← Back to Learn

Soccer Betting Analytics: A Dixon-Coles & Machine Learning Approach

We no longer publish soccer picks — this guide stays because the markets are worth knowing. Live now: NBA and NHL predictions for every game, NFL prop boards, and MLB during the baseball season. Every play is graded on the performance page.

Want soccer back? If enough readers ask, we'll bring back match projections for the top European leagues as part of our paid service. One email at launch, nothing else.

How a 1997 statistical model still beats modern markets — covering 3-way moneyline, Asian handicap, expected goals (xG), and why soccer ML accuracy caps at 55-58%.

Updated May 2026 · 17 min read

64.5%
Serie A Top Picks Hit Rate
58.7%
EPL Filtered Hit Rate (201 picks)
1,216
Graded Walk-Forward Picks
55-58%
Soccer ML Ceiling

The soccer betting markets that suit a prediction model, and why: (1) Asian handicap, because the 2-way structure eliminates draw uncertainty and the market is softer; (2) moneyline (1X2), which produced 56-65% win rates with edge filters in our backtest, the highest hit rate of the three; (3) totals (over/under 2.5 goals) is the weakest fit because most models are over-confident on goal totals.

1. Why Soccer Is the Hardest Major Sport to Predict

If you have ever wondered why soccer betting markets reward sharp models less than baseball or basketball, the answer is in three numbers: 3 outcomes, 2.5 goals per game, 90 minutes. Compared to MLB's 9 innings of repeated trials or NBA's 200+ possessions, soccer compresses the entire result into a handful of decisive events.

A single goal — sometimes from a deflection, a goalkeeper error, or a penalty kick on a soft foul — flips the outcome from win to draw or loss. This compresses the upper bound of predictive accuracy. Across the published academic literature on soccer prediction, full-time 1X2 (home/draw/away) models cap at 52-58% accuracy. Our own walk-forward tests on 1,216 graded EPL and Serie A predictions land in the same range — 51-52% on raw accuracy, with high-confidence buckets reaching 64-72%.

For comparison, a two-way market like an MLB or NBA moneyline allows a much higher hit rate on strong favourites. Soccer simply does not allow that ceiling — a fact about the sport, not a flaw in the methodology.

The Draw Problem

Roughly 22-28% of soccer matches end in a draw. That is a structural floor on uncertainty. A model that perfectly predicted home and away wins (impossible) but never picked a draw would still miss 25% of all fixtures. This is why the most successful soccer betting strategies focus on Asian handicap (a 2-way market with no draw outcome) rather than 3-way moneyline.

The Variance Wall

In MLB, a starting pitcher faces 25-30 batters. That is enough trials for true skill to surface. In NBA, a team takes 80-100 shots per game. That smooths out variance. In soccer, the average match has just 10-15 shots on target combined and 2-3 goals. Each goal carries enormous weight — a single deflection changes the outcome in a way no NBA basket ever does. Models can identify the better team, but converting that knowledge into a confident prediction runs into the wall of low-event variance.

2. The Five Soccer Betting Markets That Matter

Moneyline (1X2)

The classic 3-way market: pick whether the home team wins (1), the match draws (X), or the away team wins (2). Odds are typically displayed in decimal format in European markets, American format in US books. A typical EPL match might price the favorite home team at 1.50 (-200 American), the draw at 4.00 (+300), and the underdog away at 6.50 (+550).

This is the most popular market and the sharpest. Top books embed margins as low as 3-4% on flagship EPL matches. Smaller leagues see margins of 5-8%. The combination of high liquidity and tight margins makes this the hardest market to beat consistently.

Asian Handicap (AH)

Asian handicap is the most analytically interesting soccer market. Instead of three outcomes, it offers two: home covers the handicap, or away covers the handicap. Lines are set in increments of 0.25 goals. For example, "Manchester City -1.5" means City must win by 2 or more goals; "Manchester City -0.25" means City must win or draw with the bet half-refunded.

Asian handicap is significantly softer than moneyline because casual bettors find the quarter-line math confusing. Less recreational money flowing into the market means books spend less time sharpening the lines. Our backtest covered 1,418 EPL matches in this market. See our dedicated Asian Handicap explained page for the math.

Totals (Over/Under 2.5)

Will the total goals scored by both teams exceed 2.5? Half-line so no push. The standard line is 2.5 goals; you also see 1.5 and 3.5 as alternates. The over hits in roughly 53% of EPL matches and 50% of Serie A matches.

