Discussed: K rates, standard deviations, batted balls of varying heights, modern portfolio theory?, Sam Hinkie
When it comes to bullpens the Rockies don’t have the best but also they don’t have the worst. They’ve been dependable, especially compared to the Rockies’ starting pitching. But can the bullpen be depended upon to stay dependable?
From a certain perspective the answer looks like a pretty obvious no. They are making strikeouts at a miniature rate–7.93 per 9 innings pitched–over 2 standard deviations below the National League bullpen mean, which combined with their average walk rates put them about 1 SD below average in K/BB. This isn’t ideal because bullpen guys tend to inherit traffic on the bases or pitch in situations where avoiding traffic on the bases is extra important, making strikeouts and their .000 batting average the safest way to proceed outcome-wise.
Colorado’s bullpen has logged a FIP of 4.34, good for sixth best in the NL, but also good for just better than average in the NL (4.43). This is likely due to the group’s suppression of home runs since only 12.2% of their fly balls have resulted in an HR. That’s second lowest in the NL and about 1.4 standard deviations below average. That’s why from the standpoint of xFIP the Rockies are fifth worst (4.66), with the highest xFIP-FIP (0.32).
After assembling this data if you were to declare that there might be some less successful times ahead for this bullpen nobody is probably going to get real gruff with you. But it’s worth looking at things from multiple perspectives. Are there other ways to skin the bullpen’s cat?
One skill the Rockies’ bullpen does seem to possess is the ability to cause hitters to hit ground balls: around 46% of the time, 3% above the NL bullpen average. Ground balls are a slightly worse but legitimate alternative to strikeouts-as-a-strategy. Generally speaking, balls on the ground aren’t hit high enough to go over the fence, so they have the same safe-from-home-run feature as Ks. But there is real data1 in support of ground balls: they produce only about 0.05 runs per out, around 7x less than balls in the air. On top of that, just as a bonus, ground balls can make double plays more frequently than any other baseball event.
This groundballing of the Rockies’ relievers is also some insurance for the assumed regression to come of their home run rate. Because even if the rate at which they allow home runs per fly ball goes up the actual number of home runs should still stay relatively low. We also shouldn’t be allowed to only assume regression of one statistic but not all, which means that the bullpen’s K/9 should increase somewhat according to math.
Going forward the Rockies’ bullpen should continue to be deemed mostly dependable given the general skillset it possesses. However, is “mostly dependable” what you want for a group that comprises 33% of your payroll?2 Is being better than average at producing ground balls a premium dollar skillset? On one hand it may seem like a risk-averse strategy: pay a premium for dependability. This would make sense if we weren’t dealing with relief pitchers, the most volatile commodity in baseball with the smallest reliable dataset. Sam Hinkie3 said, “to have real success you have to very often be willing to do something different from the herd.” Allocating mucho payroll for mostly average results from aging relief pitchers is definitely different, and there are worse things than mostly dependable. But also better ones.