Most bettors use the wrong statistics. Goals, form tables, and possession figures look useful, but they are noisy, influenced by luck, and poor predictors of future results. This guide covers the metrics that actually matter (xG, xGA, PPDA, xPts, and CLV) with practical instructions for using each one.
Not all football statistics are equally useful. Ranked by their predictive power for future results:
| Statistic | What it measures | Predictive power |
|---|---|---|
| xG (Expected Goals) | Shot quality / attacking threat | ★★★★★ |
| xGA (Expected Goals Against) | Defensive quality / shots conceded quality | ★★★★★ |
| xG Difference | Net dominance per match | ★★★★★ |
| xPts (Expected Points) | Points deserved based on xG | ★★★★☆ |
| PPDA | Pressing intensity | ★★★★☆ |
| npxG | Open-play goal threat (excl. penalties) | ★★★★☆ |
| xA (Expected Assists) | Creative quality per pass/cross | ★★★☆☆ |
| Shots on Target | Attacking accuracy volume | ★★★☆☆ |
| Actual Goals | Goals scored/conceded | ★★☆☆☆ |
| Possession % | Ball time share | ★☆☆☆☆ |
★★★★★ = most predictive of future outcomes. ★ = least predictive. Based on academic research into football predictive modelling.
Scores each shot 0–1 based on location, angle, assist type. Sum over a match gives total chance quality. The core input for all modern prediction models.
Better than
Actual goals scored (noise-free over 5+ matches)
Best for
1X2, Over/Under, Asian Handicap
Practical tip
Use 5-match rolling xG rather than season totals. Recent form is more predictive than long-run averages for short-term betting.
Same calculation as xG but from the defensive perspective: total quality of chances conceded. Low xGA = defensively disciplined team generating few clear chances against.
Better than
Goals conceded (goalkeeper variance removed)
Best for
BTTS No, Clean sheet markets, Asian Handicap defence
Practical tip
Teams with xGA < 0.80 per game are elite defensive sides. BTTS No is a strong market against them regardless of opposition xG.
Passes Per Defensive Action: measures pressing intensity. PPDA ≤7: high press. PPDA 8–12: moderate. PPDA 13+: low block/passive. Determines match tempo and goal market selection.
Better than
Possession (PPDA captures intent, not just time on ball)
Best for
Over/Under, BTTS, game-flow prediction
Practical tip
When a high-pressing side (PPDA ≤8) faces a low-block team (PPDA 14+), back Under 2.5. The low-block will neutralise the press and create a slow, low-xG contest.
Points a team should have earned based on their xG in every match. A team 8 points below xPts is overdue a positive run; one 10 points above is likely to decline.
Better than
League table position (removes goalkeeper/striker luck)
Best for
Identifying undervalued teams in league winner/relegation markets
Practical tip
xPts divergence of 6+ points over 10 matches is a statistically meaningful signal. Small xPts gaps are noise.
xG with penalties removed. Penalty rates are highly variable year to year, so npxG gives a stable read of a team or player's open-play goalscoring threat.
Better than
Raw xG (more stable, less penalty rate noise)
Best for
Player goalscorer markets, long-term team quality assessment
Practical tip
Compare a striker's goals to their npxG over a season. Large overperformance (goals >> npxG) signals regression ahead.
The percentage your opening/betting odds beat the closing odds at a sharp book. Positive CLV is the strongest evidence of genuine edge, and beats analysing results alone.
Better than
Raw profit/loss (CLV identifies skill from luck far faster)
Best for
Evaluating your own betting strategy over time
Practical tip
Positive CLV of +3% sustained over 500+ bets at a sharp book (Pinnacle) is strong evidence of genuine skill.
Different betting markets require different statistical inputs. Here is the optimal data approach for each major market:
Key statistics
Method
Build or use a Poisson model with each team's xG for/against per game. Convert scoreline probabilities to 1X2 probabilities. Compare to bookmaker implied probability.
Key statistics
Method
Sum both teams' average xG per game (last 5 matches, home/away split). Run Poisson to get goal distribution. Compare P(≥3 goals) to Over 2.5 implied probability.
Key statistics
Method
P(home scores) × P(away scores) gives approximate BTTS probability. Use Poisson for precision. A team with xG 1.5+ per game scores in ~70% of matches.
Key statistics
Method
Sum all Poisson scoreline probabilities where home wins by ≥2 goals (for -1.5 AH). The exact handicap line chosen should match where Poisson shows most value vs market price.
