Arsenal are massively undervalued to win the Premier League
I’ve been a long term non-believer in Arsenal. Throughout the last 20 years they have been the team easiest to wait until hyped up or on a good run of form and then bet against them. Their fanbase is as rabid and deluded as Liverpool’s or Man United’s but unlike them the team has never put it together sufficiently to get over the line.
I think that’s the reason we see a current trend to people believing that Arsenal are in the process of throwing away a Premier League title. Recent losses in FA and League Cup and a slight downturn in form coinciding with a Man City win streak has pushed the odds at time of writing have them only as a 58% favourite to lift the trophy at the end of this season.
My own odds put them at around 90% and after some review and analysis I stand behind that Arsenal optimism.
1. Arsenal are very good, Man City are not
I run an evolutionary (proprietary and hopefully unbiased) model for rating team’s expected performance in future matches based on a variety of factors. The output of that model is XPI this is a ratio of the likelihood in a game vs a random opponent involving the team whether a goal scored is more likely to be scored by them or against them.
Arsenal this season have been sensational, while their peak XPI (2.3157) this season is not the highest since I began ratings in 2023 they have been great. They have held for the vast majority of this season both the titles of best offensive team home and away and best defensive team home and away.
They’ve been able to look so good this season thanks to the distinct lack of quality from other teams that compete for Champions League spots.
That includes Man City, they currently sit at a dismal XPI of 1.4749. While they retain a tier as a clear second best team in the league they have fallen hugely since the end of last season, a season where they weren’t good enough to even finish top 2. On global terms they’re not that much better fixture to fixture than teams like Man United or Liverpool.
While Arsenal’s form has wobbled in the last few weeks and Man City have seen a slight uptick, the gulf in class is vast. Arsenal are still putting up great xG ratios in their games where you will hear commentators describe them as “outplayed”.
My ratings however are always playing catch-up and I can understand why those who watched the League Cup final might believe that Man City have a huge edge heading into Sunday.
That brings me to:
2. It doesn’t matter who wins on Sunday
Obviously I’m being facetious, it does matter who wins between the top two in the Premier League in deciding the outcome, but interestingly it matters less than I’d expect. I ran 25,000 simulations for each scoreline below to see the resulting chances in Arsenal winning the title:
You can see that the model predicts Arsenal will be gigantic favourites should they get any type of result and they’ll still be 3 to 1 favourites should they succumb to a defeat. I think that most people would expect Man City to be favourites should they win on Sunday. The reasoning makes sense: Arsenal will be floundering and City will be motivated and on the charge. I heard a commentator say as Arsenal were losing to Bournemouth that “you look at the remaining fixtures and it gets harder to see where they’ll next get 3 points from”.
But let’s look at the remaining fixtures for both teams:
FDR is Fixture Difficulty Rating, this is my model again but independent of how well I rate Man City or Arsenal, just the Home and Away XPIs for their opposition. 1 Represents the absolute median fixture difficulty.
What you can see is barring a couple of mid table away games neither side really has any difficult fixtures left, City’s average fixture is slightly harder.
We can sit here all day and make arguments about fixture congestion and who has what to play for but on average I think City and the FA Cup and additional fixture roughly counteracts Arsenal and the Champions League fixtures.
Given my model’s heightened assessment of Arsenal we can see that their expected points from the remaining fixtures is slightly higer. But that comes with Arsenal already having 6 points more on the board. Even if they forfeited the result, my model predicts that Arsenal would still finish 3 points higher.
I’ve heard the accusation that Arsenal are flat track bullies and only beat significantly inferior opposition. Well good news for them that’s all they’ve got left to face.
3. Even if I’m wrong- I’m still right
Well what if my model is wrong? I know it’s not perfect. Perhaps you view Arsenal as nowhere near as good as I do. I can understand that.
So I decided to run some other modelled situations:
What if Arsenal and City were both as good as each other?
If I model them them both as bad as City? Arsenal win 79.9% of the time
If I model them both as good as Arsenal? Arsenal win 80.4% of the time
Hmmm okay, how about if my model is disgustingly wrong, what if I put Arsenal and City in each other’s shoes?
If I model flip their XPIs? Arsenal win 66.0% of the time
Okay fuck it, what if Arsenal and City swap XPIs and City win 3-0 at the weekend?
City win the title 54.9% of the time
If I’m that far wrong I’m happy to be humbled. City may obviously win the title in any scenario but that doesn’t change the odds currently from being hideously skewed from reality.
