Three Premier League teams whose results might give us a misleading impression

Are Brighton undeservedly low in the Premier League? Are Tottenham undeservedly high? Are Sheffield United maybe in the right area but with the wrong scores?

The Premier League is about to resume after an international break so now seems an appropriate time to reflect on what has happened so far and ruminate on what might happen next.

In the graph below you will find actual and expected goal differences for every Premier League team. The teams are listed in order of actual goal difference – highest at the top, lowest at the bottom. Beside the actual goal difference there is an expected goal difference.

Expected goals, as most of you know, are an attempt to answer this question: how many goals would be scored on average with the same number of attempts from the same positions? Answers differ, but not usually by much. The ones in the table come from the website understat.com.

The other thing you will see in the graph is the gap between actual and expected goal differences. A positive number means the expected goal difference is higher than the actual goal difference, and a negative number means the opposite.

Brighton’s expected goal difference is 17 better than their actual goal difference. Sheffield United’s is 11 better. Tottenham’s is 14 worse.

We can put that another way. Brighton rank fifth for expected goal difference but only 12th for actual goal difference. Tottenham rank fourth equal for actual goal difference but only tenth for expected goal difference. Sheffield United rank 19th for both – their expected goal difference is a lot better than their actual goal difference, but for both they are worse than everyone except West Bromwich.

What does this tell us? Let us start with some general observations.

Ten teams have a positive real goal difference and ten teams have a negative real goal difference. Eight of the ten teams with a positive real goal difference have an expected goal difference that is worse. Seven of the ten teams with a negative real goal difference have an expected goal difference that is better.

We would notice this sort of thing in any competition at any time in any season. Why?

Expected goals are supposed to tell us how many goals would be scored on average with the same number of attempts from the same positions. Some teams are better than average, others are worse. Because some players are better than average and others are worse.

So discrepancies between actual and expected goals can occur because a team have displayed above-average or below-average skill. But they can also occur because the team experienced good or bad luck

Usually teams near the top of a table have genuinely played better than teams near the bottom. Usually, though, they have benefited from kinder fortunes as well.

Past expected goals tend to be better than past real goals as predictors of future real goals. But neither is especially precise. You would get less inaccurate predictions by using a combination that gave some weight to past real goals but more weight to past expected goals. This tells us that in most instances discrepancies up until now between real and expected goals are due partly to skill and partly to luck, but more to luck.

Tottenham probably do have better players than Brighton. They should have. They can offer more. In the last three completed seasons – 2017-18 to 2019-20 – Tottenham’s wage bill was almost double Brighton’s. Those were the three seasons after Brighton won promotion to the Premier League. Perhaps this season also Tottenham have been lucky at times and Brighton unlucky.

There is no reason why a team who were lucky or unlucky in the past should be lucky or unlucky in the future. They might be, but just as likely they will not. Skill tends to be more enduring. But good players can play poorly at times, and bad players can play well at times. So even if a team deserved the results they got up until now they could play another way and deserve different results from now on.

Stats, of all descriptions, come from the past. Even if we have interpreted them well – and we can never be sure – all we could say is that we know how the team performed in the past. They might not perform the same way in the future.

This is one of the reasons why past real goals and past expected goals, separately or in any combination, are far from perfect predictors of future real goals.

The contrasts this season between attempts and scores have been striking, though, for Tottenham, Brighton and Sheffield United. Let us illustrate them in another way.

Sheffield United have taken 38 per cent of the shots in their games but scored only 24 per cent of the goals. Comparing Brighton with Tottenham is particularly interesting. Brighton have taken 60 per cent of the shots in their games and scored 47 per cent of the goals. Tottenham have taken 48 per cent of the shots in their games and scored 62 per cent of the goals. Almost the reverse.

Probably Sheffield United should be in the relegation zone, but not as deeply mired in it. Perhaps Brighton should be further away from the relegation zone, perhaps Tottenham should not be as close to a Champions League place. What they will deserve in the rest of the season, though, could be something else.

Sign up to Sky Bet and get £30 in free bets when you place a £5 bet

First single & EW bet only. Odds of 1/1 or greater. 3 x £10 bet tokens. Free bet stakes not included in returns. Free bets are non withdrawable. No free bet expiry. Eligibility restrictions. Further T&Cs apply. 18+ begambleaware.org

CLAIM OFFER HERE

MORE FREE BETS

Today's top sports betting stories

Follow us on Twitter @racingpostsport