Key Football Statistics That Actually Predict Match Outcomes
Why the Final Score Tells You So Little
A team wins 2-1 and the table shows three points. The shot count reads 18-5 against them, and their opponent's expected goals, which measures the quality of every chance created rather than just the number, came out at 2.4. They were outplayed and got away with it. Do that every week and the results turn.
This is the problem with surface-level numbers. Possession, shot totals and scorelines are outcomes, and over a single match they swing on finishing, a deflection, a red card, a goalkeeper having the afternoon of his season. A mid-table side can sit fourth in October on a run of wins from low-quality chances and a hot striker, then slide to twelfth by February once the finishing regresses. Nothing about their process changed; the results did.
Underlying metrics describe the process that produces results over time. The rest of this article looks at four of them: expected goals, shots on target, pressing metrics and possession.
The Numbers That Actually Carry Predictive Weight
Image: futureofbirth conference, by MarcSmith, CC BY 2.0
Expected goals is the anchor because it captures the one thing a scoreline cannot: the quality of each chance. Every shot is assigned a value between 0 and 1 based on where it was struck from, the angle to goal, how it was set up (a cutback versus a hopeful cross) and how much defensive pressure the shooter faced. A penalty lands around 0.76. A header from 14 yards under pressure might be 0.05. Add the shots up and you have a reasonable estimate of what a team deserved from a match.
The useful figure is xG difference, not xG itself. Over ten games, a side averaging 1.8 expected goals for and 1.0 against is creating a clear surplus of chance quality and, in most cases, will climb the table even if the points have lagged behind. Actual goals swing on finishing form and goalkeeping; chance creation is more repeatable. Two caveats worth keeping. Providers build their models differently, so compare like with like. And a single match is a small sample, which is why the ten-game window matters more than any one Saturday.
Shots on target sit between raw shot counts and expected goals in usefulness. Total shots reward volume: a side that takes 20 efforts from distance looks busy without threatening much. Shots on target at least filter out the blocked and the wild, and the ratio of shots on target for versus against over a season correlates reasonably with points. But it can still be inflated. Long-range efforts that roll tamely to the keeper count the same as a close-range header tipped over the bar.
Take a team with 20 shots, four on target and 0.9 xG. They had plenty of the ball in harmless areas and generated roughly one goal's worth of genuine threat. Compare that with a side that mustered eight shots, five on target and 1.7 xG. The second team created better chances with fewer attempts, and over a month of matches the pattern will usually show up in the results. Shot quality is the better question, and the shot counts only help when you read them alongside it. Chance creation is not the whole story, though. How a team wins the ball back shapes the chances it gets, which is where pressing numbers come in.
What Pressing and Possession Actually Tell You
Pressing numbers describe how hard and how high a team works without the ball. Pressures count every time a defender closes down an opponent. High turnovers count the times a team wins possession in the attacking third, which is the valuable version, because a turnover 30 yards from goal often produces a better chance than a patient move. PPDA, passes allowed per defensive action, bundles this into one figure: the lower the number, the more aggressively a side presses before it commits to a challenge. A team forcing turnovers in the final third both creates higher-quality chances and disrupts the opponent's buildup, so the effect shows at both ends of the pitch.
Pressing is style-dependent. A low block that concedes 60 percent possession by design is not worse at pressing than a high-pressing side; it is doing something different. Compare teams within similar systems rather than across leagues.
Possession on its own correlates weakly with points. Pair it with high xG and low xG conceded and it means something. Sixty-five percent possession spent in harmless areas is not dominance.
- xG difference — the strongest single predictor: sustained gaps between chances created and conceded track points more closely than any other number.
- Shots on target ratio — useful but imperfect, since long-range efforts inflate it and it ignores the quality of each attempt.
- Pressing and high turnovers — context-dependent: valuable within a system, misleading when compared across teams with different defensive styles.
- Possession percentage — weak alone, telling only when read alongside xG created and xG conceded.
- Goals scored — noisy over short spells, since finishing and goalkeeping swing results without reflecting underlying performance.
Putting the Numbers to Work
Image: 2010 - May - 18 - NodeXL - twitter social graph, by MarcSmith, CC BY 2.0
Start with expected goals over five to ten matches, not one. A single game's xG swings wildly on a deflected shot or a red card, but a ten-match total for and against settles into something you can trust. Then check shots on target ratio against that xG figure. If a team creates good chances but their ratio lags, finishing rather than chance creation is the problem, and that tends to correct. Use pressing numbers and possession only to explain what you saw (how high a line sits, whether a side wins the ball in dangerous areas), never to rank two teams against each other. Here is the routine in action: a side has lost three straight, yet their xG difference across those games is plus 1.8 and their shot quality is intact. They are a reasonable bet to turn results around. Treat it as a guide, though. Injuries to a keeper, an early red card or a run of fixtures against top-six sides will distort any sample, so note those before you act.
The limits are real and worth naming. Small samples mislead: three matches of strong xG can come from two penalties and a lucky rebound, so treat anything under five games as noise. Expected goals models also differ between providers, since each weights shot location, angle and set-up differently, which means a figure of 1.9 from one source might read 1.6 on another. Game state distorts the rest. A team chasing a two-goal deficit will press harder and hold more of the ball than the same team at 0-0, so pressing and possession numbers describe the situation as much as the side. Set pieces and penalties swing results on their own, largely outside open-play metrics. Keep fixture difficulty and squad rotation next to the numbers rather than beneath them. Use all of this to understand performance trends across five or more matches, not to predict a single result with confidence.
Next weekend, pick one fixture and give it sixty seconds before kick-off. Look at both teams' five-match xG difference, then their shots on target ratio, and let possession sit in the background as colour commentary. Write the two numbers down somewhere you will see them again. The habit matters more than any single reading: follow one team across a month and you will watch the underlying figures move before the results do, which is the point where this stops being trivia and starts being useful. When a side's xG difference climbs for three weeks while their points stay flat, you have noticed something the table has not caught up with yet. The World Football Scoreboard is built for exactly that kind of tracking, with the match data laid out where you can scan it quickly and compare across weeks.