Why sports player ratings are (often) broken?
And why you shouldn’t care too much about it.
Player ratings are a big things in sports games. From NBA players to NFL superstarts (link), it’s a hot topic, always there to create debates between fans, analytics and players. Is Kevin Durant really deserving the same rating as Lebron? Why is my favorite player rated 98 and not 99? No way Michael Jordan at his peak is as good as 2012 Lebron. See, debates.
Astonishing Sports games rely on ratings for things like trade value, or Hall of Fame election. But when I can avoid displaying them, I do. Why? Because the moment you let people see ratings, they’re going to consider it an absolute, true representation of a player’s skills. And I don’t blame them, that’s why they’re here for. But one thing that sabermetrics has proven is that it’s probably impossible to have a truly accurate picture of a player’s skills.
Even WAR, considered by many in baseball to be the best solution to this problem, is debated (Even though I truly think this is an amazing stat). So if you have to provide a player rating, well, you do with what you have.
Disclaimer, it’s all maths
First, let me get something straight. If you want to create a simulation, or a game, or anything really, to use a player rating, you’ll need maths. After all, player rating is a number. Imagine you have a subjective idea of what a player rating should be. Let’s say, Lebron James is 99, KD is 98, and Westbrook a solid 95 (I already hear you screaming in disagreement). Okay so you input a number into a game that is supposed to output a result. Points, rebounds, blocks… I don’t know how you would do that accurately with just one number, but in any case, it’s all maths.
How to get a player rating?
If you want to create your own player rating system, you’re lucky, because there are already plenty of ways to measure skills in most sports. WAR in baseball, PER or EFF in the NBA, PS for hockey and so on. From there you can work to create a scale to have a rating on a 0–99 or 60–99 scale, either linear or exponential, or anything in between. What I like to do is taking reference points, and work from there.
For example, in ABK20, I use Effiency (https://en.wikipedia.org/wiki/Efficiency_(basketball)). Then I take MJ efficiency for one of his seasons, and set it as the 99 rating point. I generally add a few more reference points with lower rating, and then create an equation from them. And voila! See, that was easy?
Wait, why did you use Efficiency, and not (insert any statistics here)? Well, I feel like this is good as any other stats. You would certainly see differences with others, but I think it wouldn’t change that much. Just like when you compare bWAR and fWAR in baseball. (bWAR uses ERA as a base for pitchers, the other uses FIP).
But here lies the biggest problem of this technic: you have to pick (or create. Good luck.) a stat to represent a player value. Or skill. Or power, whatever you want to call it. And it doesn’t really matter what you’ll pick, you’ll always have people who think a system would be better than another. Who knows, maybe people at Fangraphs are looking at bWAR like These noobs still use ERA. Is it still 1956 or something?
And actually, even people who don’t know a thing about statistics might complain because you ranked a player higher than their favorite. It’s sports, and when you’re passionate, it’s hard to see the truth, especially when truth itself is impossible to distinguish clearly. I know that Mike Trout is better than Mike Zunino (sorry Boy Howdy). But between Trout and Yelich? That’s another story…
But games don’t use statistics!
Well, it’s true that most games are using words that describe skills rather than using a statistics equivalent. NBA 2K uses an American grade system for things like “Steals”, “Dunks”, “Interior Defense” etc. But I believe it’s still based on statistics, and then maybe manually adjusted. Again, if you don’t use an objective way of measuring things, you’re just guessing. And I’m convinced guessing would produce even more debates.
So basically player rating sucks?
Yes. And no. They have value because they can quickly give us an overall idea of a player value. But we should probably take it as a rough estimate rather than the absolute truth. And that’s where most people would be lost I guess. Because associating one number with a player makes it de facto the way to compare it to others. Between choosing a 98-rated player and a 99-rated one, someone would probably pick the second one, without regards for strategy or actual stats (me included).
That’s one of the reasons I don’t like player ratings. I prefer to encourage Astonishing Sports players to at least look at several characteristics to form a proper idea of a player’s value in their mind. Astonishing Baseball has no visible player rating. I assume people have a basic knowledge of baseball stats. Otherwise they still have four colored chips, that provide a rough idea of a player level in one particular skill. And it has always worked well enough. Maybe people don’t need player ratings after all?