Key statistics
Method
ATSG probability ≈ P(team scores) × player's share of team goals. Compare to bookmaker implied probability. Look for players whose xG per 90 is consistently above what the odds imply.
Key statistics
Method
P(clean sheet) = P(opponent scores 0 goals) from Poisson using their xG vs this team's xGA. Teams with combined xGA < 0.7 and opponent xG < 0.9 have strong clean sheet cases.
Not all data sources are equal. These are the most trusted providers for each type of football statistic:
xG, xGA, xA, npxG, PPDA, progressive passes. Full squad and individual player data across all major leagues.
Match-level xG with shot maps and rolling xG charts for all major European leagues. Excellent for match-by-match analysis.
Team and player ratings, form guides, detailed match stats. Good for surface-level analysis though not as deep on advanced metrics.
Live match data, player ratings, expected stats. Strong mobile experience. Limited historical depth.
The industry gold standard for advanced data. Full StatsBomb data is available free for women's football, select competitions, and via FBref.
The primary professional data supplier. Powers most major football publications and betting sites. Very expensive, typically for commercial use only.
A team's season xG average includes matches from September that are irrelevant in March. Use rolling 5-match xG. It reflects current form, squad changes, and tactical evolution.
Most teams are meaningfully better at home. Season averages combine home (xG 1.8) and away (xG 1.0) into a meaningless 1.4 average. Always split by venue when making predictions.
A team conceding early often ends up with more possession than their dominant opponent. Possession % at full-time reflects game state, not quality. PPDA and xG tell the real story.
H2H records 5+ seasons old are essentially noise; squads and managers change completely. Recent tactical matchup analysis using PPDA and xG is more predictive than a 2018 result.
Three matches of high xG does not confirm a team's quality. xG stabilises over 8–10 matches. PPDA stabilises over 5–7. Act on signals when sample size is sufficient, not at the first data point.
KiqIQ's free calculators turn the statistics above into concrete probability estimates and bet sizing recommendations.
Poisson Calculator
Enter each team's xG to generate scoreline probability matrix and derive 1X2, Over/Under, BTTS, and AH probabilities.
BTTS Calculator
Input both teams' scoring probability to calculate expected BTTS Yes/No probabilities and compare to bookmaker odds.
Implied Probability
Convert any odds format to implied probability, and strip out the bookmaker margin to see the true market price.
Fair Odds
Remove the bookmaker overround from any set of 1X2 odds to reveal the true probability each outcome is priced at.
Kelly Criterion
Calculate the mathematically optimal stake size based on your edge and the available odds.
Over/Under Calculator
Enter expected goals to get the full goal distribution and exact Over/Under probabilities for any total line.
Expected Goals (xG) is the single most predictive statistic for future match outcomes. It measures the quality of chances created, not just the goals scored, making it far more stable than raw results. Over 8+ matches, xG-based models consistently outperform result-based analysis in predicting future performance.
PPDA (Passes Per Defensive Action) measures how aggressively a team presses. A PPDA of 7 or below indicates a high-pressing side; 13+ indicates a passive, deep-block team. PPDA is essential for predicting match tempo and selecting the right betting markets. Low-block vs high-press matchups tend to produce Under results and BTTS No outcomes.
Compare your xG-based win probability estimate to the bookmaker's implied probability (convert their decimal odds: 1/odds × 100). If your model gives Team A a 60% win probability and the bookmaker implies 50%, the 10% gap is a potential value opportunity. Use our Poisson calculator to derive win probabilities from xG inputs.
Expected Points (xPts) calculates the points a team should have earned based on their xG in every match: what the Poisson model says their results should have been without goalkeeper luck or striker variance. A team 8+ points below their xPts is overdue positive regression; one 10+ points above is likely to decline. A critical metric for identifying mispriced teams in match winner, league winner, and relegation markets.
In order: (1) xG difference per match, (2) xGA per match, (3) PPDA, (4) xPts divergence, (5) npxG, (6) shots on target difference, (7) actual goal difference, (8) total shots difference, (9) possession. xG-based metrics outperform all other publicly available statistics for forecasting future match outcomes.
What is xG?
The complete guide to Expected Goals: how it's calculated and how to use it
xG vs Actual Goals
Why xG predicts better than goals, and how to exploit the difference
PPDA Explained
How pressing intensity is measured and what it tells you about a match
Poisson Distribution Guide
How to convert xG into scoreline probabilities for any market
Closing Line Value (CLV)
The best way to measure whether you're finding genuine edge
Analysing Football Form
A data-driven framework for reading form with the right statistics
