Sunday, February 23, 2020
Introducing NaiveWAR
At around midnight, I had a tweetstorm on NaiveWAR. What follows is simply all my posts collected into one thread. If there’s anything new beyond the tweets, I’ll make a point of highlighting it by using the Pozterisk.
I just created NaiveWAR. It’s the absolute simplest version of WAR that I could create. I’ll see if I can do this by tweet, though I should blog it instead. Anyway, I’ll use the 2008 Orioles (68-93) for my step by step. And by the way, this method can be applied for ANY SPORT.
Let’s start with pitchers, as that’s easier. We are going to create a metric called “Runs Above Team”. The 2008 Orioles allowed 869 runs on 1422 IP, or 0.611 runs per IP. If Guthrie allowed that many runs on his 190.2 IP, he’d have allowed 116.5 runs. Instead, he allowed 82 which is 34.5 fewer runs than his team. We divide runs by 10 to get it into the wins scale, and so Guthrie is +3.5 Wins Above Team (WAT). We do this for all the pitchers. +2.4 for Johnson, down to -1.9 for Garrett Olson. The total sum is, naturally 0.
Next, we create “games”. The Orioles played 161 games, of which we assign 3/7ths(*) to the pitchers, or 69 games. With 1422 IP, that means that we hand out 1 game for each 20.6 IP. So, Guthrie’s 190.667 IP gets him 9.3 games. Since the Orioles won 68/161, or .422 wins per game, we set the baseline for Guthrie at .422 x 9.3 = 3.9 wins. And since Guthrie is +3.5 Wins Above Team, then he gets 7.4 wins. We do this for all pitchers, and we get 29 Wins.
(*) This number is not too important. You can use 1/3rd if you like, and you’ll see that it’ll still work.
Since we have Games and we have Wins, Losses is just the difference. And if we set the replacement level at .300, then our WAR is wins minus 0.3 times Games. This is how it looks like for the 12 Orioles pitchers with the most innings.
I’ll do the nonpitchers tomorrow. They get 39 wins. And 29+39 = 68, what the Orioles got.
Let’s do the non-pitchers, though I may need to pick this up tomorrow. Our Naive metric will be Runs Participated In, which is R+RBI-HR. Markakis has 173 RPI (with 414 batting outs). The Orioles had 1360 RPI on 4121 outs or 0.330 RPI/out. So the average Orioles player would have .330x414=137 RPI given Markakis’ outs. So Markakis is +36 Runs Above Team, or +3.6 Wins Above Team (WAT). We give 4/7ths of the 161 games to nonpitchers or 92 games. With 6211 PA, that means we give one game for each 67.5 PA. Markakis gets 10.3 games. The average Orioles get .422 wins per game, so Markakis starts with 4.4 wins. We had the 3.6 WAT so he’s at 8 Wins. And therefore 2.3 Losses. And 8 W is 4.9 wins above the .300x10.3 replacement level. That’s 4.9 WAR.
This is the 12 Orioles players with the most PA:
And that’s how WAR started. You realize the holes immediately.
- Hey, where is fielding?
- Hey, a SS generates more wins than a 1B given the same hitting production
- Hey, what about the leverage of a relief pitcher?
- Hey, isn’t it tougher to start than relieve?
- Hey, is it always 10 runs per wins?
- Hey, can’t we do better than RPI?
- Hey, can’t we do better than Runs/IP?
So, yes, that’s how you take it from this NaiveWAR to the WAR you see on @fangraphs and @baseball_ref. If you don’t like NaiveWAR (and you shouldn’t), but you also don’t like fWAR and rWAR, then that means they either went too far, or they took the wrong turn.
And that’s where YOU come in. Start with NaiveWAR, and now take whatever road you want. And let’s see where you get.