KATOH’s Top 100 List: Now Incorporating Multiple Years

Back in November, I published a top-100 prospect list that was generated by my KATOH projection system. Since that time, I’ve done some tinkering to improve the model. So even though we just did this, like, barely even two months ago, I’m back with another list for you. In addition to yielding lower AICs and R^2s and whatnot, this version also feels better in terms of the projections themselves. There aren’t as many head-scratchers as before, which suggests I’m moving in the right direction. There are still players who feel too high to me and others who feel too low to me, though I’d argue that’s not always necessarily a bad thing.

There’s still room for improvement. That will always be the case. But I think I’ve reached a point where I’ve gotten most of the low hanging fruit. This isn’t to say I’ll stop trying to make improvements, but don’t expect anything drastic anytime soon. This is the version I’m prepared to go to war with in 2016 (for now, at least). It’s not perfect, but I think it’s a sizable step in the right direction. Below, you’ll find a brief writeup of the changes, followed by a brand new top 100.

Moving Beyond WAR Through Age-28

In the past, a few commenters rightly pointed out that projecting a WAR total through age-28 wasn’t all that informative for some prospects. For a guy who’s already 25, his WAR through age-28 captures at most three years of team control, while it might capture all of a 21-year-old’s control years. In this iteration, I made my output a bit more flexible. For players 22 and younger, I still forecast WAR through age-28. But I expanded my horizons for players 23 and older, and now project the next six years. So 23-year-olds get a WAR thru 29, 24-year-olds get a WAR through 30, etc.

Obviously, this change benefits the players who are 23 and older. Aaron Judge, for example, grades out more favorably than he did in the past. (Though, he still fell just short of the top 100.) Jon Gray also benefited a good deal. This adjustment also makes age a bit less important in general. So the guys who have little going for them other than being uber young for their level don’t rise to the top as much. I’m looking at you, Samir Duenez. For now, at least, I’m calling the final output “WAR Thru First Six MLB Years,” even though that isn’t exactly what it’s projecting.

Other Tweaks

Players with very high stolen-base totals look better in my updated version. Burners like Tim Anderson, Jorge Mateo and Yoan Moncada benefit most from this change. Additionally, strikeout rate for hitters matters slightly less than in the previous version. Foiled again, Mr. Duenez. This tweak benefits strikeout machines like Joey Gallo. I’ll note that a hitter’s strikeout rate is still very important, but is noticeably (noticeably to me, at least) less important than in the last iteration.

Multiple Years of Data

This is the big one. For the first time, KATOH takes into account multiple years of data, rather than relying on only the most recent year. I came up with a weighting system by generating WAR values for historical players, and then regressing them against actual WAR totals. So guys who were great in 2015, but less great in 2014 aren’t automatically favored over players with longer track records. Max Kepler, who topped the list last time, is the name that comes to mind here. Best of all, I no longer need to cite “2014 KATOH” and “2015 KATOH” every time, which was confusing for everyone — including me. All players who logged at least 200 plate appearances or batters faced in 2014 have projections that consider multiple years.

So here goes. Again, I’m scouting the stat line, so this isn’t gospel. Take my math as seriously as you please. Minimum 300 minor league PA or BF in 2015 to qualify.

And so Marlins fans don’t feel left out…

…Sorry.