Welcome to Expected Villa (xV), a post-match column on the stats behind Villa’s results.
One of the reasons I really love looking at football through a mathematical lens is that it allows us to sidestep our conventional, results-based bias to take a look at how the match actually played out. Did it play out exactly how we saw? Which moments are emphasized?
Villa played really, really well, I thought
The real only time I was worried Saturday was when Villa stupidly gave the ball away, but then Barrie McKay decided to shoot from 30 yards out and everything was fine. I guess Kieran Dowell’s knuckle-balling chance was a little concerning, but not from shot quality, rather just from the nature of how the ball was moving. The xG figures weren’t that huge on Villa (more on that later), but I was really impressed by the way Villa held Forest to no clear-cut chances, despite giving up most of the ball.
This was a good tactical day from Steve Bruce, in part because it really was, and in part because it worked. Villa picked a handful of times to attack, but when they did, they created incisive, high-quality chances — be them “better” chances like Scott Hogan’s goal or John Terry’s header of the ball, or ones like Robert Snodgrass’ free kick or Jack Grealish’s try from outside the box. Villa probably generated the four or five best opportunities in this game, and that’s why they won.
Aimless possession is stupid
If you’ve followed me on Twitter for a while, you’ll remember my semi-regular #obsessionwithpossession tweets — first about Villa supporters, who always wanted Paul Lambert’s counterattacking sides to play with more of the ball, and then about Lambert himself, who suddenly decided to give up a tactic that perfectly suited his team to aimlessly pass the ball around and do nothing with it.
Possession-based football is great in theory and great when you have the talent to use it to actually control the run of play. But it’s largely worthless as a statistic without context, and Forest is a good example why. Forest had 60% possession Saturday, which they’ve now done 10 times. They’ve lost eight of those matches. Villa were more than content to let Forest hold the ball, because they knew the hosts were never going to do anything with it.
Never rely on single-match xG
Whenever I talk about single-match expected goals total in this column, I always try to contextualize it and “confirm” the values seem accurate given what our eyes saw. That’s largely because of the limitations of xG that we’ve talked about a lot: who the chance falls to, where the defenders and goalkeepers are positioned, how good the ball setting up the chance is, etc.
In more scientific terms, though, single-shot xG figures have a pretty high margin of error — the computer doesn’t know all those factors above, it just kind of has to hedge its bets and guess somewhere in the middle. Our eyes can help us tell if the computer’s made the right assumptions, or if the computer hasn’t.
On Saturday, I don’t think it did. I was a little surprised when looking at that Experimental 361 xG chart (linked again) from the weekend to see that Terry’s headed chance off the bar was only given about 0.1 xG, while Hogan’s was in the ballpark of 0.3 xG. Both were headed chances and both were taken from similar positions, a pair of factors you’d expect to see produce similar xG totals.
The potential key? Opta coded Hogan’s chance as a “big chance,” while Terry’s wasn’t. These “big chance” flags can help the model infer some of the information we were missing above — if Terry’s chance isn’t flagged that way, the model might assume it was a more difficult chance than it actually was.
(The other reason not to rely on single-match xG is that we know both Hogan and Terry’s chances had a better than 30% and 10% chance of going in, because we watched the match and saw that Hogan only had to head into an open net, and that the keeper was beaten if Terry put his header on target. Again, this is fuel for the argument that Villa were better than the xG chart suggests.)
Just get to within an arm’s reach with five to play…
Aston Villa host Cardiff City on matchday 42, then host Derby County in the penultimate week of the season. Given this is true, Villa just need to focus on getting their results and being within two or three points of both the Bluebirds and the Rams heading into the final five matches.
The fixtures are favourable for Villa, too. Their next three home matches are against three of the division’s biggest strugglers — Barnsley, Burton and Birmingham City — and that’s just the start: 10 of Villa’s next 14 matches are against teams in the bottom half. There’s no reason the Claret and Blues can’t do a couple points better than Cardiff and Derby in this run, leaving them with everything to play for at season’s end.*
*There’s also a good case to be made that Villa’s upcoming run of fixtures means sitting in second and being chased after matchday 41 is very plausible, which is true, but I’m not going to get carried away here given how an inexplicable poor run of form in March, with a loss away to Sunderland and draws against Bolton and Hull, would seem so fitting for this club.
Birkir Bjarnason played really well as a defensive midfielder
And I’ll just let Jack make my argument.
He’s super versatile, which is super valuable. This was my argument for keeping Leandro Bacuna at the club. Bjarnason is better. Let’s keep him.