September 3, 2013
Do Young Pitchers Fail to Develop When the Bullpen Implodes?
Autumn came a little early to Houston this year. You might not have noticed, but the Astros recently became the first American League team to be formally eliminated from the playoffs. It’s not that anyone really expected the Astros to contend this year, but then again, I picked the Angels to win the World Series at the beginning of the year. Shows what we know.
The good news is that the Astros can now focus their efforts on next year (or 2015). And some of their neighbors have been kind enough to offer a little advice. After a recent Rangers-Astros series, this quote appeared in a column by Randy Galloway of the Fort Worth Star-Telegram.
Bad teams have bad bullpens. That’s always a given. The Astros may have the worst bullpen in the history of baseball. The Rangers wait out the starting pitchers and then bring out the fangs on that bullpen.
The talk in the Rangers clubhouse this week centered on high praise for the collection of good young arms in the Houston rotation. Four of the five starters are considered double-plus prospects based on what I was told.
“The owner down there needs to get off his wallet and spend some money for some veteran bullpen guys next season,” said one Rangers voice. “Those kid starters have a future, but they are being beaten up mentally by pitching well and never getting a win out of it. That’s a shame for them, but, yeah, good for us, I guess.”
Pity the poor Houston starters, I guess. The Astros, as of Saturday night, had notched 25 blown saves collectively, running neck and neck with the Diamondbacks and Cubs for the dishonor of leading MLB in that category. (Kevin Chapman added another on Monday.) It must be rough for a starter, particularly a young one, to know that if he goes out there and throws six strong innings and leaves with a 5-3 lead, that all that hard work might be undone in the form of not getting a “win.”
On the surface, it makes sense that a young pitcher might suffer if he has one too many of those outings. And given that the Astros are going to need several of their young starters to develop over the next few years, it makes sense that they might want to avoid traumatizing them before they have a chance to develop. But before we get too far into this, is there actual evidence to suggest that not getting a win actually impedes development? It’s easy to write a narrative that leads from a starter having to deal with a blown save and use words like beaten up mentally. But if there’s something I’ve learned over the years, it’s that these narratives are often far too simple. This is an argument for the blown save being some sort of major trauma, and the psychologist in me knows that trauma is very complex. Before the Astros (or anyone else) goes off on a reliever spending spree in the off-season, let’s take a look to see what the evidence says.
Yes, the win is a silly statistic, but pitchers do value it. There is evidence that pitchers change the way that they pitch (if not that they pitch better) when it looks like they might need a little extra to have a chance at a win. But on the flip side, a few years ago, the Orioles went on a reliever shopping spree and signed Mike Gonzalez to be their closer. (This was back when “Mike Gonzalez” and “closer” in the same sentence still made sense.) The idea was similar: that the Orioles, who were in the middle of a decade-plus streak of losing records, needed to build a “winning culture” to allow their young position players (like Nick Markakis, Adam Jones, and Matt Wieters) to develop. I found little evidence to support that conclusion.
Well, let’s see what the evidence says about pitchers’ development without bullpen support.
Warning! Gory Mathematical Details Ahead!
To measure development, I looked at walk and strikeout rates (per batter faced) in the year in question and then the next year. Walks and strikeouts are generally the outcomes that most quickly stabilize for pitchers and are the most predictive of success. I used a method that I introduced a few months ago known as the Reliable Change Indicator (RCI). This controls for the fact that when we’re looking at yearly rates for something, the underlying sample sizes will be different and thus the reliability of that measure will be different as well.
For example, let’s say that we have two pitchers, both of whom have a strikeout rate of 12 percent in year one and improve to a 14 percent mark in year two. The difference between them is that one faces 500 batters in both years while the other faces 300. We are more confident that the pitcher who faces more hitters improved. The RCI method adjusts for that and gives us a standardized measure (think z-score, for the initiated) of how much a player has changed. (Later, I looked at just raw, unadjusted differences in strikeout and walk rates. Same findings.)
Once I had those numbers lined up, I looked to see whether the percentage of times that the bullpen blew a possible win (or the raw number of times that they did so) predicted whether the pitcher would increase or decrease his strikeout or walk rate in the next year. If the unnamed Ranger source above is correct, we should see that starters who are victimized a lot by blown saves should show a reduction in their strikeout rates and an increase in their walk rates.
There was just one problem. It didn’t happen. The correlation between blown win percentage and change in strikeouts was .055, and the correlation between blown wins and walks was -.009. Whatever effects there might be were actually pointing in the direction of pitchers getting better. I looked two years out into the future, and blown wins were actually correlated with pitchers seeing their walk rate going down (r = -.104, p = .018).
I did want to double check was to make sure that we weren’t dealing with an overly biased sample. One problem with baseball research is that the sample is always biased. Good players get to play more often, and bad players are simply eliminated from the data set (that is, sent to the minors or just not brought back). Because I had initially required that the pitcher have 15 starts both in the initial and follow-up season, maybe there was a group that grew so dispirited in the first season from all the blown saves that they simply fell apart and were just not around to be studied the next year.
So I took the group of young pitchers that made 15 starts in the initial year and coded whether they got to 15 starts in the following year. I controlled for initial strikeout and walk rates (because another really good way to be eliminated from the data set is to be a bad pitcher). Blown save rates did not predict survival to the next year as a 15-start guy.
Quick! Change That Narrative!
What I found giggle-worthy was that what (minor, insignificant) effects were present actually pointed to pitchers getting a little better in the year following a season in which they endured a bunch of blown saves. Quick, let’s change that narrative! Blown saves provide adversity for the pitchers to overcome, this adversity builds character, and that character leads to strikeouts! It’s funny how no matter which way the arrow is pointing, we have a ready-made, all-too-simple MadLib that the facts of the story can be shoved into.
If our starter is being rational, he will say to himself that he did what he could and ran into some bad luck. Short of sitting at the top of the dugout step and tapping the triangle button over and over, it’s not like he can control his teammates. If he can’t make that rational jump, then that’s where some sort of mentor or (dare I invoke him again) “clubhouse guy” comes over and talks to him. As I’ve said before, a baseball team is a small society, and every society has a way of dealing with loss. If a team does not, they need to fix that.
Should the Astros sign some veteran relievers for next year to shore up the late innings for the benefit of their young starters? The evidence here suggests that it will make little difference. It’s not that the unnamed Rangers source who provided the quote is entirely wrong. There probably is some disappointment about losing a win to a blown save, and for some guys, it might be an issue. But the way to address that isn’t by signing relievers whom you might not need. There will be guys who regress after having a bunch of their hard-won leads vanish, but it might not be because of that. The lesson to be learned here, once again, is that the impact of psychology on performance is much more complicated than we often acknowledge.