Attack and defence Elo ratings: how to read a team's strength

Why two numbers instead of one?

The classic club Elo ranking says how strong a club is overall. That is a good answer to "who will win?", but it does not say how a team wins. Two clubs with the same overall rating can look completely different: one wins 3-2 because it scores a lot and concedes a lot, the other wins 1-0 because it gives opponents almost nothing.

Splitting strength into attack and defence shows this difference. It matters above all for goals: whether a match looks open or tight, and which side has the better chance of scoring.

How to read the scale: 1500 and the hundreds

Both ratings start from the same base:

• 1500 is an average team in the league at that moment,
• attack 1600 means the team scores about 1.2 times as many goals as an average team of that league would against the same opponent,
• defence 1600 means it concedes about 1.2 times fewer goals than an average team, that is roughly 83% of what an average side would concede,
• attack 1400 means about 1.2 times fewer goals scored, defence 1400 about 1.2 times more conceded.

The effect multiplies. A 200-point gap is already about 1.44 times (1.2 × 1.2), and 300 points about 1.73 times.

Important: the ratings are per league. An attack of 1600 in the Premier League and an attack of 1600 in the Championship both mean "clearly better than the average of its own league", not the same thing in absolute terms. To compare clubs from different leagues, use the overall club rankings, which have one common scale. There you can also switch to the experimental Elo with closing odds, which learns from the market's prices as well as from results.

Where the ratings come from

Before every match the model works out how many goals each side should score. It takes into account the league's average number of goals, home advantage, one team's attack and the other's defence. After the match it compares these expectations with what happened:

• if a team scored more than expected, its attack goes up and the opponent's defence goes down,
• if it scored fewer, the opposite happens,
• the closer the result was to expectations, the smaller the change.

This way the ratings account for the strength of the opponents on their own. Three goals scored against the leaders lift the attack more than three goals against a team near the bottom, because fewer were expected against the leaders.

The second important thing is xG. Goals are rare and random, so where we have expected goals (xG) data for both teams, we compute the ratings half and half: half from the actual goals, half from xG. In our tests this blend worked better than goals alone or xG alone. Where there is no xG, the ratings rest on goals only.

At the start of a season, the ratings of teams that stayed in the league are pulled part of the way back towards the average, because squads and managers change over the summer. A promoted team starts at the level of the teams that were relegated from the league the season before.

Team profiles

Next to the ratings we show a short profile. It follows directly from the numbers:

A profile is shorthand, not a verdict on quality. An "attacking" team can be weaker than the league average if its defence is very poor, and a "defensive" one can be a top club that simply wins by narrow scores. Always look at both numbers, not just the label.

How to read the league chart

On a league page all the teams are plotted on one chart. The horizontal axis is attack, the vertical one defence, and the lines cross at 1500, the league average. That makes four quadrants:

• top right (complete): teams that score more than average and concede fewer than average. This is usually where the title contenders are,
• top left (defensive): a good defence and a weaker attack. Such teams more often play low-scoring matches,
• bottom right (attacking): a good attack and a poor defence. Their matches tend to be more open,
• bottom left (struggling): below average in both. This is where the teams fighting relegation usually are.

The distance from the centre shows how clearly a team stands out from the average. A dot right next to where the lines cross is an average team, even if it formally sits in the "attacking" quadrant.

How a strength gap turns into results

Attack and defence ratings describe goals. For the question "who will win?", the overall club Elo ranking fits better. To show what a gap in that ranking looks like in practice, we checked 35,455 league matches from 37 leagues played from July 2023 to June 2026. The Elo gap is the home team's rating minus the away team's, including home advantage. How to turn it into match chances is shown in our guide on working out match odds from Elo ratings.

Two practical points follow from the table:

• roughly 5 percentage points for every 50 Elo points. Between the 0–50 and the 150–200 bands the home side's chance rises from 40% to 54%. That is a good rule for a quick read of a match,
• a big strength gap means more goals. When one team clearly dominates, the match more often ends with over 2.5 goals, while with evenly matched teams that share falls below half.

If you want the chances for a specific pair, use the Elo calculator. You pick two clubs and get the chances of 1, X and 2, with home advantage included. In its "Attack and defence" mode it also gives the chance of every correct score from 0-0 to 5-5, which helps with Skrill Final Score picks.

On the football tips list, the attack and defence ratings show up twice: as expected goals in the dashed squares by the team names and as the blue chances in the Goals column, among them the chance that both teams score. We show how to read them on a real match in how to read football tips. The match page adds the "Attack vs defence" tile, which sets each team's attack against the opponent's defence, as we describe in how to read the match page.

What the ratings do not tell you

• They do not know the line-ups. An injury to the best striker will not lower the attack until it shows in the results.
• They do not say when a team will score or concede, and nor does its record by minute: in our late goals test on 11,142 checkpoints, the timing of past goals did not predict the next matches.
• They react gradually. After a change of manager or big signings it takes a few rounds before the numbers catch up with reality.
• They describe strength, not a ready-made tip. They are a starting point for your own view of a match, not a pointer to what to bet on. Why even a good model is not enough against the bookmaker, we explain in a separate guide: why a betting model alone is not enough.

FAQ

What does an attack rating of 1500 mean?

It is exactly the league average. A team with an attack of 1500 scores as many goals as an average team of that league would score against the same opponent at the same ground.

Can I compare the attack of teams from different leagues?

Not directly. Attack and defence ratings are calculated separately for each league and relate to its average. For comparisons across leagues, use the club Elo ranking.

Why did a rating change after a match the team won?

Because what counts is the result against expectations. If the favourite won 1-0 and the model expected about two goals from it, its attack may have dipped slightly, even if its defence went up.

Do the ratings take xG into account?

Yes, wherever xG is available for both teams. Then goals and xG count half and half. In matches without xG the ratings rest on goals alone.

How often are the ratings updated?

After every finished league match. A club's page also shows the change over the last 30 days and a chart of its attack and defence over time.