clock menu more-arrow no yes mobile

Filed under:

Premier League Game Projections, Week 35

A statistical projection of the games of week 35, with bonus plus-minus analysis

Clive Rose

After a tough match with one of England's elite sides last week, Tottenham Hotspur will travel to face Wigan Athletic for a must-win week 35 fixture. While this seems a very favorable fixture, Spurs have already seen the effects of losing a must-win after their pitiful performance against Fulham several weeks ago. On top of that, Spurs were beaten 1-0 at home by Wigan earlier in the season.

I think fans should consider the Fulham game as a reminder that this club is entirely capable of a true stinker. But I'm going to argue that we shouldn't worry too much about the precedent of the previous Wigan match in early November.

Spurs played that game without either Moussa Dembele in midfield or Hugo Lloris between the sticks. While I think any argument that Gareth Bale isn't the best player on this team is just anti-social contrarianism, a plausible argument can be made that either Dembele or Lloris is at least as important as Bale. This roster is very thin in central midfield, and Dembele is both the club's best defensive midfielder and the source of our best direct attacks out of midfield. In goal, Brad Friedel is not a bad keeper, but he's simply a poor fit for anyone who's going to play a high line and compress space in the center of the pitch like Spurs do. Lloris, by contrast, appears to have been created in a lab specifically for the purpose of playing in such a system.

This subjective analysis can be partly confirmed objectively. I'm using the same kind of plus-minur analysis I did last week to look at Gareth Bale, but now applying it to Lloris and Dembele. Spurs played 20 games with both in the squad, 7 with neither, and 6 with only one or the other. (This isn't quite correct, since Dembele has both come on as a sub and been substituted later in games. My data isn't time-stamped, so I am categorizing games based on whether or not Dembele started. This isn't optimal, but it's the best I can do.)

With both our first-choice keeper and our best central midfielder, Tottenham are a truly elite club, nearly 50% better than league average on both attack and defense. Without either Dembele or Lloris, Spurs have been roughly the same quality as Fulham, about 10% worse than league average on both attack and defense. So the side that lost to Wigan in November is just radically different in quality from the side that is expected to take the field at the DW on Saturday.

What I find most striking is how similar the overall numbers are when either Dembele or Lloris isn't starting. You would expect the loss of Dembele to affect the attack and defense equally, since his responsibilities are evenly balanced between the two. And indeed, Spurs have been somewhat worse in both areas when Dembele has sat, measuring about 25% better than league average on attack and 15% better on defense. You might expect the +/- Lloris numbers to be more heavily weighted to defense, but instead they're practically identical to the numbers with or without Dembele.

We're looking at a very small sample here, so this could just be random variation. I think we're seeing a real effect though. This result, suggesting that Hugo Lloris affects Spurs' attack as much as their defense, is exactly what Jonathan Wilson described in his column on Tottenham, Bale and Lloris back in February:

So why are Tottenham more compact than they were five months ago? Why does the defence feel able to push higher? In part that is down to confidence and to Villas-Boas's training. But it is also down to the change of goalkeeper. Brad Friedel remains a very fine keeper but he is part of the reactive school of US keeping that tends to stay deep. There is nothing wrong with that – it is a perfectly legitimate way to play – but it does have ramifications elsewhere on the pitch.

If a team want to play with a high line they need a keeper who is comfortable coming off his line, sweeping up behind the defence, making sure the space behind an advanced back four is not a yawning void into which opponents can play through balls. Hugo Lloris is much more adept than Friedel at that and, since he has become the regular No1, Spurs have played much higher and been much more compact. That has got the best out of Bale. (It has also worked defensively; no team in the Premier League concede as few shots on goal as Tottenham.)

Lloris makes the whole system tick. The forward players can launch their attacks at speed coming out of a tough press only when the defense can effectively play a high line and shorten the field, and the back four can only do that when a true sweeper keeper is ready to pounce on balls sent behind the lines.

Here is the data in tabular form on Spurs with or without Dembele and Lloris:

Lineup Team+ Att+ Def-
Spurs w Dembele and Lloris 150 142 69
Spurs w Dembele wo Lloris 125 130 90
Spurs wo Dembele w Lloris 124 126 88
Spurs wo Dembele or Lloris 86 94 115

Week 35 Projections

Ok, so what are the numbers on the games? Like Spurs, Chelsea face a must-win home to Swansea, while Arsenal have a match roughly equivalent to Spurs-City last week. They face a Manchester United side who have been on brilliant form, but who aren't playing for anything besides pride and a paycheck. The season projections have already accounted for the likelihood that Arsenal will drop points, so a draw or even a loss would not put them out of contention.

Club xG xGA %W %D #L
Arsenal 1.8 1.8 37% 22% 41%
Chelsea 2.3 1.1 64% 19% 17%
Tottenham 2.3 1.4 58% 19% 23%

Because the race is so tight, "must-win" isn't exactly the right terminology. No one is going to be eliminated based on this'week's results. But a loss in a should-win game makes the next should-win an effective must-win. And no one wants that to happen. (I mean, I want that to happen to Chelsea. But you get the idea.) Here are the effects of the different game outcomes for each club on their respective top four hopes:

Outcome W D L
Arsenal 85% 67% 55%
Chelsea 79% 58% 47%
Tottenham 73% 52% 40%

For y'all in the prediction league, the most common projected scores in Tottenham-Wigan were 2-1 (9%), 1-1 (8%), 2-0 (7%), 3-1 (6%), and 1-0 (6%).