Are Barcelona really worse off when Xavi plays?

Yesterday from friend of the blog 11tegen11:


The link is to this article. In summary the author has identified a set of players they deem to have positively or negatively affected their team this season. Amongst the more surprising claims is that Barcelona are a worse team with Xavi on the pitch than without. Fortunately I’ve recently run some numbers for La Liga so it’s easy for me to test this hypothesis.

First off I’ll say that I think this kind of ‘with or without you analysis’ (a tangotiger concept that has been around since at ~ 2006) has massive potential for teasing out the contribution of individual players. I even did an analysis that was remarkably similar to that in this article a few years before I started writing this blog, using the same ten game cut-off. Amongst other things it told me United were worse at attacking when Cristiano Ronaldo played. For reasons including this I decided I needed more detail before I could make anything of it.

So to some analysis. First the traditional counting numbers:


Well what do you know; Barcelona’s results were extraordinary without Xavi and merely very good with him. Maybe he is a detriment.

How about some of the more advanced numbers I usually use to evaluate teams?


An explanation of total shots ratio can be found here and here, for PDO see here and here, and for qualcomp see here.

Oh well, that was fun while it lasted. Barcelona are basically the same team regardless of whether Xavi plays. They take a similar ratio of shots, hence controlling the ball in the attacking zone just as often. Their opponents had similar total shot ratio’s, suggesting they were of even strength.

What changed is so obvious that the highlighting probably wasn’t necessary. If one of Barca’s opponents put a shot on target in a game Xavi started then it was 47% more likely to result in a goal than in a game he didn’t. Barca themselves scored with 10% more of their shots in games Xavi didn’t start. Try pinning that on Xavi. Seriously, give me a viable working theory, I’d love to hear it.

I guess in summary I’m saying that whilst the numbers may not suggest that Barca are better with Xavi on the pitch, there’s certainly nothing here to suggest they’re any worse.

The crazy thing is that I’m doing this in an hour with some pretty basic, freely available data and using maths that a seven year old is capable of. The post I’m responding to was written by someone who is in the privileged position of having access to opta stats and who’s website claims to be “revolutionising football statistics“. It’s truly maddening just how little of the surface is being scratched here.

9 thoughts on “Are Barcelona really worse off when Xavi plays?

  1. just read the other guy’s blog … what frustrates me are a number of things …

    1. There is no differentiation between games … ie. maybe all games without Xavi were at home (one extreme) and I’m sure that there is a difference in home/away performance. Furthermore, were all games without Xavi against the bottom teams (ie. Racing)? That night explain it.
    The point is that games with/without have to be weighed in some way that relates to opposition quality, location, etc. …
    Your Qualcomp goes a way towards this, especially if we know what kind of deviations we could expect … ie. is the 0.7% difference above a lot or not?

    2. This really is a small sample size … if Xavi had did not play in 6 games for Barcelona and the record was 6W 0T 0L would the conclusion be any different? Hey they have a perfect record without him! Never play the chump! 🙂
    What would have been better is if the guy had taken ALL games over the past seasons in which Xavi could have played and then balanced out the results. Xavi has been a regular since around 2000 … now that’s what I call a decent number of games to draw a conclusion from.

    As for opponents scoring more with Xavi on the pitch …
    I’m assuming the same guy is in goal and that Barcelona play the same style and thus have the same possession stats etc. with/without Xavi.
    – Maybe Xavi’s weakness is his inability to transition quickly from an offensive to a defensive position thus when there is a loss of possession an opposing team can score easier on the break because he doesn’t get back and help out the 3-man defence enough … therefore maybe the attempts on goal come from better scoring positions such as a 3 on 1 or 2 on 1 break.
    The player that replaces him could be faster, more defensive minded, less likely to find himself high up the pitch etc.
    Xavi scored 10 goals this year, highest of his career, so he must have been higher up the pitch on average (better scoring position) or been insanely accurate with his distance shooting. A quick check of ESPN Soccernet stats (37Shots, 16SoT, 10G) this year, last year (34Shots, 11SoT, 3G), the year before that (45Sh, 19SoT, 3G), the year before that (58SH, 16SoT, 6G) etc. … would have to look more in detail as to where the shots/goals came from to really make a solid conclusion.

    As for Barcelona having a better goals/shot% without him …
    no idea.

