Expected Points (xPts) Explained: Which Teams Are Lucky?

Every season has one. A side wins three on the bounce, climbs the table and suddenly gets talked up as a European contender, though nobody who watches them every week can quite explain how. Another team dominates games, hits the woodwork and keeps trudging off with nothing. Over 38 matches those two stories usually meet somewhere in the middle, and the bettor who saw it coming gets the better prices.

Expected points (xPts) is how you spot it early. It takes the quality of chances each side created and conceded, works out how often that match "should" have been won, drawn or lost, and adds up the points a team deserved.

Below there's a worked example you can follow with a calculator, the reason data providers never quite agree on the number, and the 2026/27 Premier League expected table after eight matches per team. Then we look at what the last two seasons say about luck evening out, and how to turn an xPts gap into a sensible bet.

Expected goals (xG) puts a value on every shot: the chance it ends up in the net, based on thousands of similar shots. Expected points takes that one step further and asks what those chances were worth in the currency that actually decides the league: wins, draws and points.

The logic is simple. A team that out-creates its opponent most weeks should rack up points, whatever the scorelines say in the short term. Add up the deserved points across a season and you get an "expected table", the league as it would look if finishing and goalkeeping had been perfectly average for everyone.

The gap is where it gets interesting. Real points minus xPts is the headline over-performance number (call it the luck number if you like). A team well above its xPts has had the margins go its way: a deflection here, a worldie there, an opponent missing a sitter. A team well below has had the opposite.

None of that makes the expected table the "real" table. It describes what already happened on the pitch rather than predicting the final standings. Over-performers do tend to drift back towards the pack, but it's a tendency, not a law, and telling a lucky team from a genuinely good one is what the rest of this guide is about.

Every version of xPts starts from the same scoring: 3 points for a win, 1 for a draw, 0 for a loss. The hard part is estimating the probability of each result from the chances in the match. That's the whole xPts vs xG difference in a sentence: xG counts the chances, xPts turns them into results.

xPts = 3 x P(win) + 1 x P(draw)

P(win) and P(draw) come from simulating the match many times,
with each side's goals drawn from its xG
(a Poisson model: goals ~ Poisson(team xG) v Poisson(opponent xG))

Plug in a 50% win chance and a 30% draw chance and you get 1.8 xPts. Opta's model, for example, simulates each match 10,000 times from the xG of the shots and records the share of simulations won, drawn and lost. If the chance values underneath are new to you, our guide to using xG for smarter football betting walks through them.

Our worked example

Here's what four typical xG lines are worth, using independent Poisson goals for each side.

Two things jump out. A side that racks up 2.5 xG and concedes 0.4 banks about 2.6 xPts even if it loses 1-0. The model doesn't care that the ball wouldn't go in.

And an even game is worth about 1.35 xPts each, not 1. A draw is only around 31% likely even when the chances are level, so between them the two teams collect about 2.7 expected points rather than 2.

Shot by shot or match total?

The better models work shot by shot rather than from the match xG total. One 0.8 xG chance and eight 0.1 xG chances add up to the same xG, but they don't produce the same xPts. Eight half-chances give you more ways to score at least once; the single big chance is all or nothing. When several shots come in one move (a save and a rebound, say), the models combine them so the move can't count for more than one goal.

Look up the same team on three sites and you'll get three xPts figures. That's normal. Each provider runs its own xG model, with its own view on headers, set pieces, long shots and shots under pressure, plus its own simulation assumptions. Everyone watches the same games and still gets different outputs.

The model has blind spots too:

• Game state. xPts doesn't know the score when a chance arrives. In Arsenal's 3-0 defeat by Brighton in 2026/27, 1.3 of their 1.6 xG came while chasing the game at 2-0 down. Manchester United's 1.8 xG at Hull included 1.2 with Hull already 2-0 up.
• Finishing quality. It inherits xG's average-finisher view of the world, and elite finishers can beat that for years.
• Chaos. Deflections, red cards and refereeing calls don't sit neatly in a shot model.
• Sample size. Over a handful of games xPts is noisy. It earns its keep over longer runs.

So the practical rule is this: always know which provider and which date you're looking at, and compare the direction of the gap across sources rather than fussing over the decimal places.

The 2026/27 Premier League kicked off on 21 August 2026, and by late October every side had played eight league games. These are the ten biggest gaps between real points and Understat's xPts at that point, read on 30 October 2026.

Nottingham Forest are the story. Six points from eight games, yet Understat rates their performances as worth 15.18, not far behind Manchester City's 17.02 and level with Brighton (15.21 xPts, 16 points). On these numbers Forest have been the unluckiest side in the division by a distance.

Bournemouth catch the eye too. Nine points is a modest haul, but 14.86 xPts would put them right among the leading pack. Going the other way, Newcastle and Ipswich have each picked up roughly three more points than their chances justified, while Brighton (−0.8 on the gap), Leeds (−0.1) and Brentford (+0.4) sit almost exactly where they deserve to be.

Three providers, three sets of numbers, one clear direction: City and Arsenal above their xPts, Forest and Sunderland below.

