FanPost

Unicorn Hunting - WAR/600


Before entering esoteric stat space, i decided to just look at hitter WAR again, this time on a rate basis. I simply divided WAR by plate attempt (PA) for every season/player, and multiplied by 600 to give a WAR per 600 plate attempt value.

This is important to consider, since WAR is a 'counting stat'. the more you play, the more WAR you can accumulate. Anything that takes playing time away from a player will reduce the opportunity for that player to accumulate WAR. this means, in theory, that it WAR doesn't just measure player quality, but to some extent, that player's quantity. Removing 'quanity' from the metric by normalizing to plate attempt should give a slightly more pure metric of 'quality' - at least as reported by WAR.

36FG9aM.0.jpg

This plot is again a histogram, this time of WAR per 600 PA. you can see over these five seasons (i have eliminated 2020 from the data, this now includes 2018, 2019, 2021, 2022, and 2023), is that there is a wide range of 'quality' as measured by this statistic, anywhere from ~ -5 to 10 WAR per 600 PA.

vN2FZqu.0.jpg

We see many of the same unicorns as we did with the WAR, as is expected. however, we can also see a new name on the list - Mike Zunino's 2021 season. The 2021 season WAR total is actually not that out of line with a few prior seasons with seattle, which included a 3.8 and 4.6 WAR season in 2014 and 2017. However, the way he attained the accumulated WAR was somewhat different. In 2014, WAR was primarily derived through defense. In 2017, it was still driven largely by defense. In 2021, it is almost an almost identical balance of offense and defense. this is driven by an an anomolous 0.342 isolated power. this is his age 30 season, not out of the realm of a late peak in power, but maybe not expected, given a career ISO of just north of .0.200.

Wanted to also compare the distrubutions of WAR vs its rate counterpart:

Tz7bjK2.0.jpg

The top WAR/600 plot is much more normally distributed. The left side and right side of the plot have similar shape. The bottom plot has what is called 'right skew', where there is tail to he right which shifts the bulk of the values to the left. By removing the playing time factor of WAR from the metric, we in theory get a slightly purer description of player skill. However, we also run the risk, in this case, of observing increased noise. you can't accumulate a lot of WAR without a lot of playing time, and the more PA we observe, the more stable the metric will become. The flip side, is that you can get players with fewer plate attempts who fall into a hot streak over a 300-400 PA period and become unicorns. This is still unlikely to happen without talent - over his career he has accumulated about 3.5 WAR/600 PA. I am guessing this is a good deal of injury time. i have to figure out whether that is in the fangraphs data i am use. TBD.

And for fun, the reverse unicorns:

sSA1B6j.0.jpg