I have been thinking a lot about J.P. Crawford’s midseason ranking and the quotes from unnamed scouts and executives thrown at him as different public sites went to bury him. On some level the criticism was fair. When Crawford missed 10 days of games in mid June he was batting .194/.313/.252. As is well documented in Phillies circles, since coming back on June 20th, Crawford is hitting .284/.388/.585. He has been hitting even better of late, carrying his hot July right into August. I then saw this is in Baseball America’s Hot Sheet Chat
Darren (Philly): It is usually a mistake to scout a stat line, but is it better to view JP Crawford's season by doing just that and moving on rather than overanalyzing the awful start or more recent month and a half or sustained success?
Kyle Glaser: It’s best to be moderate in your expectations of J.P. Crawford. You have to remember he struggled badly for a full year at Triple-A, it was the most of last season in addition to this season too. But he plays a very good defensive shortstop, gets on base and is making needed approach adjustments to be a threat at the plate, as he has been this past month. Is he a perennial all-star shortstop who is a franchise cornerstone? No. But an SS who plays good defense and gets on base, even a bottom of the order one, is valuable at the big league level.
It made me want to go deeper into some theories, so warning right now this is going to be more free form and it is going to not really be about J.P. Crawford, it is more about how we evaluate baseball players.
It has been drilled into us by statistical authorities that sample size is king. That over a long enough period of time stats should stabilize. The thing is that in reality there is a difference between stabilization and future prediction. We also need to reevaluate what is a large sample size, and to do this we need to ask these questions:
- Am I measuring past performance or trying to predict future performance?
- Am I looking for what happened or for what should have happened? (performance vs true talent)
- Do I think my full sample size is of the same person in the same state of being?
It is a very different question if I ask you “Who was the most valuable pitcher in baseball this year?” than if I ask you “Who is the best pitcher in baseball?”. The answer may be the same, but the methodology to getting to that answer is different. We can deal with that third questions by looking at a player who played a month with a known minor injury and their performance suffered. If we want to know how valuable they were to their team we can look at the full season. If we want to know what a healthy version of that player is capable we can cut the injured part of the season out, or if we think that the injury is recurring and we can predict this in the future we can leave that part of their season into our analysis.
When we deal with the minor leagues it becomes clear that stability is not really something we get afforded. Players are constantly challenged by a changing environment and changes aimed at making them better major league players and not at making them better minor league players. Then you add in contextual factors like ballpark factors and an ever changing strength of schedule and you have a lot of noise in a standard stat line.
What happens if the player completely changes? We do this somewhat with pitchers, because we can ignore their stats more easily and focus on things like a velocity increase. For example last year it was very obvious that Sixto Sanchez throwing 95 to 98 was not the same pitcher who was throwing 91 to 95 start the year. Sometimes it is more subtle. This year Adonis Medina scrapped his curveball in favor of a slider and his strikeouts took off. It seems logical then that are sample size be contained to 2017 when the slider is part of the arsenal rather than looking at the past two seasons and worrying about his lack of strikeouts. With a pitcher we can measure things like velocity and pitch usage, and so we can tangibly grasp the change in profile. But how do we do the same thing with a hitter?
We saw this early in the year with Scott Kingery. Kingery hit for much more power this year than he ever has in his career, and it was completely unsustainable. The unsustainability did not come from his past without power though, because ultimately, it was irrelevant. We knew that Kingery had changed his swing, and we knew that he gotten stronger and the change was clear in his forearm strength. We could safely say that the power production of 2016 was a completely different player. We then had to evaluate his 2017 season in its own context, and we found that his power output was unsustainable due to the kind of contact he was making and how hard he was hitting the baseball.
This brings us full circle, because outside of injuries, one of the most tangible things we can see change in a hitter is their swing. There has been a lot written to show that basic physics work and that if you hit a ball hard and you hit it in the air, it tends to go far, and extra base hits need to be hit far. We also can visually see how swing can affect ability to make contact, not just good contact, because it affects timing and ability to adjust to a ball in flight. What we do know about J.P. Crawford is that his swing has changed. I am not going to rewrite Jeff’s breakdown of the process, so go read it here. We also know that Crawford has a great batting eye and we know that his body control has always been there.
The big question then becomes, does J.P. Crawford’s first half matter at all? The answer is of course it matters, but is it more of a cautionary tale than something equal to his second half. Unless we think that Crawford’s swing and approach at the plate from April to June will be part of his game going forward, we cannot include it in our calculations. Now are Crawford’s second half numbers going to translate to the majors? Probably not, but that is once again, not because of what he did in the first half. Pitchers have yet to really force him to make adjustments over the past two months, and they are going to make him uncomfortable, because advanced scouting is really good these days.
Baseball players are humans, and we are constantly surprised when they change. I don’t think the answer is to try and smooth over all of the changes and hope they even out. Evaluation is a constantly evolving process, and the past is important because it gives you context, but it does not set destiny. You can’t predict breakouts blindly, the best you can do is hope to spot them and adjust accordingly.