Model Audit: October 29-31 Soccer Results and Performance

Across a three-day evaluation window from October 29 through October 31, 2026, the dixon_coles_v1 soccer model generated 44 picks across 28 games, finishing with an 18-24-2 record and a net loss of -3.42 units. A 42.9 percent win rate over a short fixture slate is noise rather than signal; three days of output cannot validate or invalidate a quantitative architecture. The divergence between spread performance and totals performance provides the primary analytical backdrop for the period.

Market Performance and High-Probability Divergence

Spreads drove the overall deficit, producing a 6-12-2 record across 20 selections for -6.34 units. Totals delivered positive drift, posting a 12-12-0 record over 24 picks for +2.92 units.

Market	Picks	Record (W-L-P)	Units	Win Rate
Spread	20	6-12-2	-6.34	30.0%
Total	24	12-12-0	+2.92	50.0%
Overall	44	18-24-2	-3.42	42.9%

The model's highest-probability calls yielded mixed results when evaluated alongside decimal market prices:

• Estac Troyes 2, Le Mans 0: The model assigned an 0.8117 probability to Le Mans +0.5 at a 1.709 decimal price, but Troyes claimed a 2-0 home victory, taking a -1.0 unit loss.
• FC Schalke 04 0, RB Leipzig 2: The model posted an 0.7992 probability on under 2.5 goals at a 2.820 decimal price, returning +1.82 units.
• SV Elversberg 4, FSV Mainz 05 1: The model hit both selections, backing over 2.5 goals at an 0.7598 probability (1.571 decimal price, +0.57 units) and SV Elversberg +0.5 spread at an 0.7382 probability (1.800 decimal price, +0.80 units).
• Hull City 2, Ipswich 2: Hull City 0.0 spread carried an 0.7510 model probability at a 1.962 decimal price, resulting in a push at 0.0 units.

High-Scoring Matches and Standout Performances

In Frosinone's 5-1 win over Lecce on October 29, the attack generated heavy box activity. Frosinone got goals from Giacomo Calò (2 shots on target), Antonio Raimondo (1 shot on target), and Matteo Chichella (1 shot on target). Playmaking came from Giorgi Kvernadze (2 assists, 1 shot on target), Romano Schmid (1 assist), and Farés Ghedjemis (1 assist, 1 shot on target). Joel Monteiro scored Lecce's single goal with 2 shots on target. The model won on Frosinone -0.5 spread (0.6348 model probability, 1.909 decimal price, +0.91 units) but dropped under 2.5 goals (0.4364 model probability, 2.480 decimal price, -1.0 unit).

Fiorentina and Atalanta played to a 2-2 draw on October 29. Fiorentina's goals came from Alieu Njie (1 goal, 1 assist, 1 shot on target) and João Mário Lopes (1 goal, 2 shots on target), with Álex Jiménez adding an assist. Atalanta countered through Charles De Ketelaere (1 goal, 1 assist, 2 shots on target) and Nicola Zalewski (1 goal, 1 assist, 1 shot on target). The result produced a push on Atalanta 0.0 spread (0.6576 model probability, 2.240 decimal price, 0.0 units) and a loss on under 2.5 goals (0.4791 model probability, 2.200 decimal price, -1.0 unit).

Long-Odds Selections and Sample Size Context

The model occasionally identified value on long-odds under totals:

• Bayern München 2, Borussia Dortmund 2: Under 2.5 goals held an 0.3398 model probability at a 4.630 decimal price, resulting in a -1.0 unit loss.
• Barcelona 6, Alaves 1: Under 2.5 goals sat at an 0.2246 model probability against a 4.480 decimal price, taking a -1.0 unit loss.
• Manchester City 0, Brighton 1: Under 2.5 goals carried an 0.3991 model probability at a 2.730 decimal price, generating +1.73 units. Brighton +0.5 spread (0.4994 model probability, 2.150 decimal price) also won, yielding +1.15 units.

A 44-pick window across three calendar days represents a narrow slice of performance. Reporting a net -3.42 unit outcome reflects what transpired in this specific set of fixtures without establishing a long-term statistical trend.