Fulham go moneyballing for gold

Football and analytics fight!

This has been going on forever in baseball, so no shock to see it happening here, now, either.  Worse, much of what the likes of Bill James discovered in the 70s and 80s was more or less indisputable, but it still took until relatively recently for his ideas to take hold.  Now all baseball teams (it really is all now, I think) embrace analytics, looking for an edge anywhere they can find it.

Football is harder.  Anyone unconvinced with the whole notion will tell you that: instead of discrete batter v pitcher matchups, football is a fluid game with indefinite happenings.  Chain reactions abound. There’s more noise than… not noise.  It’s a minefield, it really is.  But still, you look for edges where you can find them, right? Better to spend £250k on a middle of the road midfielder from Slovakia than £2m for a middle of the road midfielder from Leeds.  If you can pull it off.

Anyway, much excitement on twitter over this recent piece.  I won’t go into the details, but what fascinates me is what the powers that be can have seen in Madl and Mattila in the first place.

Now, we can’t retro-fit these two at all.  The data that must be in use is not in the public domain as best I can tell, so we must go top-down (e.g. look at a team level).  I don’t think this is the worst idea, anyway: after all, a top down approach means you’re looking at what happened to the team the player played in.  It takes a brave man to see a poor defensive side and decide that there’s a centre-back in there who’s doing a fabulous job.

So…

I can’t see anything.  Have a look at these numbers, which are the games Mattila played in before we signed him.

Mattila.png

This is what I’m getting at.  Here we have a team that, in 2015, was beaten by five goals twice and four goals twice.  Now that could be a systematic failure from any number of perspectives, but from an analytics POV, what can you see in there that makes you think a defensive midfielder is worth a punt?  He’s making lots of tackles?  I bet he is! His team’s being attacked relentlessly.  He passes well?  Maybe, but if so he must be a lone beacon of positivity in what seems to be a very poor side.  We know, we think, that even the best players are only worth a few points over an average player a season, so it’s possible he has been playing brilliantly, but using a top-down viewpoint at least, it’s a mess, to the point where I’d be nervous about unpicking an individual’s contributions.

The sensible retort might be that the team has used several years of data, which I hope is the case.  The comeback might be that he’s bounced around from team-to-team for a while, only settling in his recent stint at Aalesund, which itself is confusing.  There must have been pedigree there or Udinese wouldn’t have bought him in the first place (they are scouting masters), but after that, nobody seems to have been convinced.  A true diamond in the rough if he works out.  From here, it’s hard to see.

Michael Madl is half-similar: he’s come from a team (Sturm Graz) that appears to have been comfortably top half for the last couple of seasons.  They’ve scored and conceded goals at a perfectly normal rate in the last couple of seasons.  So what do we take from that?

Madl

This uses standard deviations to unpick where a team’s strength lies.  I can see here that Sturm Graz have been a better defensive side than might initially have been guessed: they were a pretty good side in 2014-15, and much of that was because they had the second best defensive team in the league.  Now, sometimes there’s an alarm bell around this, because the attacking play isn’t all that.  What it could mean is that this is just a team that sets itself up defensively (e.g. keep men behind the ball even in possession): goals dry up and so any top-down measure will make defenders look good and attackers less so.   This could be what we see here, as the Sturm Graz games have been among the lowest scoring in the league for the last couple of seasons, implying more closed, cagey games.

All of which is a long way to say that I really don’t know.   Madl, I can see, maybe there’s a sniff of something, and if Fulham got Opta numbers to probe further maybe they did see that Madl’s leading a defence that is perhaps a bit better than average in Austria.  It’s a reach though.   As noted from the outset, it’s very hard to unpick an individual contribution from his team’s performance, and there’s nothing I can see (which doesn’t mean much, of course!) that suggests we might be on to some kind of dynamite.

The key here is scouting.  If the club have enough analytically to peer deeper, watch the player a few times, then that’s fabulous, and hopefully what’s happened.  Otherwise, well, your guess is as good as mine.

We do know that previous signings have had an analytical ‘calling card’, e.g. the key passing acquisitions of the summer, and to an extent Ream and Stearman doing their thing (what was it, blocks?  I can’t remember), so perhaps it’s as simple as that: they’ve found players who, in ways I can’t see, are statistical outliers, and, at lowish cost, have taken a punt.  Time will tell.

5 thoughts on “Fulham go moneyballing for gold

  1. The trouble with this approach it assumes that you have the football knowledge and intelligence to correctly interprete the data. Based upon the last two years; the Fulham Board have not demonstrated the ability to do this. Time for a radical change of the Board. Please wake up Mr Khan.

    1. Also, the Jaguars still suck, so I don’t quite understand their confidence that their approach will bring success to Fulham. I’m not saying I don’t trust analytics. But I certainly don’t trust Tony Khan’s buddies’ use of analytics.

  2. Oh boy. Kline’s not exactly a great salesman for advanced metrics, is he? He’s a little Bond villain-like in that article. You can see him rubbing his hands together in front of a monitor. At this point Kline isn’t new either–it seems he goes back to Summer of ’13–so fair to say his model isn’t really having the desired effect. I’m not saying there isn’t a player in the Austrian league worthy of playing for a team with dreams of being promoted but it’s not the first place I’d look! I’m far more sympathetic than most to the integration of advanced metrics into scouting but I’m left scratching my head here.

    Incidentally Rich, I thought of you recently when advanced copies came into the office for this. Interesting read:
    http://www.bloomsbury.com/us/soccermatics-9781472924124/

  3. Oh my goodness! Incredible article dude! Thanks, However
    I am going through troubles with your RSS. I don’t know why I am unable to subscribe to it.

    Is there anybody getting the same RSS problems? Anyone who
    knows the solution can you kindly respond?

    Thanx!!

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s