I have spent this Spring Training rooting for JA Happ to earn the 5th spot in the rotation. It is not that I wanted Chan Ho Park to fail, but I have generally felt that Spring Training numbers are overvalued in making roster decisions. I have generally believed that the sample size is too small to base any decisions on, and so I thought Happ was the better option and wanted him to succeed and Park to fail. After all, Chan Ho Park faced 8,074 hitters in the major leagues, demonstrating that he is in fact mediocre. Right? Maybe not.
When we say things like "small sample size", we need to make sure that it's not just an empty sabermetric talking point. What a person means when they say "small sample size" is that the resulting outcome is within two standard errors of what would be expected on average. As the size of a standard error is inversely related to the sample size, the idea is that the range of possible outcomes is too vast to trust that any one outcome represents the truth.
So, as I thought about all of this, I took a look at Chan Ho Park's spring numbers. His ERA is quite low at 2.53, but so is JA Happ's at 3.15. But then I looked at Chan Ho Park's peripheral numbers and something shocked me.
Hitters faced: 85
Strikeouts: 25
Walks: 2
That seemed extreme. After all, that is 29.4% K/PA and 2.4% BB/PA. In his career, he has 19.7% K/PA and 10.6% BB/PA. I checked for statistical significance-- in other words, I checked to see what the odds were of someone with a true skill level of 19.7% K/PA and 10.6% BB/PA getting 29.4% K/PA or more and getting 2.4% BB/PA or less by random chance. As it turns out, the odds of being 9.7% above his K/PA by luck when his skill level remained the same is 1.2%. The odds of being 8.2% below his BB/PA by luck is 0.7%.
These are both statistically significant-- very statistically significant. What this means is that something is very, very probably different. Some of this could be the quality of hitters faced or the level of competition of the hitters he does face, but it does seem to imply that something larger is different. A K/BB ratio of 12.5 for Park and just 2.33 for Happ is a huge difference. In general, I tend to believe that the team would be better off with a guy with league minimum salary anchoring the back of the rotation, since he would be able to do so for the next couple of years, but there is something different-- something statistically significantly different-- about this new Chan Ho Park. But who is he?