Hai-Binh Ly over at The Volleyball Analyst blog posted an article on the subject of what he calls the Point Efficiency Ratio (PER). He does a lot of stats work and periodically writes about it. I previously commented on some of his other posts with respect to blocking and serve reception. Generally, his articles give you something to think about, even if you might not agree.

So what is the PER?

It’s defined as “Points Scored (Kills, Block Kills, Aces) divided by Points Lost (all Errors)”. I don’t know whether blocks against count among the Points Lost. It perhaps depends whether you refer to those as errors or not.

So basically PER is the ratio of the points you score over the points you give away. Opponent earned points do not factor. I want to address the usefulness of this metric in a couple of ways.

###### The statistics

Ly dedicates a fair bit of the article to defending his claim that PER is “the best stat in volleyball”. It’s the basis for his Masters dissertation, so it makes sense that he rates it highly. 🙂

His reason to say PER is best comes down to its very high statistical significance per his analysis of various different performance measures. It’s worth noting that high significance doesn’t itself mean much. You can have an extremely high significance, but have a variable that doesn’t really impact outcomes very much. I’m not saying that’s the case here. It’s just something to keep in mind when looking at any kind of statistical analysis.

The funny thing is even Ly’s own research shows that PER comes in second behind Point Efficiency % in terms of the top metric. This efficiency is probably what you think it is.

PE% = (Points Scored – Points Given) / Total Attempts

In this case I have to assume the Attempts part is total rallies, which includes points earned/given by the opposition. This immediately strikes me as likely being a stronger metric even without looking at the statistics as it at least captures in the Total Attempts the points the opponent scores and gives away. After all, our team doesn’t play against itself.

###### PER’s value

Now let’s talk about whether there’s really much value to the PER from a coaching perspective.

I thinking it’s probably blindingly obvious that the bigger the gap between the number of points we earn and the number we give away correlates well to how much we win. This is the case any time you focus on points. Naturally, winning teams score more than they give up. That’s a fundamental aspect of our game, as Ly notes.

So does something like PER (or PE%) give us anything to work with? If we have some kind of benchmark, then yes. By benchmark I mean we know the PER of our desired level of performance. For example, if we want to finish in the top three of our league, what’s the normal PER for team at that level?

If we don’t have that benchmark, then we’re left with just a self metric that lets us gauge improvement over time. Of course that’s useful as well since it does generally correlate to more wins.

###### PER’s limits

What I outlined above is pretty much where PER’s (or PE%’s) usefulness runs out. It’s fine to say we want to improve PER, but where do we need to focus our efforts? Are we not scoring enough, or do we need to cut down on our errors (or in the case of PE%, are we not stopping them scoring enough)? That’s where you need to have more granular metrics.

And let’s assume you decide you aren’t scoring enough. What does that mean? You need to drill down even further to decide whether you have to address attacking, blocking, or serving to get the PER improvement you want. Of course, from there you then need to go one layer deeper to figure out within the scope of that particular area what needs work or where you can get the most impact.

You see why I say setting priorities are one of the two primary jobs of the coach and why if it’s easy you’re probably doing wrong? 🙂

### 6 Steps to Better Practices - Free Guide

Subscribe to my weekly newsletter today and get this free guide to making your practices the best, along with loads more coaching tips and information.