Despite being a binary market like Asian handicap, totals is harder for Dixon-Coles to beat. Our calibration testing shows the model becomes over-confident at higher probability buckets — predicting 74% over but actually hitting 58%. We do not currently include totals in our shippable picks for that reason.

Both Teams To Score (BTTS)

A simple yes/no market: do both teams score at least one goal? Yes hits in roughly 50-55% of top-five European league matches. BTTS is mathematically derivable from a Dixon-Coles joint goal distribution (1 - P(home shutout) - P(away shutout)) but books offer it as a pre-priced market with a 5-6% margin.

Player Props (Goals, Shots, Cards)

Player-level markets — anytime goalscorer, shots on target over/under, total cards over/under — are the softest soccer markets but also the hardest to model accurately because they require lineup data that is only confirmed about an hour before kickoff. Most published soccer prediction models focus on team-level markets for that reason.

3. The Dixon-Coles Model: Why a 1997 Paper Still Wins

In 1997, Mark Dixon and Stuart Coles published "Modelling Association Football Scores and Inefficiencies in the Football Betting Market" in Applied Statistics. Their model, often shortened to DC, predicts soccer match outcomes using just three ingredients per team: an attack rating, a defense rating, and a global home advantage parameter. Twenty-eight years later it is still the most widely-used soccer prediction model in academic research and competitive prediction markets.

The mathematical core: home goals are modeled as Poisson(λ_h) where λ_h = exp(α_home + β_away + γ). Away goals are Poisson(λ_a) where λ_a = exp(α_away + β_home). Here α is the attack strength, β is the defense strength (negative is better at preventing goals), and γ is the home advantage that applies to every match equally.

The Tau Correction

The clever insight that distinguishes DC from a naive Poisson model is the τ (tau) correction. Pure Poisson independence underestimates how often games end 0-0, 0-1, 1-0, and 1-1 — the four lowest-score cells of the result matrix. DC introduces a single correction parameter ρ that adjusts these four cells upward. The result is a joint distribution P(home_goals=i, away_goals=j) that matches empirical low-score frequencies and produces much better-calibrated draw probabilities.

Because the tau correction makes draws more likely than naive Poisson predicts, DC models tend to assign higher draw probability than ML classifiers trained directly on the H/D/A label. Our walk-forward comparison showed DC predicting 8 draws across 1,216 matches versus a calibrated XGBoost predicting 0 draws. DC's draw recall remained low (under 1%) but its calibration stayed honest in the sense that probability buckets matched actual outcomes.

Time-Decay Weighting

The original DC paper introduced exponential time-decay weighting to give more weight to recent matches. Each historical match contributes to the likelihood with weight exp(-ξ × days_ago), where ξ controls the half-life of relevance. We use ξ = 0.0065, which corresponds to a 107-day half-life: a match from 107 days ago contributes half as much information as a match from yesterday. This handles squad changes, manager turnover, and form swings without requiring explicit feature engineering.

Why DC Beats XGBoost on Soccer

We tested adding an XGBoost classifier trained on rolling shot statistics (shots, shots on target, corners, fouls) and ensembling it with DC. Pure DC outperformed every ensemble configuration we tried. The likely reasons: goals are downstream of shots, so DC's attack/defense ratings already encode the shot-related signal. Adding raw shot features to the model adds noise, not orthogonal information. Without xG (which the FD historical data does not include), shot-count features are too noisy to help.

Read our deeper Dixon-Coles methodology page for the full math, code references, and reproducible backtest setup.

4. Expected Goals (xG): The Most Important Stat in Modern Soccer

Expected Goals (xG) measures the quality of each shot based on its location, angle, defensive pressure, and shot type. A tap-in from 3 yards out might have xG of 0.85 — meaning historically, similar shots convert into goals 85% of the time. A speculative shot from 35 yards might have xG of 0.02. Summing all shots in a match gives total xG: a more stable measure of attacking quality than actual goals scored.

The reason xG matters: actual goals are extremely noisy. A team that creates 2.0 xG worth of chances might score zero, one, two, or three goals in any given match — randomness dominates over single-game outcomes. But over 10 matches, total xG and total goals converge. A team underperforming their xG over 10 matches is statistically likely to regress upward; a team overperforming is likely to regress down. This regression signal is one of the most predictive features in modern soccer ML.

When xG Data Is Available

xG availability varies dramatically by data source and league. API-Sports, our primary source, has xG for top-five European leagues from 2023 onwards. Older seasons have only basic stats (shots, shots on target, corners) without xG. Football-data.co.uk free historical CSVs do not include xG at all. This constrains how far back you can train a model that uses xG as a feature — without losing the benefit entirely on older training data.

