The public opinion of stats people on defensive statistics has varied over the years, or at least it isn’t very consistent. Up until Ted Knutson and I (independently of each other, I think, although there’s a chance I copied his idea and forgot) started possession/dominance-adjusting defensive stats there wasn’t much advisory thought out there about them, because not many people in the public dealt with them at all.
I personally have toned down my stance on this (possession adjusting helps a bit but not to a hugely meaningful extent), and the advice now seems to be that defensive stats are just plain hard, and the style of the team needs to be heavily taken into account. Whether this is an adjustment to the numbers or just something to bear in mind when looking at them is up to the individual.
There was, of course, a brief lapse when some people (myself included, probably) went ‘oh well if Leicester signed Kanté because of some one-off statistical category then that must be a sensible thing to do’. We have since repented for our sins.
Anyway, the point is that it’s long been said that ‘counting stats’ are far too simplistic for defenders. Looking at a striker’s shots is ok – even if you do get an Andros Townsend looking like a superstar for taking numpty shots – but tackles or tackle success percentage, for example, are different.
But, even when including interceptions, we (well, I) have never been counting all of the defensive actions that it makes sense to count. As I understand it, if a player attempts a tackle and misses, this goes down as an unsuccessful tackle – but if they foul the player it goes down as a foul instead, not in addition to. Awkwardly, not all fouls are missed tackles, but given the amount of fouls which are likely to be bad tackles, it would make sense to include it.
As well as this, blocked passes are something that you’d be forgiven for not having heard of. I wrote about how Opta-defined interceptions encompassed a range of different types of interceptions here, and blocked passes are almost a separate subset of them (the name is self-explanatory, but I think they’re passes that are stopped but possession isn’t kept, or something).
A collection of these defensive actions has similar problems to using tackles and interceptions alone, of course, but logically at least it makes more sense. You are finally, and even so not wholly accurately, actually measuring all of the things that you want to be measuring.
Of Premier League centre-backs to play more than 5 lots of 90 minutes, the top 10 for the new ‘Defensive Actions per 90’* looks like this:
Nicolas Otamendi – 6.11
Curtis Davies – 5.70
Eric Bailly – 5.67
Jake Livermore – 5.28
Cesar Azpilicueta – 5.21
Shkodran Mustafi – 5.16
Papy Djilobodji – 5.00
Virgil van Dijk – 5.00
Jordi Amat – 4.93
Laurent Koscielny – 4.92
*only starting minutes at centre-back are counting; and clearances aren’t counted within this defensive action grouping because even when possession adjusted they correlate positively with conceding more shots.
As mentioned earlier, the same problems as using tackles alone applies. Where would sheltered centre-backs like Toby Alderweireld and Gary Cahill come? (Azpilicueta, even when starting at centre-back, operates higher and in different ways to Cahill, which is why he appears in the top 10). The answer, is that Cahill is 50th in the list of 66 and Alderweireld is 59th.
As for the top 10 for a composite ‘Defensive Action Success Rate’ – successful tackles, interceptions, and blocked passes on one side with unsuccessful tackles and fouls on the other – things change a lot (as would be expected).
Gary Cahill – 92.38%
Nathan Aké – 86.49%
Curtis Davies – 85.53%
Mamadou Sakho – 85.00%
Phil Jones – 84.38%
John Stones – 83.70%
Joel Matip – 83.17%
Maya Yoshida – 82.61%
Angelo Ogbonna – 81.94%
Steve Cook – 81.43%
There are some notable guys just below the top 10 who were in the other list too: Koscielny is at 11, Livermore 12, Van Dijk 14, all on 80+%. Alderweireld, seeing as I mentioned him earlier, is at 30 on 75% (a pip above the sample average of 73.9%).
Again, I am required to point out the drawbacks. Gary Cahill, the best in the league by a distance? That’s odd, and doesn’t account for the fact he seems to be the least fancied of Chelsea’s 3 starting centre-backs among the general public (if not the PFA team of the year voters).
Curtis Davies appears highly on both lists, and I believe he’s stood out by basic defensive stats in previous seasons too – but this isn’t necessarily to say that he’s a hidden world-beater. The absence of Toby Alderweireld, considered by a lot of people to be the best CB in the league, from either top 10 is the opposite. Alderweireld is clearly a good player, and the fact that he doesn’t show up means we have to think a lot about why – but maybe that’s something for another article.
Finally, we go back to Cahill at the top of the Success Rate list, by nearly 6 whole percentage points. He doesn’t make many actions, but when they do they’re successful – this sounds a lot like a defender in a relatively deep defensive bloc, and that sounds a lot like the type of defender Cahill is.
It’s possible that he is doing so well at the moment because the system he’s in, when defending, matches his preferred style of play quite well. This then raises the question of how you interpret stats of good players who are in a system which doesn’t suit them but, again, that’s a subject for another time.
This article was more to make the point that using tackles and interceptions alone doesn’t quite make logical sense, because not all tackles and interceptions are included in those categories, rather than to say that any particular centre-back was good or bad. Stats are a funny ol’ thing, eh.
This article was written with the aid of StrataData, which is property of Stratagem Technologies. StrataData powers the StrataBet Sports Trading Platform, in addition to StrataBet Premium Recommendations.
Eastbridge Soccer Betting Broker, bespoke sports betting for the discerning client.