Wednesday, February 8, 2012

Percentage and Rate:How Advanced Statistics Fail with Basic Math

In both advanced and basic game statistics, you need to know how much or how often something occurs -- player makes a free throw X amount of times out of Y opportunities, or grabs A rebounds out of a total of B. Typically in the mainstream media that's represented by the percentage, which is well understood by virtually everyone. Shooting percentages are commonly expressed in that form where everything is based on 100. There's also the simple ratio or rate -- X out of Y, or 0.90 free throws per free throw attempted instead of 90%. However, even with so-called advanced metrics the two are often confused with one another, and the consequently mistakes can happen from the mislabeling.

First of all, I realize this is nit-picking, but as someone who parses through data both for basketball statistics and a profession, consistency is extremely important. If something is called percentage, it better actually be percentage or an error could be made. For example, is an incredible resource, but when you use the season finder most of the statistic percentage filters actually require a ratio form. If I'm trying to find list of players who have averaged 30 points a game with a field-goal percentage over 50, then I will receive an empty list unless I used the incorrect "0.50," which is not a percentage. With more filters and more complicated searching processes, it can be harder to find the error. Consistency in the usage of percentage would save time.

Both ratio and percentage are easy definitions to remember. Ratio or rate is just blank divided by blank -- miles per hour, people per square mile, heart beats per minute, etc. Percentage is more specific. It's a ratio scaled by 100. If there are 80 people per 100 seats in an arena for a rate of 0.8 people per seat, then the arena is at 80% capacity. All you do is move the decimal.

Again, I realize this is pedantic, but the reason for the specific definitions is so we can tell them apart from something similar. On ESPN's advanced stats page, all the rebound stats, the usage, the assist, and the turnover statistics are called rates yet they're out of 100 while true-shooting percentage is a rate. They got the definitions completely reversed, and this is the most prominent sports site.

All I want is a little consistency in basic math terms so I know what I'm getting.

1 comment:

  1. If you're going to be pedantic, you should at least be correct. "0.50" is not an incorrect way to represent making 5 of 10 field goals. In fact "50" would be incorrect unless you append it with the all-important percentage sign. So when you find yourself typing "0.50" into B-R, the only way typing a whole number would make sense is if it was "50%".