I have an irrational fear about Pete Kozma's upcoming season.
It's a fear that stems in large part from closely following politics and economics in the news. It's spread, however, to most any application of math at this point. The resurgence of disproven or unlikely ideas that wait for a perceived opening to make a come back -- or ones that don't wait for an opening at all.
Part of my fear stems from the fact that a surprisingly large number of off-blog Cardinals fans I know are perfectly content with Pete Kozma at shortstop. It's not that they're content because he's part of a larger vision of the team that permits decreased output at one position given the increased output at others. Rather they are content because they're pretty sure Kozma is going to be great.
That's a valid opinion and there is some evidence to suggest it's possible. We should note, however, that the overwhelming amount of data predicts a very subpar year is ahead of our current starting shortstop.
That's the math side of things though and math rarely predicts with a certainty. (Perhaps that is better said that math never does predict with a certainty though it's results are sometimes interpreted as such.) When Dan Szymborski runs a ZiPS model, the output for Pete Kozma is not .226/.284/.328 but rather a range of outcomes each one with a probability attached. That final slash line of stats that gets posted at Fangraphs and other places is simply the mean outcome. The average.
So if Pete Kozma overperforms, that shouldn't be terribly surprising. If he underperforms, that also shouldn't be terribly surprising.
This is part of why the argument has evolved beyond either scouting or sabermetrics into scouting and sabermetrics. The reason that people have come to believe Pete Kozma will be better than his mean ZiPS projection is, in my estimation, two fold:
- Recency Bias - Pete Kozma was a pretty special player for 82 at bats last year. It's hard to get that memory out of our mind or least to place it in the proper context of the larger body of work.
- Subjective Scouting - There does seem to be something different about Kozma's approach at the plate. That may be a fluke of the hot streak. It may be a cause of the hot streak. Projection systems like ZiPS are ill-equipped to register and properly weight fundamental changes in approach like swing mechanics.
So this is where I lay the groundwork for context on the season. If Pete Kozma blows us all out of the water and hits his 95% projection, it's important to remember that result doesn't invalidate ZiPS or math or sabermetrics. (Somewhere, someone is hitting their 5% projection and someone else is hitting that mean projection right on the nose.) Rather it means that there's a lot dependent on the person. This year we get to see what Pete Kozma can offer in terms of beating his projection at a personal level.
I still pine for Elvis Andrus though.
* * *
It looks like Mitchell Boggs will step into the Jason Motte sized void at closer. This is a bit of a downgrade though I'm unconvinced it is a significant one. Relievers work in such small sample sizes that it's hard to differentiate between luck and talent aspect of their role. If Mitchell Boggs blows the first 10 saves of the season, there will undoubtedly be questions about his mental fortitude and how well suited he is for the role. If he is a perfect 10-for-10 to start the season, Motte will be largely unmissed. It's hard to imagine a scenario where the loss of Motte and his replacement by Boggs shows in a (non-lucky) way that can be deciphered over the duration of Motte's projected time on the DL.
The more interesting result would seem to be that the move allows for the inclusion of both Joe Kelly and Shelby Miller on the roster. As Dan noted a few weeks ago, this might be Joe Kelly's last best chance at starting. I think Shelby Miller is the better pitcher though it's unclear at this stage in their development how wide that gap is.
Risk mitigation is what concerns me most about the potential move of Joe Kelly to the pen. In a year where the Cardinals do not have a high degree of certainty in their starting pitchers or the length of their outings -- I'm looking at you Jaime Garcia -- it strikes me as more risky to move Joe Kelly to the pen where he's pitching for shorter durations and is less equipped to step back into the rotation should disaster strike.
That statement presupposes evidence, however, that I've not seen in a statistically rigorous study. The question is whether a) Kelly will be less effective transition from reliever to starter than from minor league starter to major league starter mid season and b) is Kelly more at risk for his own injury transitioning from reliever to starter than from minor league starter to major league starter. The answer to both seems to speak in favor of keeping him stretched out but, again, I've seen nothing conclusive.
This also has the interesting ripple effect of opening a rotation slot in the Memphis rotation. Look for wonderboy Michael Wacha to step into that hole.