by Ben Pugsley
Here at StatsBomb we are constantly looking at new and creative ways by which to evaluate teams and players. This innovation often leads us to writing about new ‘metrics’ that may not be all that easy to grasp on first read. Ted and I are often using short hand terms like Shelling, game states, and per90 and expecting the reader to pick up on what these metrics mean.
This may be a little unfair for, not many of these new metrics have clearly defined explanations behind them. So what Ted and I have decided to do is write about some of these weird new metrics, explain the methods behind them and why they are important.
Shelling (link) has already been written about. Explaining Game States is next on my list, but right now I want to walk the reader through Per 90 Minutes and why it should be used in every evaluation of a player.
Most football data sites are excellent sources of information. We can easily discover a player’s shot count, their goal tally and the pass%. These are useful numbers to have to gain a basic understanding of how a player performed in any given game or season.
But there’s a problem with these raw, basic numbers and I am going to use Arjen Robben and Frank Ribery to illustrate why I created (stolen from hockey) the per 90 minutes metric last year. *It has since transpired that Daniel Altman (link) had been ran per90 goals in 2005!!
2012/13 Bundesliga Stats
These two wide players are both outstanding footballers, but using the data above, it looked like one player was far superior to the other. Frank Ribery thrashes his Bayern team-mate in the goals and assists column. But as you may have guessed, the problem with the chart above is that it does not factor in the individual’s minutes played.
Minutes Played in 2012/13
Now, once we have the added context of minutes played the first table, which features the raw numbers, just cannot be taken at face value. The minutes played table dramatically changes what we thought we knew about the gap in performance between Ribery and Robben.
Ribery played more than double the amount of minutes that Robben did and hence Ribery’s goals and assists totals were far higher.
So what we need to do is marry the raw numbers (table 1) with the minutes played (table 2) to give us a greater understanding of how each player in terms of goals and assist given the number of minutes each player was actually on the field for.
One simple calculation later……
|Goals p90||Assists p90||Shots p90||SoT p90|
Per 90 minutes is born!
Once we have factored in the minutes played Robben and Ribery are an awful lot closer in terms of goals and assists than table 1 would ever have us believe. In fact, Robben is a far better shots and shots on target player than his team-mate Ribery.
This is why I use per90 and it’s why we should all use per90. It gives us context when evaluating players who play wildly different minutes over the course of a season.
I have sorted this table by goals per90.
|Goals p90||Assists p90||Shots p90||SoT p90||Scoring%||Passes p90||ToP%|
For a sortable version click here (link)
One of the reasons I included midfielders in the table seen above is to look at which players are ‘hub players’, which means high volume passing players. The type of player that springs to mind when I think of a hub player is Xavi or Carrick, and maybe Luis Suarez.
Let’s find out which players were high volume passes per 90 players.
Obviously there is a Barca team effect at play here, but Xavi would post a high passes per90 in any team he played for. Carrick is high on the list, as are Toure and Andrea Pirlo.
I have walked the reader through the steps of why a player’s raw numbers can be misleading and the need to use a players minutes played in order to give us added context. It really is vital to convert a player’s numbers into per 90 format if we want to understand how well a player performed given his minutes played.
Per 90 may take a little getting used to, and the numbers in the per 90 chart may look a little odd at first glance, but once one looks at player stats in per 90 format, it’s awfully difficult to go back to using a player’s raw stats for evaluation and comparison purposes.