Thursday, June 13, 2013

Passing %, Passes Per Turnover, and the Presentation of Data

Passing % is a fairly ubiquitous statistic.  You see it flashed across the screen every halftime, right next to Possession % (which is really just total pass share (%) but that is another discussion).  The problem with Passing % is that the public has no idea how to interpret it:  What is a good passing percentage?  80%, 75%,  85%?

When we hear that Barcelona passes 89.5% as a team we have a sense that this is a good thing, but we also have no context: you would want to account for the difficulty of the pass (length, forwards, backwards, defensive pressure, attacking third, own half, etc.) to get a true understanding of a team's passing success.  Presenting such an advanced metric to the general public - assuming one even exists - would be problematic.  If most people don't know how to interpret something as simple as Successful Passes divided by Total Passes then how can we expect them to understand a complex algorithm? There must be a simpler solution.

Below are six stylistically different teams (Barcelona, Liverpool, Portland Timbers, QPR, FC Dallas, Stoke) ranked according to Passing % (out of 117 teams from Big 5 UEFA leagues + MLS).

There are some big differences, but it is likely hard for the average fan to comprehend the vastness of the disparity.  That is because the chart looks something like this in your head:

But what if we took the same metric (passing %) and presented it in a different manner. What if instead of looking at Successful Passes / Total Passes we made it a ratio instead:  Successful Passes to Unsuccessful Passes.  And let's rename it something more digestible and intuitive: Passes Per Turnover.

As more soccer statistics become available it is imperative that the community of soccer analysts (some professional, many amateur) continue to think of new ways to present the data that the average fan can understand and appreciate.


  1. Teaching statistics, I have found that while often students cannot intuitively distinguish between a proportion (pass%) and a (odds) ratio (pass/TO), one of the two is usually better understood.

    Also, like you cleverly showed, the ratio here spreads the data out more so the differences are easier to see.

    I guess I would say that the only advantage to the pass% is that it is a proportion, and thus the margins of error are easier to calculate if we're willing to accept the binomial probability assumptions...

    1. Thanks for the comment. I do still like the passing % as a metric but find that, for whatever reason - perhaps they can imagine the actual passing sequence, passes per turnover seems to resonate more with the average soccer watcher.

  2. Quick question here: You have determined the Passes per Turnover Ratio from looking at Passes Successfull and Passes Unsucessful. What about other ways of losing possession i.e. Unsuccessful Dribble? Those wouldn't be included in your Turnovers and would skew the ratios.



    1. Absolutely. Don't get me wrong I would love to see that data included too as turnovers, but calculating that figure is much more difficult. I am able to do it for MLS, but in other leagues the requisite data is nearly impossible to come by.