Get Your Stats Right: August 2024

Our new monthly column looking at all the stats and data behind the U's

It’s an absolute pleasure to have the brilliant Ben Griffis writing these new monthly columns for UTAS, delving into the numbers that make up the U’s performances using his own statistical data and modelling. I’m sure you’ll agree this is a fascinating start with just a small sample size, made all the more impressive by the fact Ben is a US-based U’s fan. We’re massive now, it seems.

Give Ben a follow to discover more of his brilliant work, and come back next month for the second instalment.

---

Three games played, one point, 20^th place… but it doesn’t feel as bad as it looks on paper so far for the U’s. There have been some wonderful stretches of performances not just in these opening 3 league games, but also in the game against QPR. But what’s happening “under the hood” of the results? That’s what I’ll be diving into across this season using a more stats-based view. What’s happening, how’s it happening, and maybe a little insight into the “why” of it all.

While I’ll introduce a fair few numbers here, don’t worry: I’ll explain everything I talk about, link to good resources for more deep dives if it piques your interest, and try to keep it all grounded and not claim United are the best team if the table shows 24^th, or the worst team if it shows 1st (at that point, there’s no such thing as “statistics”… only vibes).

You’ll be seeing my words come across your phone, tablet, computer, and more as the season progresses. As more games are played, richer data becomes available and we’re able to actually start drawing conclusions. So, for this first piece just at the end of August, I want to chat a bit about some basic numbers I’ve found interesting in this early U’s campaign.

Basic Statistics

The most basic statistic in football is goals. United scored fewer than Stockport, fewer than Crawley, and the same number as Blackpool. That puts us 20^th.

That concludes this article, see you next month, right?

Of course not, let’s build a little bit of a base so that we can start investigating, especially in the coming months, perhaps how and why those goals are coming (and I expect them to come much more frequently this season than they have recently).

Possession is a foundational concept in data, showing who had more of the ball in a match. While possession is pretty correlated with points per match, it’s not a given that high possession teams win all their games. Check out Las Palmas in last season’s La Liga who finished 16^th despite ranking 2^nd in the possession tables, with 60% (only behind Barcelona).

Last season, United ranked 21^st for possession in League One, with an average of 44%. This number was 46% in 22/23 and 44% in 21/22. This season it’s nearly 48%, ranking us solidly mid-table (as an aside, much of the info I’ll talk about here is available on my free web app if you want to play around yourself).

These may be small differences at the moment, but moving closer to 50% possession is a very positive mark on United and on Garry Monk’s tactics. One of the chief reasons possession is so correlated with results is that not only does having the ball help give a team more opportunities to score, but it also helps keep the ball away from their own net.

Of course, there’s more to tactics than that. It’s what you DO with possession that matters. And I think the wild comeback vs Blackpool illustrates something we’re all feeling that we may not have felt as much in recent seasons: United can actually score a few goals now. Of course, the U’s were kept scoreless in the opening two fixtures, but the attacking flows are there, even if the finishing may be naturally inconsistent early on.

Advancing The Statistics: xG

This naturally brings us to a little something called “expected goals”. While a common term now, if you haven’t gotten a good explainer yet let me try to remedy that. Expected goals, almost always abbreviated to xG, is essentially the probability that a shot becomes a goal. A shot with an xG of 0.05 can be thought to have a 5% chance of going in, and a shot with an xG of 0.55 can be thought to have a 55% chance of going in. xG is never perfectly 0 (literally never a goal), since there’s always a chance a shot goes in, and never 1 (100% a goal), since I can guarantee that if you stuck me in front of an open goal, ball sitting on the goal line, I’d still find a way to sky it.

We also add xG up for both players and teams. This added sum of xG can be thought of as the number of goals a player/team “should” have scored. So, if a team generated 1.2 xG in a single game, we might “expect” them to score about 1 goal. Of course, a tricky little concept called “variance” means that the actual goals in a single match can often be a bit higher or lower than xG. But know that, over a longer time-frame, xG and goals do tend to converge.

Alright, enough explainers, let’s get back to the numbers. In this article, I’ll be sharing my own xG model’s numbers. It will differ from Wyscout or FotMob, just like how Wyscout and FotMob differ from each other (sometimes by a LOT), since they are all different models.

United rank 13^th in terms of xG per game, with about 1.5. This means that, on average, we could expect the U’s to score about 3 goals every 2 games. Overall, my model has the U’s total xG at 4.7, and we’ve scored 4. For reference, Stockport’s total xG is 7.5 and they have 7, Shrewsbury’s total xG is 2.0 and they have scored 1.

We can see United are 12^th for possession and 13^th for xG per game, which is a really good indicator even if it’s so early on in the season. However, the table below clues us in to a little bit of the negatives of the opening performances.