    No idea, grasping at air here really, that’s all I can come up with. 🙂

  2. ah … forgot the following … this really irritates me about the other guy’s blog …

    3. Two scenarios, one real and the other fictitious but the point is; how are they integrated into the stats …
    a) Real: Barcelona played away at Real Zaragoza this season and won 1-4 … The game started with Xavi on the bench, he came on in the 90th minute at which point Barcelona were winning 1-3. In the second minute of extra time they scored their 4th goal and a minute later the game ended.
    Xavi on, 1 shot , 1 goal … Is this counted as a Win WITH or WITHOUT Xavi? Is it 4 goals for Xavi or 1 for and 3 against? (the other guy uses goals per game stat)
    b) Fictitious: Barcelona are playing at home against Racing, Xavi starts for them but gets injured in the 60th minute when they are leading 1-0, he is subbed and in the remaining 30 mins of play Racing miraculously score 2 goals and win 1-2! 🙂 Is this counted as a Loss with or without Xavi? How are the goals allocated? He left the field in a winning position with 0 goals against, maybe the game counts as a Win for Xavi?

    Anyway … you get my point … 🙂 A game is 90 mins and hardly any player completes the full 90 mins each game and there could be changes in goals/game state before/after they come on/leave the field … how are those changes incorporated into the other blog’s little stat table?
    One option he could do is change to minutes per goal, that’s a bit better than goals/game.

    What is especially irritating is that blog post presumes is that the player analysed is KEY to a team. Therefore, that player leaving the field or entering the field would be a KEY moment in a game and that blogger should assume it to have an impact as he uses the with/without premise … aarrrgh! Just makes me want to tear my hair out. 🙂

    I don’t feel like signing up to that blog so can’t post the above there … ah well.

  3. Hi Bart. I’ll come back to the other points later but just wrt qualcomp. A team with a TSR of .494 would expect 51 points a season, with a TSR of 0.487 they expect 50. There’s a margin of uncertainty here but they’re essentially equal

    • cheers 🙂

      I don’t have twitter … but I read that the guy who wrote the article twitters … “exactly, I didn’t say worse” but he did in the article … he relates Xavi’s with/without numbers “The difference of 1.21 fewer goals scored per game is just behind that of Gameiro” and just above he writes the following …

      “meaning the Parisian team score 1.32 more goals per game in matches where Gameiro did not start – the worst rate of any non-defender in Europe’s top 5 leagues to have started at least 10 games and also not started 10.”

      Just below he puts … “When it comes to positive effects …”

      AND he starts the article by saying … “who exactly is responsible for their side’s peaks and troughs in form … hindsight gives us a frustrating yet extremely useful chance to look back over the season to see where it all went so wrong. Here, we do exactly that, focusing on the differences in teams’ results when they have, or are missing, the players that have the biggest effect on their performances.”

      Put 2 and 2 together and yes he does say worse. 🙂 Unless he defines a peak as a time when a team scores less goals, concedes more and gets a lower win%.

      Maybe I’ll sign in to that site now and rip the writer and that M.Laurence guy new ones cuz man oh man … idiots.

  4. Pingback: Liverpool’s crossing addiction in 2011/12: a desperate measure? | 2+2=11

  5. I’m trying here: when Xavi doesn’t play Barcelona have less control over the match, so the match is different. The opposition is able to hold possession a little longer and can create some chances with more build-up, as opposed to the counter-attacking, one-on-one with the goalkeeper stuff. This could account for a lesser conversion rate. This loss of control by Barcelona however, also lures out the opponent, and with Barcelona’s attacking power they in turn create more clear-cut chances. Barcelona’s conversion rate goes up. The ability Xavi has to control a match, you could say, is just too much. However, in my opinion, this logic does not apply to matches against truly great opponents (El Classico or Champions League). Here Xavi’s ability to control is one of the main reasons Barcelona have been so consistent over the last 4 years.

    Are there possession stats for Barcelona with and without Xavi? Significantly greater amounts of possession with Xavi could back my theory up.

    Again, I’m just trying to think of something, I’m a huge Xavi fan.

  6. Coming into this very late, having read your most recent blog post, but another factor that doesn’t seem to have be taken into consideration is the fact that the matches in which Xavi is not involved are, injury/suspension aside, generally those against weaker teams that Barcelona expect to beat and can therefore afford to rest him for.

    • If that were the case then the Qualcomp would be vastly different. As it is the opponents when he did play would expect to score 51 points over the season, whereas the opponents when he didn’t play would expect to core 50. In other words they’re essentially the same.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s