Where to Get xG Data (Free and Paid Sources)

The four authoritative xG data sources, ranked by cost and coverage:

  • Understat — free, covers EPL/Serie A/La Liga/Bundesliga/Ligue 1/RFPL from 2014. Per-shot xG with location data, no API key required, and easy to scrape. The most popular free xG source for hobbyist analysts.
  • FBref — free, powered by StatsBomb. Comprehensive match and player xG with detailed advanced stats. Covers top-five European leagues plus MLS, Champions League, Europa League. Best for player-level analysis.
  • StatsBomb Open Data — free GitHub repository with detailed event-level data including their proprietary xG model output. Covers select competitions including Women's Super League and historical World Cups. Best for academic research.
  • Opta / Stats Perform — paid commercial. The xG source that most professional sportsbooks and broadcasters use. Coverage includes nearly every professional league globally. Cost runs into thousands per month for API access.

For a hobbyist or independent bettor, Understat plus FBref is sufficient to replicate the analyses on this site. For production betting at scale, paid Opta access or API-Sports football is the practical choice — they provide xG plus odds in a single feed.

The Practical Limit

xG is necessary but not sufficient. Our research showed adding xG-derived features (xG ratio diff, xG overperformance) lifts model accuracy by roughly 1-2 percentage points compared to raw goals alone. That is meaningful but not a silver bullet. Combined with team-strength ratings from Dixon-Coles, xG brings models into the high end of the 52-58% accuracy range — but not beyond.

5. Why Asian Handicap Suits a Goal Model

Of the three soccer markets we tested (moneyline, totals, Asian handicap), AH maps most directly onto the model's goal-margin output. With combined probability + edge filters (model probability ≥ 50%, edge ≥ 6%), the EPL AH picks hit 53.4% across 650 picks over 4 years.

Why AH suits the model:

  • It is a 2-way market, not 3-way. Dixon-Coles produces a clean joint goal distribution, and AH probabilities derive directly from summing margins across that grid. No draw uncertainty to fight.
  • Casual bettors avoid it. Quarter-line math (e.g., -0.75 meaning bet split between -0.5 and -1.0) confuses recreational bettors. Less casual money means books sharpen the lines less aggressively.
  • Integer-line pushes smooth results. When the line is -1.0 and the favorite wins by exactly 1, the bet pushes (refund). This smooths variance compared to moneyline where every pick is win-or-lose.
  • DC naturally models margin. AH bets are bets on the goal margin distribution — exactly what Dixon-Coles produces. There is no translation step where signal can get lost.

6. The Per-League Reality: Not All Soccer Is the Same

One of the biggest mistakes in soccer betting is assuming a model that works on EPL also works on Bundesliga or Ligue 1. We tested Dixon-Coles across the top five European leagues with identical methodology. The results were dramatically different by league:

LeagueAccuracyVerdict
EPL52.0%Ship
Serie A51.6%Ship
La Liga51.7%Marginal
Bundesliga51.9%Skip
Ligue 150.2%Skip

Notice that raw accuracy is roughly the same across all five leagues (50-52%). The very different verdicts come from how each league's market reacts to the model's edge picks. The model held up in EPL and Serie A and did not in the Bundesliga and Ligue 1.

Why Bundesliga and Ligue 1 Break the Model

Both leagues have extreme team concentration — Bayern Munich has won 12 of the last 13 Bundesliga titles; PSG has won 10 of the last 11 Ligue 1 titles. Markets correctly price these teams as massive favorites (often -700 or worse). When Dixon-Coles, fitted on team-strength ratings, says Bayern should win 80% of the time but the market implies 87%, the model thinks it has found a value bet on Bayern. But the market is right. The 7pp edge the model thinks it has is structural skepticism the market correctly applies to dominant teams.

The fix would be league-specific adjustments — discounting model confidence on extreme favorites in dominance-skewed leagues. We have not implemented that yet, so for now we recommend skipping Bundesliga and Ligue 1 entirely.

7. Edge, Calibration, and the Combined Filter

The single biggest mistake recreational bettors make is filtering on probability alone. "Lock pick: model says 78% confident" sounds great until you realize the books also know it is a heavy favorite and price it at -400. To win at -400 you need to hit 80% — a higher bar than the model's 78% prediction. This is why probability-only filters lose money in the long run.

The right approach combines two filters: a probability floor (so you do not bet coinflips) AND an edge requirement (so the market price still has value). The filter we settled on for EPL is p ≥ 55% AND edge ≥ 8% — meaning bet only when the model's probability exceeds 55% AND that probability is at least 8 percentage points higher than the market's implied probability. This produces 201 picks over 4 years at a 58.7% win rate.

