+0.78 units at a 1.775 price).

Ultimately, three days of results are merely a small sample size. The +8.42 unit gain reflects a favorable short-term run across 56 selections, but evaluating true model performance requires monitoring these market edges over a much larger body of work.