United’s expected goal difference per match (xGD, the sum of xG and xGA, or expected goals against) is -0.23, which is a pretty decent number all things considered. We can’t expect the U’s to be top-table in this respect, however amazing that would be. But the actual goal difference per match is -1. That essentially means that, through poor finishing or leaky defending, the U’s are under-performing by a bit in these 3 games. That’s simultaneously positive and negative, as it’s possible these results are not representative of our future results… but it’s also never good to under-perform by so much (only Bolton, Blackpool, and Rotherham have under-performed by more so far).

In terms of individual players, United actually have a player in both the top and bottom 20 in terms of their non-penalty xG over- or under-performance. Danny Andrew has over-performed by 0.9 (although it’s so early in the season that any defender scoring a goal will likely over-perform by a lot…) while Sully Kaikai has under-performed by about the same margin. It’s a wash. This tells me the under-performance is not due to a single player missing chances but instead likely a team-based issue both going forward and in defence. I personally prefer a system-based issue than a player-based issue, but with the caveat being system-based issues only get sorted out if they’re known, player-based xG under-performance usually corrects itself over time.

Taking Another Step: xG Buildup

I’m trying not to spend too much time on overly-advanced things in this first monthly stats review, but there’s one xG-related metric I’m very excited about that I want to include. xG Buildup. This is essentially the “quality” of a player’s buildup-play, or rather, how they are able to consistently facilitate good-quality possessions for their team. Please read this StatsBomb article for a much deeper explanation. The math behind xG Buildup is simple. We sum up the xG of every possession a player is involved in (with a recovery, pass, or shot), and then subtract the xG from their own shots, as well as the xG they assisted. This number then is the player’s xG Buildup, or the xG they have been involved in helping to indirectly create. Messi, Xabi Alonso, and Toni Kroos are 3 players who always had very high xG Buildup, for instance.

United have two players in the top 20 so far this season for xG Buildup per 90 minutes. Danny Andrew ranks 8^th with 0.97 xG Buildup per 90, and Paul Digby ranks 17^th with about 0.87 xG Buildup/90. The reason I love to see two United players so high this early in the season is that it highlights the ability the U’s have to create quality possessions. Going back to my opening bit about solid possession numbers, it’s not just possession for possession’s sake, for United it does seem to be quality possession.

The Final Note: Field Tilt

The final piece of data I want to share for this article relates to possession, and shows possession quality in a slightly different light: field tilt. If you’re not familiar with the concept, field tilt is essentially the possession number but only considering possession in each team’s respective final third – the area that almost every goal is scored and/or assisted from. While the defensive third is where the foundations of an attack are set, the final third is where they come together.

Field tilt is, mathematically, the number of passes a team has in their final third divided by the sum of their final third passes & their opponents’ passes in their respective final third (the focal team’s defensive third). Here’s a nice article with more detail if you’re curious. Like possession, field tilt is naturally correlated with more points.

I’ve got slight concerns about United’s field tilt, although there’s still not enough games for me to conclude anything yet. Below is a table which shows the U’s field tilt per match, as well as how our field tilt has compared to possession.

While we rank 12^th for possession, having about 48% of the ball each match, we rank 23^rd for field tilt, with about 39% of the share of possession in the final third. United’s field tilt is about 8.7 percentage points below their possession, ranking 22^nd in this respect.

The reason this concerns me is that the possession is deep. Early days, of course, but these numbers hint at an initial difficulty of breaking opponent lines and pushing into good areas of the pitch consistently. We’ve seen the quality of possession that Andrew and Digby can help create, but there’s still tons of room for improvement so far. In fact, United’s field tilt has been lower than their possession in each of the opening 3 league games. Possession needs to be translated into field tilt, into xG, for it to be “useful” to scoring more goals.

Wrapping Up

Hopefully the data I’ve shared so far paints a picture of the *potential* added attacking threat that we now see from the U’s this season. But it also hints at the club’s performances teetering on the fence at the moment. If Monk, the fans in the Abbey & those traveling each week, and maybe sometimes the opponent, can give the squad a little nudge into the “green grass” side of the fence, we could be seeing a new-look U’s for 24/25.

The transfers give me hope. The manager’s tactics give me hope. The performances (albeit maybe not the results… yet) give me hope. And the initial underlying data gives me hope. The trick is to keep up the great work, get all the new players gelling, and keep improving the system. Easier said than done, of course, and I’ll be here with you this season each step of the way sharing cute graphs and data illustrating the process.

If you’d like to follow along each week, I’ll be updating my analysis web app each night and in there you can find single-page statistical reviews of matches, league rankings for many statistics, match-by-match breakdowns of each statistic, and more.