Analytics, Marginal Goals and Matej Vydra

The Numbers Game, the book written by two American social economists (me neither) that was published earlier this year, aimed to do for football analytics what Moneyball did for Baseball. Though the book’s ambitions to highlight what we can learn from statistics and data in the stubbornly traditional sport are praiseworthy, it often lapses into mindless evangelisation of wishy-washy statistical hypotheses that detract from the heart and soul of the game, the thing that makes it the subject of such universal passion.

One section that did get me thinking, though, was one entitled ‘They Should Have Bought Darren Bent’. In it, the authors pry into the relative value of goals; whether we can put a greater value on a striker’s return that simply their ‘goals scored’.

Their reasoning was thus: all goals are not equal, the fifth in a rout is less valuable than the late winner in a 2-1 victory. Using results from the top four European leagues (Premier League, Serie A, La Liga and Bundesliga) for the last decade they plotted an average of how many points were won by a team against how many goals that team scored – though there are, of course, no guarantees in football, the conclusion was that scoring five or more goals was pretty much enough to ensure victory, while one goal tended, on average, to be enough for a point.

The team used this data to create an exchange rate of sorts for each goal. The most valuable goal, they deduced, was the second, as it increased the team’s predicted point haul by 0.99.

What’s this got to do with Watford? Well, when Matej Vydra was going through his hot-spell in January, I suggested that it seemed that his goal tally was perhaps painting a rather complimentary picture and that the bulk of his goals were padding to the big victories that the side were churning out with ostensible ease at the time.

With talk that the Czech striker might return, I thought I’d return to this theory, and look at how valuable his goals really were. Using the figures from The Numbers Game, I was able to imitate their calculation for “marginal points” won by any one player over the course of the season.

First, it was a case of recording who scored what goal in league action for Watford in the 2012/13 season:

1st Goal

2nd Goal

3rd Goal

4th Goal

5th Goal

6th Goal

Total

Vydra

9

7

4

2

0

0

22

Deeney

10

5

4

0

0

1

20

Abdi

5

5

1

1

0

0

12

Forestieri

3

4

1

0

0

0

8

Chalobah

2

3

0

0

0

0

5

Yeates

1

1

1

1

0

0

4

Anya

2

1

0

0

0

0

3

Geijo

0

1

1

0

0

0

2

Hoban

2

0

0

0

0

0

2

Battocchio

1

0

0

1

0

0

2

Taylor

1

0

0

0

0

0

1

Doyley

1

0

0

0

0

0

1

Hall

0

1

0

0

0

0

1

Murray

0

0

0

0

1

0

1

Pudil

1

0

0

0

0

0

1

Ekstrand

0

1

0

0

0

0

1

Briggs

0

0

0

1

0

0

1

After assigning the respective point values of each goal, we were left with a total number of theoretical points that each player had earned:

Vydra – 17.04

Deeney – 15.50

Abdi – 9.87

Forestieri – 7.00

So it appears I was wrong. Vydra nabbed seven of those all-important second goals and gained Watford a hypothetical tally of 17.04 points. All this is theoretical, of course. Lloyd’s attacking exploits over the course of the campaign are said to have a ‘marginal points’ tally of 0.83, whereas his lone strike at Bolton, though spectacular, meant absolutely bupkis in the end.

Had Vydra had such a season in the Premier League in 09/10, that total would have seen him fifth on the table behind Wayne Rooney (20.64), Darren Bent (20.02), Didier Drogba (19.59) and Carlos Tevez (17.67). Whereas if he was transposed to the following year, he’d be a clear winner ahead of Tevez, who only managed 15.70.

Troy’s exploits would have seen him just below his strike partner in 09/10, above Fernando Torres’ 14.34 (that being his last year at Liverpool, when he was still good), while Forestieri’s goalscoring prowess puts him in Dirk Kuyt territory.

For the sake of argument, let’s look at some of the Championship’s top scorers last year.

Glenn Murray – Crystal Palace – 24.16

Jordan Rhodes – Blackburn – 21.60

Charlie Austin – Burnley – 20.42

Chris Wood – Millwall/Leicester – 16.56

Thomas Ince – Blackpool – 13.97

David Nugent – Leciester – 12.59

The top three were all their clubs main penalty takers, but their importance to their teams cannot be understated. As far as goalscoring goes, Palace and Blackburn were something of one man teams moreso than any of the Premier League teams over the two seasons that ‘marginal points’ were recorded in The Numbers Game.

One of the conclusions that Sally and Anderson arrived at after this analysis is that Chelsea would have been far better served spending their £50 million on Darren Bent rather than the mopey Spaniard they ended up with – their decision-makers should have looked at the importance of the goals rather than the total number, they say.

It is easier, as Rhodes, Murray and Austin show, to score the important goals when you’re the only person in your team scoring, but all three had undoubtedly great seasons. If we are to follow the authors’ theory, then Vydra might be waiting for this prospective big move for some time – Jordan Rhodes and Charlie Austin, should, so the numbers say, be higher up any big teams’ priorities.

But it’s not true, is it? A fine goalscorer Darren Bent may be, but there’s a reason that he’s never featured for a big club, and never had much of a go in international football. The cream rises to the top, whatever numbers say. Analytics and statistics are a fantastic resource, and enable us to learn more about the game than we could know when football was only watched by bare eyes and not machines.

But using them to completely subvert thinking just doesn’t work.

Football is a complex sport. Baseball has had a rebirth since Bill James’ sabermetrics were popularised by Moneyball, but that’s because all of the innovative statistical formulae and logarithms are being applied to one guy chucking a ball at a bloke with a bat again and again and again. Football is a game of feel, of complexities; off-the-ball movement and indefinable athleticism is just as important in creating a goal as having someone boot the ball in the net.

Matej Vydra is better than Charlie Austin.

What have we learned? Nothing. But I’ll be damned if I did all that maths to not get an article out of it.

I’d like Vydra back. It turns out he doesn’t just score icing on the cake, he does a fair bit of the baking as well.

Advertisements

One comment

  1. Very interesting article, would be helpful to measure contribution to goals scored (and to wins when playing) to supplement this. unfortunately it becomes a complex equation if you try to do this.
    For arguments sake you can prescribe values to; being in the final third (and hence occupying a defender), to assists or secondary assists (as in Ice hockey), to the number of minutes played previous to the goal (particularly in conjunction with mistakes made by the defender during that time i.e. fouls, misplaced passes, slips etc), and so on. Unfortunately as you say it is all (to quote Dr Ben Goldacre) ‘a bit more complicated than that’.
    What we can instinctively see without statistical analysis was that Vydra was very important to us last season (Brighton away showed how he was perfect in the counterattacking system).
    p.s. try reading ‘Soccernomics’ (previously ‘Why England lose’) for another attempt at statistical analysis of Football.

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

%d bloggers like this: