Using Modelling to Review of 2024-25 English Premier League to date – one word BORING

22-Jan-2025 by Kerry O’Sullivan

Not the model, the league, I need to clear that up first.

Let me explain.

Almost 2 years ago, following with the demise of the 538 Soccer** prediction model (https://projects.fivethirtyeight.com/soccer-predictions/global-club-rankings/) I decided to do something new. Using similar methods they described; I started building my own football prediction model. Quickly, I could output the similar predictions individual game outcomes and predictions of final league placement as did the 538 model.

Now having access to all the model’s data, this has allowed me to start doing other things.

I could now start measuring and plotting individual teams defensive and offensive capabilities throughout a season.

I could run experiments to see what better, having the best defence or offence? (defence if you’re looking for an answer).

It is also I discovered a great way to take a high-level review of a league and see what’s really going on away from the day-to-day transfer windows, injuries, managerial firings and the general Monday-morning quarterbacking (yes, I know it an American Football term, but there really is no English football equivalent version)

So, what’s (beyond Boring) is the high-level review of the 2024-25 season English Premier League (EPL) as of 22-Jan-2025?

It’s over already. Granted that is what most people bar Arsenal and some deranged Chelsea fans think, but I’m afraid it is and it’s going to need a Deus Ex event to change it now.

It’s not only that Liverpool looks certain to win, but the relegation is sorted, and Champions League positions are almost finalised. There is nothing left to play for, ok there is a little battle for the Thursday night UFEA European League position (be still me beating heart).

Let me show you.

Below is a modelling of the probability of a team winning the league calculated after every game, using the outcome of the previous game and looking ahead for the rest of the season.

Manchester City can out of the gate strong at the start of the season but faded away quickly and were out of contention by Game 12 (late Nov-2023), Chelsea never really got going. Arsenal around game 17 (before Christmas) looked like making a challenge but by New Year it was fading.

This is not to say Liverpool are a better team than Arsenal. Really, they are both as good as each other.

This next chart predicts the average point per game a team is likely to win for the remaining of the season, calculated after every game.

Her you can clearly see Liverpool have been projected to get 2.1 to a now 2.3 point per remaining games. Usually in the big European Leagues, 2.2 is the ceiling. Arsenal have been playing at this level since mid-November 2023 but will never overcome the 9 point lead Liverpool has built up (assuming they win the game in hand).

For reference a projection of 2.2 points per game equates to playing 5 games, winning 3 games and drawing 2 and thus winning 11 points (2.2*5)

This chart also explains the relegation scenario. Ipswich Town, Leicester City and Southampton have been projecting 0.5 to 0.85 point per remaining game since the start of the season. There is a gap to the rest of the league. Everton, West ham and Wolves are all projecting 1 and above points per remaining game.

You also clearly see the fall of Manchester City.

But, on the bright side the strength of Bournemouth (likely Champions league place) and the rise of Newcastle are good omens for 2025-26

Oh yeah, below is a chart of the probability of gaining a UFEA European League place.

Brentford 0.1%, Crystal Palace 0.1%, Fulham 1.4%, Brighton 6.6%, Aston Villa 4.5%, Manchester City 11.4%, Chelsea 21.4%, Nott'ham Forest 15.6%, Newcastle Utd 19.1%, Bournemouth 16.2%, Arsenal 3.6%.

A Battle Royal Indeed.

My next article will be a review of the Spanish La Liga, which is very very exciting.

** Yes I used Soccer, because 538 is American