Laidig Speaks: All He Wants for Christmas

santaThis time of year is widely considered the season of giving.  However, as a parent to three young girls, I also know that this is also the season of wish lists.  And it is in this spirit that I offer: what I want in soccer statistics. I made this list because I hope these metrics will be developed and made available to soccer fans (and stat geeks). If my wishes are granted, broadcast analysis would be more informative and interesting, and the understanding of the game would grow.  And as understanding of the game grows, sports fans become casual soccer fans, and casual soccer fans become season ticket holders.  In addition, I think this information would be useful for the teams that analyze performance, which would lead to better performance across the league.

First, I’d ask Santa (or Don Garber, or Opta, or Stats Inc…you get the idea) for information tracking 2nd Tier passing.  In other words, I’d like to know if the recipient of a pass was then able to complete a pass or attempt a shot (defined as an assist or “key pass” by Opta). Devin Pleuler, author of the Central Winger column for, has published some of this data.  But one person compiling the data by hand is not the widespread availability I am wishing for here.  The bottom line is, good players complete passes.  Great players make passes that allow their teammate to do something with the ball.  Let’s keep track of that.

Second, as long as we talking about passing information, let’s separate the square/back passes from the more aggressive forward passes.  The forward passes are more difficult to complete, and are similarly more likely to unlock a defense.  Although conservative passing may be wise in a specific situation, treating all passes equally loses important information, and just doesn’t fit our understanding of the game.

Third, let’s ask for a defensive metric. I’d like to know if a player was able to pressure opponents, shut down the attack etc.  We have stats for dispossession, such as tackles, interceptions, and recoveries.  But a defender may harass an opponent all night, reducing the opponent’s production in a way that’s less visible. Thus, let’s report an opponent’s passing percentage allowed by each defender/player (preferably separated into the types of passes as above).  Sure, this requires some judgment in determining the closest defender, and whether the player is close enough to get the credit (or blame) for a pass.  But it’s the season to think stats

Finally, and this is the biggest ask of all.  I want a metric to quantify off-the-ball movement.  I know there is some data on the number and speed of sprints.  But we should also add location information; where the run is on the pitch, how far, relation to the ball, etc.  We know that off-the-ball movement helps the offense by drawing defenders, wearing down opponents, screening defenders and the like.  However, if we don’t measure it, we have a tough time recognizing these efforts.  And I think the collective wisdom of the soccer universe can determine which actions are the most meaningful, and then follow that up with accurate measurement.

I’ll be the first one up on Christmas morning to see what I get.

6 Responses

  1. I wonder how accurate or effective adopting hockey’s plus/minus stat would be. (Player gets a point when his team scores when he’s on the field, but loses one when other team scores when he’s on.) It might be a way to show all-around effectiveness, especially for midfielders.

  2. When on a broadcast would you actually communicate all this data?

  3. Plus-minus data is attractive to some coaching staffs, but supported less by stat people. Some issues involve whether a player should really get the credit or blame for a score. Does a Forward get knocked when an opponent scores three goals? It’s not so clear. And considering pragmatism in developing a statistic, there is less variability for plus-minus in soccer. In hockey, a shift change takes place every couple minutes. In basketball, players sub out for stretches at a time. There is more player movement to measure to differences between players. In soccer, some players regularly take the field for 90+, and others may sub in for 10 or 15 minutes every other week. What difference a player makes to results is a natural and intuitive question, but the plus-minus metric is rather blunt for an all-around useful tool.

    However, all those issues aside, I have combined plus-minus into a player “dashboard” of metrics in specific situations. In cases where little information is available (i.e., some non-MLS, Non-Euro leagues), I used plus-minus for CB pairs (under the assumption this tandem plays a large role in the defense, and in turn helps set up the offense), comparing each pair to an expected value based on the opponent and location (Home v. Away). I used the expected value to deal with the small sample issue in that some CB pairs may have played in 4 games, and of a different quality of opponent. Evaluating this plus-minus with other basic metrics, like goals allowed, shots on goal allowed, can tease out the style of play for each tandem and their results (versus expectations). I’m working on an MLS example of this to allow some critical evaluation (and perhaps improvements), but am not quite there yet.

    And as for broadcasts, I think a natural place to insert some of this data is in pre/post-game and halftime analysis segments. And I also think that a properly supported color commentator could work it in as well. For example, “Donovan has not scored tonight, but he has been a key offensive facilitator by completing 85% of his forward passes, with a season-high 60% secondary pass rate.”

    I don’t expect an announcer to read a box score with all this data; that would be quite dull, even for me. But baseball announcers don’t read out of the Baseball Prospectus line by line either. A statistic is a fundamentally a quantitative summary of some observed behavior. By reporting a statistic, the person is implicitly stating that the stat accurately describes what is going on. And I would expect soccer announcers/analysts to select stats that they feel accurately represent the story of the game. Right now, all they have is their narrative for the game. Some more statistical definitions and data, used to support the narrative, would only add to the experience and place moments in context. We know Wondo had one of the best years in MLS history because we keep track of goals scored. Better data would allow us to capture the contributions of other players, and their own unique roles. (And I imagine some marketing guru could make use of this as well)

  4. I looked at traditional +/- for Man City in 2010/11 and it didn’t add anything as there was no real way of telling who was involved in goals and the ‘usual suspects’ who played all the minutes were just bunched at the top.

    This season though I’ve started looking at recording (manually) scoring chances like a couple of hockey guys have done; recording plus marks and minus marks for those involved in scoring chances and those at fault when conceding one. Early days but some promising stuff.

    There are plenty of other areas I think are worth looking at but the difficulty is (here in the UK at least) not having access to game tape as not very game is shown or covered live which restricts being able to manually record data.

  5. I think Danny’s summary explains the problems with the plus-minus. It doesn’t mean much for most field players. There is, arguably, some meaning for central defenders and perhaps mids. But again, I’d only use it in low-information scenarios. I prefer to use my adjusted Castrol scores for the leagues that put out such information. And even this data has issues that hopefully will be addressed with Stats 2.0 (or by Santa).

    And I like the grading of each player on a play, which is the most informative and most labor intensive analysis. In football (NFL, not footie), lineman may be graded on each play with a 2 (did job), 1 (partially did job), or 0 (failed). And over the course of the game their scores would be tallied. Applied to soccer, this is a little more difficult due to the dynamic nature of the game, but worth a try I think. here, there doesn’t seem to be the interest (or resources) to implement such a system though. Someone with knowledge of that day’s tactics and game plan is required to review…a coach. And coaches are busy enough as it is.

  6. Youth soccer clubs need to promote MLS on TV every week. Encouraging the kids and their families to watch games on TV would go a long way toward ratings. I see most clubs ignoring the TV MLS schedules.

Comments are closed.