Brief Notes on Forward Evaluation and Metrics

Football analysis is in a crazy time right now in that so many things are brand spanking new to the sport that things are changing week by week. I’ve only been working on player stats for about a month and have already changed my approach a couple of times after working through various problems.  At some point in the future, I’ll probably write and re-write a best practices piece on what I’ve discovered doing player analysis, but today I just wanted to cover two really big things for anyone else that is interested in doing this.

Use Non-Penalty Goals
This is the big one for forwards because it levels the playing field and focuses on what really matters.

Penalties are converted at a standard 75-80% clip. Yes, there are some guys like Mario Balotelli or Matt LeTissier who are absurd at taking them and add to the rate, but the point here is that this is a very basic skill that many players have. It is also a different skill to goal scoring from open play, just as free kick taking is a special skill in and of itself. Each club likely has four guys on the pitch at any one time who can convert penalties at the standard rate, so why would you give any extra benefit to the forward who does the converting? If a defender is your penalty taker, and he scores 10 goals a season from the spot, but zero from anywhere else, does that make him good at scoring goals?

Put this another way. AC Milan win an absurd amount of penalties in Serie A every season. (Historically, their rate is a bit crazy. Berlusconi couldn’t inflate the Euro, but he’s done numbers for Milan’s ref bias.) Mario Balotelli had 12 goals in 13 matches in Milan, an incredible rate.

But!

Six of those goals were penalties. In numbers we actually care about, Balotelli had six goals in 13 matches after his move to Milan. This number, though impressive, isn’t quite as staggering as 12.

Assume for a moment Rooney and Robin van Persie convert spot kicks at the same rate.  Van Persie is the penalty taker at United and converts 8 penalties to Rooney’s 0 over the course of the season. Meanwhile Rooney scores 20 open play or free kick goals and RVP scores 12. At the end of the year, they both have 20 goals. However, Rooney’s 20 would have been far harder to achieve and more valuable than RVP’s penalty-aided 20.

I understand some people care about penalty takers in terms of having a leadership or mental strength characteristic, and that’s fine. You can just tick a box on the player evaluation form that says “Takes Penalties” and be done with it. In terms of statistical analysis, you really want to filter out the penalty goals from the much harder earned goals that come via the rest of the game.

Benjamin Pugsley is the first one I saw start to slice data in this way for forwards, and deserves a lot of credit for making me aware of its importance.

Per 90
This one is fairly intuitive once you see it, but not everyone notices it at first. All “rate” stats like tackles, passes taken, completed, goals, assists, key passes, interception – basically anything that is just counted – should not be done on a Per Game basis, but instead should be done at Per 90.

The reason for this is simple: 8 out of the 11 guys in every match play 90 minutes. Thus 90 is the standard unit for football activity. That said, some players get subbed regularly, and some (especially younger players) often appear mostly as subs. How do you normalize their contributions? By dividing those contributions into 90 minute segments.

This isn’t groundbreaking – nearly every sport that is being analysed in this day and age has some sort of time normalization to help compare players. Per 90 is the correct way to do it for football/soccer.

Advertisements

2 responses to “Brief Notes on Forward Evaluation and Metrics

  1. Ted,
    Agree with your conclusions. Would be interesting to know more about how your evaluation model has changed.

  2. Ted, I agree with both notions. Not taking them BOTH into account will bias all analysis. I would add two more
    – Take the league into account (only relevant for cross-league comparisons)
    – Take the strength of the teammate and opponents into account

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