Calibration: Trust the Model Where It Earns Trust

We tested how predicted probability matches actual hit rate across the bucketed range. The findings were sharp:

  • 0.50-0.55 bucket: predicted 0.525, actual 0.523 — well-calibrated
  • 0.55-0.60 bucket: predicted 0.575, actual 0.599 — slightly under-confident
  • 0.60-0.65 bucket: predicted 0.626, actual 0.614 — well-calibrated
  • 0.65-0.70 bucket: predicted 0.673, actual 0.675 — essentially perfect
  • 0.70-0.80 bucket: predicted 0.745, actual 0.593 — heavily over-confident
  • 0.80-1.00 bucket: predicted 0.855, actual 0.825 — decent

The lesson: trust Dixon-Coles in the 50-65% probability range. Above that, especially in the 70-80% range, the model becomes over-confident and its picks hit well below their stated probability. This is why our Top Picks tier filters at 55-65% rather than 70%+ — we are explicitly avoiding the over-confidence band.

8. Frequently Asked Questions

What is the highest accuracy a soccer model can achieve?

Published academic models cap at 55-58% on full-time 1X2 results. High-confidence subsets (top 10-15% of picks by model confidence) can reach 65-72% but at vastly reduced volume. Anyone advertising 75%+ accuracy on full-volume soccer betting is either cherry-picking, overfitting, or marketing.

Should I use Pinnacle, Bet365, or DraftKings for soccer?

Pinnacle has the sharpest lines (smallest book margin, ~3-4%) but limits sharp action quickly. Bet365 and DraftKings have wider margins but accept larger bets without flagging accounts. The optimal strategy is best-line shopping: take whichever book offers the highest odds for your specific pick. A few cents of price on every bet adds up over a season, so compare books before every bet.

Does the Dixon-Coles model work on lower-tier soccer leagues?

Probably better than top-tier leagues, in fact. Lower divisions have softer markets (5-8% book margins versus EPL's 3-4%), less sharp money flowing through, and book operators who do not invest as much in modeling these leagues. The trade-off is that statistics like xG are often unavailable for lower divisions, so the model falls back to goal-only Dixon-Coles. We are testing several softer markets including Saudi Pro League, J1 League, and the Norwegian Eliteserien.

How should I size soccer bets?

Use flat-unit sizing (or fractional Kelly) and never increase stake size after losses. Even a well-calibrated model goes through long losing runs: the worst streak in our 4-year EPL Asian handicap backtest was 7 in a row. Only bet what you can afford to lose.

Why does Prediction Engine focus on EPL and Serie A first?

Three reasons: First, both leagues held up across multiple filter configurations in our backtests, where the Bundesliga and Ligue 1 did not. Second, both have wide bookmaker coverage at every major US sportsbook (HardRock, DraftKings, FanDuel, BetMGM, Caesars), so users can actually place bets. Third, both leagues have the data quality required to fit Dixon-Coles cleanly — full match histories, half-time scores, advanced statistics on modern seasons. Lower-tier leagues lack one or more of these requirements.

Why we don't publish soccer picks

Soccer is one of the hardest sports to project, for the same reason baseball is: very little scoring. Top leagues average under three goals a game and roughly one match in four ends level, so a single deflection, penalty or red card decides a result, and the better team drops points often. A model can read team strength well and still be at the mercy of one moment.

We ran EPL and Serie A models into September 2026. Our members used the MLB, NFL and NHL boards far more, so we retired soccer picks to put the work where people use it. The guides stay: knowing how these markets are priced is useful whatever you bet.

Want soccer back?

If enough readers ask, we'll bring back match projections for the top European leagues as part of our paid service. Join the waitlist and we'll email you once, when they launch.

What we publish now

  • NBA — moneyline, projected margin, totals and player props for every game
  • NHL — moneyline, puck line and player props for every game, every day
  • NFL — moneyline picks, plus free touchdown and yardage prop boards
  • MLB — moneyline, player props and pitcher strikeouts during the baseball season

Every play is graded in public on the performance page, wins and losses alike.

Responsible Gambling

Educational content only. Past performance does not guarantee future results. Sports betting carries financial risk and may be illegal in your jurisdiction. Must be 21+ in legal US states.

If you or someone you know has a gambling problem, call 1-800-GAMBLER or visit ncpgambling.org.

See plans — NBA, NHL, NFL

See Plans