The useful test is whether the other models agree, and they do. statz.ai, with 80 matches played, has Manchester City as the luckiest side on xPts (21 points v 15.3, +5.7) and Arsenal next at +4.7, with Arsenal top of its expected table on 16.3. Its unluckiest are Nottingham Forest (6 points v 13.4, −7.4) and Sunderland (−5.4). Opta's version, published by The Athletic on 22 October 2026 after seven rounds, had Arsenal leading on xPts and City nearly eight points above expectation, the biggest over-performance in the league by a distance.

The size of the gap moves around from model to model. The direction doesn't, and that agreement is what makes a gap worth acting on. It's the same distinction we draw in our guide to form, variance and luck: one model shouting is noise, three whispering the same thing is a signal.

The unlucky ones

Look under the bonnet and the under-performers' numbers are brutal. Sunderland were bottom of the real table after seven rounds but 15th on underlying numbers in Opta's model. At 0-0 they converted one of seven big chances, while opponents scored four of the 13 big chances they created against Sunderland at 0-0. Forest were 19th with five points from seven, more than six points below expectation. Spurs created 10 big chances at 0-0, worth 2.3 xG, and didn't score one.

The ones who might be genuinely good

City are the catch. They have beaten their xG in each of the last eight league seasons, and Erling Haaland has scored 120 league goals from 112.8 xG. Elite finishing is a skill, so some of City's gap is quality rather than fortune.

Brighton show the same idea at the other end of the pitch. Their opponents scored just five goals from 10.2 xG over the first seven rounds, the biggest under-performance by any side's opponents. Some of that is luck, some may be goalkeeping, and an xPts gap on its own can't separate the two.

The classic xPts story is Nottingham Forest in 2024/25. Understat had them on 65 points from just 50.01 xPts, the biggest over-performance in the league, and they finished seventh. Every bettor asked the same question that summer: would it last?

Forest, 15 points above their xPts, dropped 21 points the next season. Manchester United, about 10 below, gained 29, though plenty else changed at Old Trafford, so it would be lazy to put that down to luck alone. Now look at the other two. Bournemouth were nine points under their xPts and barely moved the following year. Spurs were 12 under and improved by only three.

That's the honest picture: regression is real but partial and noisy. Some gaps close, some close slowly, and some reflect things the model can't see. Our guide to regression to the mean in football betting digs into why the pull is a tendency rather than a promise.

2025/26 threw up fresh examples. Aston Villa finished on 65 points from 51.07 xPts, and newly promoted Sunderland collected 54 from 42.03. Wolves went the other way, with just 20 points from 35.44 xPts. Champions Arsenal sat a believable five or so points above their expected 79.87.

xPts measures team strength over a season, so it suits some markets far better than others.

Where it helps

• Match result, draw no bet and Asian handicap. If a side's xPts says it's better than its league position, its true strength may be higher than prices built around recent results suggest. Handicap lines show that difference most cleanly.
• Outright markets. Top four, top six, relegation and "to finish bottom half" markets are priced partly off the table. A team sitting far from its xPts is exactly where the table and the underlying level disagree most.
• Finishing position specials. A big gap tells you which way a team's real points total is more likely to drift.

Where it doesn't

For a single game's correct score, both teams to score or goals markets, go back to xG itself. xPts squashes everything into win, draw or loss and throws away the scoring detail those markets need.

Be realistic about the edge

Bookmakers see expected tables too, and their prices usually already shade for big gaps. Spotting that Forest have been unlucky in 2026/27 isn't an edge on its own; anyone with a laptop has seen it. Any edge lies in the size of the adjustment: whether the market has moved too far, not far enough or not at all. Compare prices across several bookmakers, track your bets against the closing line, and never assume a gap automatically makes a bet value.

When to look again

A rolling view beats the season total. A promoted side that wins its first five while creating relegation-level chances is the textbook fade; a big side losing matches it dominates is the textbook buy. The gaps also get more trustworthy as the sample grows, so the second half of the season is when an expected table carries the most weight.

The January window makes a natural checkpoint. By then every side has 20-odd games of data, and new signings can change a team's chance creation enough to make the old xPts numbers less relevant.

• Back the direction, not the decimal. If Understat, statz.ai and Opta all have a team on the same side of its xPts, take the gap seriously. If they disagree on direction, leave it alone.
• Forest are the 2026/27 buy-low case. Every model we follow rates their play as worth far more than six points from eight games. Premier League relegation betting and bottom-half prices built on the real table are worth checking against that.
• Don't fade City on xPts alone. Eight straight seasons of beating xG says finishing quality is part of their gap, so expect some of it to stick.
• Bournemouth and Sunderland deserve a second look. Both sat well below their xPts in October 2026, and handicap lines built on recent results can lag behind underlying form.
• Ask why the gap exists. Game state, a red card or one freak result can inflate a gap. Check the matches behind it before you bet.
• Revisit the gaps after January. A bigger sample and a reshaped squad make the expected table far more useful around the turn of the year.

Whatever the numbers say, keep stakes level and modest. xPts sharpens your judgement; it doesn't remove the variance it measures.

• How the 'Football Supercomputer' Really Works: Monte Carlo Simulation Explained
• Set-Piece Betting Guide: Why Dead Balls Decide Matches
• Luck vs Skill: Which Sports Are the Most Unpredictable?
• Premier League Relegation Betting 2026/27
• Home Advantage vs Team Quality: Advanced Football Betting Guide

18+. Please gamble responsibly. For free, confidential support visit BeGambleAware.org.