Which of MLB's most improved hitters of 2019 will keep their gains?

Though he entered Tuesday having collected just two hits over his past five games, Cody Bellinger is in the midst of a remarkable season. After setting a National League rookie record for home runs in 2017 but then falling victim to something of a sophomore slump last year, Bellinger spent the winter not only working to restore the mechanics of his swing to their 2017 state but to gain a better understanding of why they worked and how pitchers approached him. The results have been eye-opening. The 24-year-old slugger is tied for the NL lead with 42 homers, and leads in the FanGraphs version of wins above replacement (6.8 fWAR). He may well win NL MVP honors, and he even more clearly deserves to be called the league's most improved player.

Toward that end, last week I attempted an objective search for the majors' most improved players via a deep dive into the pages of FanGraphs' Season Stat Grid, which allows users to view as many as 11 seasons worth of data in a single category and track and rank year-to-year totals. I chose 10 statistical categories in which we might look for significant changes and awarded 30 points for the largest change in each category, 29 for the second-largest and so on down to the 30th. There were a couple other wrinkles, but the take-home was that Bellinger did indeed come out on top, with the Red Sox's Rafael Devers -- whose strong 2019 season had inspired my search in the first place -- running second.

This time around, I've rerun the rankings, eliminating one category (clutch) that's not really-skill based and eschewing my double-weighting of wRC+ and WAR, both of which are already double-crediting some of the other categories I did include. (Namely batting average, on-base and slugging percentages, strikeout and walk rates, out-of-zone swing rate, and fielding -- ultimate zone rating plus the positional adjustment.) I've also eliminated a tedious step of divvying up rankings points in the case of virtual ties; where two players might appear to have improved by 3.5 WAR but in one case it's 3.52 and the other 3.48, those players are already ranked correctly on the site. With all that in mind, here's the leaderboard and their points in the given categories:

The results aren't much different than before; Tim Anderson slipped out of the top 10, replaced by Mark Canha. But this time, I'm looking not just at the rankings but at which players are more likely to hold onto their gains, and which aren't, bringing considerations such as Statcast metrics, batted ball distributions, and so on to bear on my findings. Saying that a player won't maintain all of his gains shouldn't be considered a slight. These guys still qualify as among the most improved, I just believe that certain indicators suggest they're less likely to remain at this level.