One of the biggest purposes of numbers in sport is to separate players by their performance. While it’s really easy to tell the difference between the truly elite and the incredibly awful, things get a little blurry as you get more into a “middle ground.” Bill James, the famous baseball thinker and “father of sabermetrics,” once mentioned that the difference between hitting .300 (good for the sport) and .275 (average for the sport) is merely one hit every two weeks. While it’s quite easy to see the difference between a guy hitting .333 and .200, finding the difference between good and average or great and good becomes much harder is very hard to find simply by watch. That’s where statistics really can be helpful. They make that difference much clearer between two players. They “see” that hit every two weeks.
From there, it’s a matter of finding which statistics really mean something – and for which time periods they’re applicable. For me, greatness isn’t just based on a single season. Greatness, to me, is defined by consistent performance at a level that distinguishes a person. So, when you look for stats to define greatness, it’s necessary to find stats that are applicable moving forward. While having a great season is awesome, it doesn’t necessarily mean that performance can or will be replicated in the future.
Take, for example, finishing. While someone who finishing 20 out of 80 shots is really impressive, it doesn’t necessarily mean that they will consistently finish at – or around – 25% for their career. They might, but it doesn’t necessarily mean they will. It’s really easy to find out whether a stat is indicative of “true level” performance or if it’s just an anomaly in a single season – all you have to do is compare the stats between seasons for a player.
For my test, I looked for players that took more than 40 shots in Spain in both 2013/14 and 2014/15. By keeping the league the same, you remove another variable in the style of the league and how those changes can cause differing results. Given those parameters, there were 21 players to look at. Next, you run a comparison between the numbers between the years. For now, I’ll stick with finishing.
When I ran the numbers, this is the graph that was produced:
The r^2 value is pretty low, and the trend line suggests there’s no correlation. When you’re looking at r^2 values, the closer to 1.0 the better the correlation – so a r^2 of 0.01576 is really weak.
What does that mean for conversion rates themselves? Well, they show what happened during the season. They definitely help differentiate between what closer calls (for instance, a player scoring 15 goals on 100 shots probably didn’t shoot as well as someone scoring 10 goals on 50 shots), but they don’t really show “true talent.” Seeing Neymar finish at a 23% clip last season doesn’t really indicate that he’ll do that again in 2015/16 or moving forward. That doesn’t mean he’s inconsistent or not great, just that the trait of finishing itself is very unstable – for all players.
The challenge then becomes to find a stat that indicates true talent to differentiate players. If we know that goals per shot isn’t a good option, and goals themselves are dependent upon shots taken, it seems the next best route is to look at shooting statistics. I took the same group as before, but this time I looked at shots per 90 minutes (to put all the players on an even playing field. Again, here’s the graph that produced:
That looks much better. It seems that shooting is incredibly repeatable, so someone who shows some improvement in shots per 90 really may have improved overall – and that may be something to look for.
My biggest concern there is just that shooting, in and of itself, isn’t necessarily indicative of greatness. To give you an idea, in 2013/14 Mario Balotelli took more shots per 90 minutes (5.96) than Lionel Messi (5.76) – and I haven’t heard anyone claim that the former is the better player. So there has to be a way to see whether or not a forward is truly performing well with a stat that brings consistency.
Again, we know that goals per shot isn’t good, but what happens a step before that? What has to happen before the shot can go into the goal? It has to hit the target. There are a lot of variables that can happen between the shot being on target and it finding the back of the net (for instance, a great or poor reaction from a keeper), but the shot on target itself indicates it was a good shot. So what does the graph look like for shots on target per 90? Glad you asked.
Again, this has a very high correlation. It isn’t quite as high as shots per 90, but this seems to mix consistent, repeatable performance with something that’s very much going to be a positive event for the team. While shots on target per 90 isn’t perfect, and it’s not all-encompassing (there’s much more to being a forward than shooting, let alone any player), but it’s a good step. When looking for performance moving forward, this seems to be a good indicator of a player who is on the verge of scoring big and one that will bring positive attributes into the team.