Why xG Is a Better Predictor Than Form Tables

Every football fan does it. Your team has won three on the bounce, you check the form table, and you start thinking the season's turning around. Or your rivals have lost twice in a row and suddenly they're "in crisis." It's natural — results are the most visible thing in football, so we use them as our compass.

But here's the problem: results lie. Regularly. And if you're using recent form to predict what happens next, you're building on sand.

The Problem With Form Tables

A form table — the last five or six results — feels like it should be a reasonable indicator of how a team is playing. They won four, drew one, and lost one? They must be in decent nick. But this approach has a fundamental flaw: it treats all wins as equal and all losses as equal, with zero nuance about what actually happened on the pitch.

Consider two teams, both with W-W-W-D-L in their last five matches:

Team A created chances worth 1.8, 2.1, 1.6, 1.9, and 2.3 xG in those five games. They won their three matches comfortably and were the better side in the draw and the defeat. Their underlying performance has been consistently excellent.

Team B created chances worth 0.6, 0.8, 0.4, 1.1, and 0.7 xG in those five games. They won their three matches through a combination of goalkeeper heroics, opposition wastefulness, and a couple of scrappy set-piece goals. They were outplayed in four of five games and got lucky.

The form table says these teams are identical. The xG data says Team A is significantly better and Team B is heading for a correction. In the next five games, which team would you back?

Why Recent Results Can Be Misleading

Football is a low-scoring sport, and that's precisely what makes results unreliable over small samples. In basketball, the better team on the night wins most of the time because there are so many scoring opportunities that luck evens out within a single game. In football, a single deflected shot or a referee's decision can flip the result entirely.

The variance problem

A team that creates 1.5 xG per game and concedes 1.0 xG per game is, by any reasonable measure, a good side. They're creating half a goal more than they're allowing. But in any individual match, there's a significant probability they lose. Goals are discrete events — you can't score 1.5 goals. Sometimes their 1.5 xG produces zero goals, and sometimes their opponent's 1.0 xG produces two.

Over five games, it's entirely possible for this genuinely good team to go W-L-L-D-L. That's a rotten run of form. The form table screams relegation battle. But the underlying numbers say they're fine, and over the next twenty games they'll almost certainly bounce back.

The "clutch" illusion

When a team wins a series of tight games — 1-0, 2-1, 1-0 — fans call them "gritty" or "winners." And sometimes that's fair. But often what's actually happened is they've been in a series of close games where small margins went their way. The goalkeeper made one world-class save per game. The striker converted their one decent chance. The opposition hit the post twice.

None of that is repeatable. You can't count on hitting the post going in your favour or the keeper pulling off a miracle every week. xG strips away these random fluctuations and shows you the quality of chances created and conceded — the thing that IS repeatable.

The Brighton Case Study

No team illustrates this better than Brighton under Roberto De Zerbi in the 2022/23 season. Brighton's xG numbers were extraordinary — they were creating chances at a rate that would put them among the top four or five sides in the league by underlying performance. Their xG per game regularly topped 2.0 in many matches.

But their actual goal output was considerably lower. They drew games they should have won. They lost games where they'd created enough chances for a comfortable victory. The form table had them as a solid mid-table side. The xG data said they were one of the best attacking teams in the division.

What happened next was instructive. While Brighton's points tally didn't always match their xG-predicted levels (partly due to squad changes and other factors), the underlying quality of their play was evident to anyone watching. Teams with those underlying numbers are always dangerous, and dismissing them based on actual results would have been a mistake.

The lesson: if you'd judged Brighton purely on results and form, you'd have underestimated them. The xG told the real story.

How xG Smooths Out Variance

The mathematical reason xG outperforms form tables is that it uses continuous data rather than binary outcomes.

A form table reduces every match to one of three outcomes: win, draw, or loss. That's three data points across 90 minutes of football. An xG model, by contrast, captures every shot from every game, with the quality of each chance preserved. A match where a team had eight shots worth 2.4 xG and lost 0-1 is recorded very differently from a match where a team had two shots worth 0.3 xG and lost 0-1. In the form table, they're identical.

This matters enormously for prediction. When you have more granular data, you need fewer games to get a reliable signal. A team's xG numbers over five or six games give you a much clearer picture of their true quality than their five or six results do. And over a full season, the correlation between xG and final league position is consistently stronger than the correlation between early-season form and final position.

Real-World Predictive Power

Academic research and betting market analysis have repeatedly shown that xG-based models outperform results-based models for predicting future outcomes. Here's why that matters in practical terms.

Early-season predictions

After five matchdays, the league table is essentially random. Small sample, high variance, and often wildly unrepresentative of where teams will end up. Remember when Everton were top of the league after a few games in 2020/21? They finished tenth. Leicester famously won the league in 2015/16, but that was a genuine once-in-a-generation outlier — and even then, their underlying numbers were better than most people realised at the time.

xG data after five games isn't perfect either, but it's far more useful than the table. A team that's created 2.0 xG per game but only has four points is a much better bet to improve than a team that has ten points but has been creating 0.8 xG per game.

Mid-season adjustments

By Christmas, the league table starts to become more meaningful because the sample size is larger. But xG data still adds significant value, particularly for identifying teams whose results are about to shift. A mid-table team with top-six xG numbers is likely to climb. A team in the top four with mid-table xG numbers is likely to fall.

Identifying regression candidates

This is where xG analysis is most powerful. Teams that significantly overperform their xG (scoring many more goals than their chances suggest) tend to regress — their results get worse even if their performances don't change. Teams that significantly underperform their xG tend to improve.

This isn't a guarantee for any individual match, but over a ten-game window it's one of the most reliable patterns in football analytics.

When Form Tables DO Have Value

To be fair, form tables aren't completely useless. There are situations where recent results carry information that xG doesn't capture:

Confidence and momentum

There's a genuine psychological element to football. A team on a winning run carries itself differently — players make bolder decisions, the crowd is behind them, and the manager has more tactical flexibility because results are buying time. xG can't measure confidence. But this effect is smaller than most fans think, and it dissipates quickly after one bad result.

Injuries and squad changes

If a team has been on a good run and then loses their best player to injury, neither the form table nor historical xG will capture that. You need to factor in context that no single metric provides.

Tactical changes

A new manager or a shift in formation can change a team's underlying performance level overnight. Historical xG data might not reflect a team that's playing a completely different system. You need to weight recent xG data more heavily when there's been a significant tactical change.

The Practical Takeaway

If you're trying to figure out which teams are genuinely good and which are flattered by their results, here's the approach:

1. Look at xG for and xG against over the last ten games — this gives you a reliable picture of the quality of chances a team is creating and conceding
2. Compare that to their actual goals scored and conceded — a big gap suggests regression is coming
3. Check the form table last, not first — use results to add context, not as your primary signal
4. Weight recent xG data more if there's been a managerial or tactical change — the last five games under a new manager matter more than the first fifteen under the old one

The form table tells you what happened. xG tells you what's likely to happen. And if you're in the business of prediction — whether that's a bet, a fantasy team, or just bragging rights with your mates — "likely to happen" is what you want.

Conclusion

Form tables are the fast food of football analysis — quick, easy, and ultimately unsatisfying. They give you a surface-level view that's distorted by luck, variance, and the low-scoring nature of the sport. xG isn't perfect, but it's a far more reliable foundation for understanding team quality and predicting future results.

Next time someone tells you a team is "in great form" because they've won three in a row, ask what their xG numbers looked like. You might find the story is very different underneath.