In previous articles, I’ve shown defensive ratings using basic play-by-play data from Retrosheet. All I really need to know to get good defensive ratings is what type of hit it was (ground ball, fly ball, line drive) what plays were made defensively and by whom, and who ended up fielding every hit. I’ve used this data to produce major league fielding ratings for seasons 2003 to 2008. (The method is a little different before that due to data availability.) There is no reason the same method cannot be applied to the minor leagues, except that I didn’t have a data source.
Enter Jeff Sackmann, fellow THT writer and the owner of Minorleaguesplits.com. Jeff was able to produce exactly the data I needed from the MLB gameday data he uses to update the minor league splits. Since he sent me the data, I’ve generated fielding ratings for every player from rookie ball to the majors for years 2005 to 2008.
Here are the top three fielders, in runs above average compared to their league, for 2008, by position:
Ramirez is an incredible fielder, though his bat appears to be inadequate for the major leagues. Over the last four seasons, he is +86 runs in center field at various minor league levels; he has the best minor league defensive stats of that time. Hernandez, only 20 last season, has shown enough patience to make you think he could develop into a decent top of the order hitter, though his bat is nowhere near ready. He rates at +53 runs over two-and-a-half minor league seasons.
Are these ratings useful?
For the ratings to be useful, they need to correlate from one level to another. If we knew a player had a +15 rating one year, that would be of no use unless it told us that he was likely to continue to post good ratings in future seasons, at higher levels of the minors. Preliminary investigation shows that these ratings are more useful for infielders than outfielders. The correlation is much lower for outfielders, though at least they are (usually) positive. I’ve found that you can usually get a correlation of 0.50, meaning you regress 50 percent to the mean, at about 350 chances for infielders. This represents less than a full season of chances for second, third and short. For outfielders, you need about two full seasons of data to regress 50 percent, or about 1,000 chances.
The low correlation for outfielders could be due to park effects (at this point I have not measured these), and data quality. For the most part, a ground ball is a ground ball, and there should be little question on how infield plays were scored. For outfielders, it is probably impossible to ensure that the standards for what is a line drive and a fly ball are scored consistently. It would improve data greatly to ditch the classifications and record hang time instead—how long between the time a ball hits the bat and it hits the ground, wall, or glove of a fielder. These ratings for outfielders may not be very useful for these reasons, you might do better by just looking at player speed scores, but I have not et studied this.
How do they translate from one level to the next?
For infielders, players perform worse, relative to the league, as they move up in levels. In other words, major league third basemen are better fielders than Triple-A third basemen, who are in turn better than players in the lower minors. For outfielders, the picture is less clear. For center fielders, the relative performance is relatively flat. Give or take a run or two, the average center fielder in Low A is about as good as the average major league center fielder. For corner outfielders, it appears that the quality of fielding, by looking at players who move up in levels, is lower as you move up in level, with major leaguers being the worst!
This was a bit hard for me to believe. There are some reasons that it could be possible—players lose speed relatively quickly, and outfield range is highly dependent on speed. Also, the lower levels of the minors use the DH in every game, which probably cuts down on the Adam Dunn and Pat Burrell types in minor league outfields. Another thing to consider is that players like Dunn were not the same lumbering plodders in the minors. At age 21, he was likely a good bit lighter on the scale and faster in the field. Major league outfielders, especially at the corners, are selected more for their bats than their gloves, and while having a lot of bulk muscle helps you hit for power, it does not help you chase down fly balls.
I tried looking at it from another perspective. Instead of looking at the guys who switch levels, I just looked at the total flyball and line drive out rates in the major leagues, as opposed to Low A. It’s not even close; fly balls and line drives are more likely to become outs when major leaguers are in the field. This doesn’t consider parks (both dimensions and quality of lighting), batters or pitchers, but it is at odds with the translations from level switchers. For this reason, I think the best option is to not adjust minor league outfield ratings, but to treat them as equal to major league ratings. A +20 outfielder is a damn good outfielder, regardless of what level of professional baseball he’s playing at.
Here are the infield “MLE” translations used, per 500 chances, by level of play from rookie ball to Triple-A:
|Position||Rookie||Short season||Low A||High A||AA||AAA|
How does my top prospect rate?
Alcides Escobar, a shortstop who is the Brewers’ top prospect, was +22 runs last season. Scouts rave about his defense. He was about average in 2007 and rated poorly in 2006. This could be a case of a player with great tools who just recently developed consistency.
Braves outfielder Jason Heyward was +13 in Low A last season. St Louis third baseman Brett Wallace was –6 runs at two levels; he may not be able to stay at third. Red Sox first baseman Lars Anderson was –4 runs at two levels, an improvement from –18 runs in 2007. He might be better off as a DH.
Rangers shortstop Elvis Andrus was +2 runs last year and about average over the last four seasons. Scouts love him, but defensively, he’s not the king yet. Marlin center fielder Cameron Maybin’s ratings have been all over the place. After a +18 2006 season, he followed it with a –14 in 2007. He was about average last year. Matt LaPorta was –14 runs for Huntsville in right field, but improved to +4 in left for Akron after the trade. He’ll probably wind up back at first. Blue Jay outfielder Travis Snider, another who doesn’t look like an outfielder, rated average last year but was an awful –28 in 2007.
Cardinal Colby Rasmus was +9 last year and has consistent ratings, for a total of +30 runs over four years. Brandon Wood ranks a bit below average at both short and third, with four-year totals of –11 at short and –5 at third. He probably lacks the range to play short but with experience should settle in at third. Andrew McCutchen is +21 over four years in center, he may push Nate McLouth to a corner this season. Jay Bruce rates as +4 in center and +8 in right, with a little more than a full season’s worth of chances in center and less than a full year in right.
Where can I find more data?
Jeff Sackmann has already loaded the data to the player pages on minorleaguesplits.com. My goal is to have defensive pages available on my site, baseballprojection.com, but I cannot give you any timetable for this.