NEWSLETTER
Every year, Spring Training rolls around, baseball fans dust off their score cards, players show up in the best shapes of their lives, and we are reminded that none of the stats we are about to witness matter.
Surely, most of us agree with that last statement, having been burned by the “Super” Joe Mathers and the (2014) Mike Olts of the Cactus League one too many times. Those might be emotionally charged examples, but the message behind them has been driven home for years: Spring Training statistics don’t predict future regular season performance.
In Spring Training, the argument goes, players are working on different areas of their game, there is an erratic fluctuation in the quality of the competition, and everything is tied up neatly in a super small sample size. Spring stats, then, can’t possibly have any meaning whatsoever.
But maybe we are wrong. Maybe Spring Training statistics aren’t completely useless.
At the outset of Spring training, the Economist took to the issue of Spring Training stats and determined that they do have some correlative power with future regular season stats. One such identifiable case arose with strikeouts per at-bat (quick! what’s Baez at?). Much like a strong correlation from season to season, the Economist’s research found, one can generally rely on Spring Training strikeout rates to predict the upcoming season.
You may have already known or anticipated that, though. More interestingly, the author utilized peripheral Spring Training stats to augment and revise ZiPS projections. Surprisingly, the resulting projections proved to be more accurate than ZiPS, alone. The implication here is that some Spring stats do wind up correlating, at least a little, with some regular season performance. That appears to be especially true for younger/inexperienced players (which sounds about right).
The only rub there is that the article isn’t entirely clear about how this methodology was employed, or precisely how we could further analyze or build upon the work to verify the value of the peripheral Spring stats.
Ultimately, I don’t think we can come to a conclusion just yet on the true value of Spring Training statistics. Although the analysis here is certainly interesting for discussion purposes, we have years and years of examples of poor correlation between Spring Training performance and subsequent regular season performance. The lesson here is certainly not “Spring Training stats matter!”
Instead, the broader lesson here is simply to keep an open mind. It is always in our best interest to use the best available data to better understand, analyze, and predict the game. That should be the predicate for challenging deeply-held beliefs, whether those beliefs emerged from “traditional” schools of thought, or “modern” ones.
In other words, even if your well-developed sabermetric background has taught you that Spring Training stats don’t matter and aren’t predictive, it’s always good to be open-minded about the possibility that you might be wrong in some small way.
