Model documentation & methodology

How the model works

We compute probabilities using a Dixon-Coles bivariate Poisson model. We publish every expected-goals figure, source rate, and data quality check behind every evaluation.

01

Recency-weighted goal rates

Every match evaluation begins with team-level scoring and conceding rates (λ for home expected goals, μ for away expected goals).

Early in a season, single-match samples are volatile. We use a recency-weighted blend against prior season data:

Goal Rate Calculation Example:
Bournemouth 1.47 goals/game at home
1 match in 2026/27, weighted 10% against 2025/26 (1.53)

As the current season progresses, the weight automatically shifts from prior season stats to current season matches.

02

The Dixon-Coles Scoreline Engine

Standard independent Poisson models underestimate low-scoring outcomes (such as 0–0 and 1–0 draws/wins). The Dixon-Coles model introduces a low-score correlation adjustment parameters (τ) to solve this structural bias.

A single Dixon-Coles solver pass generates a 10×10 scoreline probability matrix for the fixture. From this single matrix, all supported market probabilities are derived:

  • 1X2: Sum of Home Win, Draw, and Away Win cells.
  • Over/Under (1.5, 2.5, 3.5): Sum of scorelines where combined goals exceed or fall below the line.
  • Both Teams To Score (BTTS): Sum of scorelines where both home and away goals > 0.
  • Double Chance: Combined 1X, X2, 12 probabilities.
03

Data Quality Audit & Component Rules

Before any match is assigned a signal tier, three underlying datasets are audited:

1. Recent Form
Min 3 matches per team
2. Head-to-Head
Min 3 past meetings
3. Home/Away Split
Min 3 home matches played

If any component falls short of minimum match thresholds, the fixture is flagged as Blocking or Not enough data rather than outputting misleading estimates.

04

Empirical Calibration & Per-Market Confidence Floors

We do not group all betting markets into single shared buckets. Because Over 1.5 Goals has a high natural occurrence (~77%) while Draw has a low natural occurrence (~25%), pooling markets distorts decile bands.

Below is our out-of-sample calibration audit evaluated strictly PER MARKET across 727 completed top-flight fixtures (2026 season). Any decile band with fewer than 50 fixtures is flagged as Insufficient sample (<50) rather than displaying a percentage a reader should not trust:

MarketPredicted BandSample (N)Model MeanObserved Hit RateDeltaHeadline Status
Over/Under 2.550–60%49853.7%50.8% (253/498)-2.9%Suppressed (<60%)
Over/Under 2.560–70%19263.5%66.7% (128/192)+3.2%Eligible (≥60% Floor)
Over/Under 2.570–80%3273.1%75.0% (24/32)+1.9%Small sample (N=32)
Both Teams To Score50–60%48054.4%52.9% (254/480)-1.4%Suppressed (<60%)
Both Teams To Score60–70%22463.7%62.9% (141/224)-0.8%Eligible (≥60% Floor)
Over/Under 1.560–70%4667.3%84.8% (39/46)+17.5%Suppressed (High Baseline Noise · N=46)
Over/Under 1.570–80%40875.3%77.2% (315/408)+1.9%Eligible (≥70% Floor)
1X2 (Home Win)50–60%9153.5%56.0% (51/91)+2.5%Suppressed (<60%)
1X2 (Home Win)60–70%2464.2%45.8% (11/24)-18.3%Display-Only Excluded (N=24)
Over/Under 3.560–70%23164.9%62.3% (144/231)-2.5%Display-Only Excluded
Over/Under 3.570–80%33273.3%66.9% (222/332)-6.5%Display-Only Excluded
Team Totals 1.560–70%54665.0%59.7% (326/546)-5.3%Display-Only Excluded
Team Totals 1.570–80%13473.7%67.9% (91/134)-5.8%Display-Only Excluded

Our Evidence-Based Selection Rules

  • 1. Small Sample Protection (N < 50 Threshold): Any decile band with fewer than 50 recorded match outcomes is marked as Insufficient sample (<50). We never present un-statistically significant small-sample frequencies (like 4/5 or 4/4) as evidence.
  • 2. 1X2 Moved to Display-Only: Top-flight match-winner predictions above 60% are rare (N=24 in 60-70%) and suffer from high draw/upset variance (-18.3% delta). 1X2 is strictly display-only on /fixture/[id] until sample volume builds.
  • 3. Over/Under 3.5 & Team Totals Exclusion: Over/Under 3.5 displays a -6.5% overconfidence bias in the 70–80% band, while Team Totals runs -5.3% to -5.8% overconfident and lost to closing lines in backtests. Both are reserved for display-only.
  • 4. Double Chance Composite Exclusion: Double Chance (1X, X2, 12) is a derivative composite of 1X2 with an inherently high base rate (~70–80%). It is kept display-only on detail views to avoid diluting primary single-outcome picks with derivative combinations.
  • 5. Core Primary Market Reliability: Core primary markets (Over/Under 2.5 and BTTS) calibrate closely within -0.8% to +3.2% of predicted probabilities in the 60–70% band with large sample volume (N=192 and N=224).
Transition to Live Ledger Settlement

In addition to our 727-fixture out-of-sample backtest, all generated predictions are stored in our Postgres database (predictions table). When live post-fixture settlements reach N ≥ 50 settled match predictions per decile band, the published calibration table dynamically updates to reflect live real-world performance from our public ledger.

05

What we don’t claim

No market in our model has passed Gate 2 calibration. We do not beat the bookmaker’s closing line on any market at any confidence level we can demonstrate.

Grade labels (Strong signal, Mixed signal, Contradictory signal) reflect statistical consistency across inputs — not an assertion of financial edge or guaranteed winning picks.

Every published selection is logged to our immutable public ledger before kickoff and settled automatically when results complete.