Sabermetrics and the Cy Young Award

Visits: 1

Zack Greinke of the KC Royals

I think we’re all aware of which stats voters use to determine Cy Young winners. For the most part, it’s stats like ERA, wins, K’s, losses, and WHIP. There are a number of reasons why using only those stats doesn’t come close to telling the whole story – each of them is seriously flawed.

Let’s start with ERA. It’s used to calculate how many earned runs a pitcher gives up, on average, per 9 innings pitched. An earned run is a run that reaches base and scores without the assistance of an error. It’s pretty easy to see why this is flawed – all it takes into account are errors. It doesn’t consider range, BABIP, or luck. All of which are extremely important factors for a pitcher. If an outfielder falls over while attempting to catch a routine fly ball, and it ends up going for an inside the park home run, it’s the pitcher’s stats that suffer, as it’s considered an earned run, since no error occurred.

Then, we come to wins. I’ll keep this one short and sweet, as I think most people know by now why wins are a bad statistic. We’ll use Zack Greinke as an example here. He has 16 wins, compared to CC Sabathia’s 19. Has Sabathia been a better pitcher this year? Absolutely not. But Greinke plays for the Royals – a team with an awful offense, and an awful defense, and that has greatly contributed to both his low win total, and his high loss total. If he played for say, the Yankees, he would probably have 21-23 wins.

WHIP is a bad stat for evaluation pitcher’s for the same reason ERA is. It’s designed to measure, on average, how many hits + walks a pitcher allows per inning. Sure, walks are a useful stat to look at, but hits allowed really doesn’t tell you much, because of all the factors that can contribute to a hit (bad range of a fielder, luck, etc.).

There’s nothing wrong with using strikeout’s when considering the Cy Young award, but it has extremely limited use. It tells you, generally, how well the pitcher both misses bats and fools hitters, but that’s only one aspect of being a great pitcher.

So, I’ve explained which stats AREN’T good, but “which ones are?”, you might be asking. The two biggest ones, that really need to break into the minds of the voters, are tRA and FIP.

Now, both of these are extremely complicated stats to calculate, and I’m not so much a sabermetrician in the aspect of calculating statistics, I’m more of a sabermetrician in the aspect of putting these stats to use. So, if you really want to understand these stats, I’d advise you to check out the Statcorner.com glossary or the Fangraphs explanation of FIP.

Basically, tRA was designed to cover every aspect of pitching, rather than selective statistics that don’t come close to telling the whole story. The idea is to assign a run value to every type of event under a pitcher’s control (ground ball, line drive, home run, swinging strike, etc.), and then use it to come up with an expected number runs allowed, and outs allowed, without factoring defense or park factors. That way, it narrows it down only to things under direct control of the pitcher.

FIP, or fielding independent pitching, is a little bit different, and a little bit less advanced. It is designed to figure out how much responsibility a pitcher has for the runs he allows, based only on walks, strikeouts, and home runs. The formula for FIP, per Fangraphs.com, is (HR*13+(BB+HBP-IBB)*3-K*2)/IP. Granted, FIP does have it’s shortcomings, and isn’t as reliable as tRA, but it’s significantly better than any of the basic, more common stats. Also, FIP is one of the few advanced statistics, and probably the only advanced pitching statistic, that has started to break it’s way into mainstream sports coverage.

So, when do I see these stats becoming more popular with Cy Young voters? I have no idea, it could be 2010, it could be 2030. From what I’ve seen, though, the majority of fans/writers that aren’t interested at all in advanced stats, are older people. I would expect, that once the slightly younger generation starts to fill up more of the voter spots, the inevitable transition to advanced stats will speed up.