What Minor League Stats Really Mean
by David Luciani
Published April 11, 1998

It has been more than fifteen years now since Bill James first wrote that minor league data meant something and it could be understood. James told us all that inevitably, major league baseball teams will eventually have to accept that. Surprisingly, major league GMs have paid too little attention to James’ philosophy and are suffering as a result of it.

Quite simply, minor league statistics do mean something and perhaps the difficulty in accepting them has been because no one really knows how to read them. Even the so-called equivalencies that have become popular are useful but tend to over-reduce some columns and under-reduce others.

I thought that if minor league numbers do mean something (and they do), then it would follow that some categories would translate to major league ability more than others. Before looking for data, my hypothesis was that a batters’ home runs would drop but that stolen bases would drop less, except as a result of a likely drop in times on base. Walks, I thought would go down and strikeouts would go up. Intentional walks would virtually disappear, at least in a player’s first year or two.

In order to “solve” the problem, we took data on every player who appeared in the major leagues the past four seasons and compared what they did in the majors to what they did at any one of a number of minor league levels in the same season. What a player did at Triple-A a year before playing in the majors doesn’t tell us what we want to know. A player can improve or decline during that year. We wanted years in which the player played in both the minors and majors and played a significant part in both seasons.

We then adjusted data for the environment in which the player appeared. If a player played half of his major league games in Oakland and half of his minor league games in Edmonton, we needed an adjustment for that. For pitchers, we needed to look at the league itself. Was it a hitter’s league or a pitcher’s league?

We made adjustments for playing time so that a player who had 500 plate appearances at Triple-A but 100 plate appearances in the majors has the 500 plate appearances at Triple-A reduced accordingly to match the context. We also performed a probability analysis on it to determine whether we had enough significant data to draw a conclusion.

What we discovered was that for batters, anything as far as down as the most competitive Single-A leagues, such as the Florida State League, can include significant data. For pitchers, anything below Double-A did not have any consistency whatsoever in relation to major league performance. Excluding playing time, we then examined everything on a per plate appearance (for batters) or per batter faced basis (for pitchers). Here were the results for the Triple-A level, results we often use in making projections of recently-promoted minor league players:

AAA BATTERS (% TRANSLATION PER PLATE APPEARANCE)

AAA PITCHERS (% TRANSLATION PER BATTER FACED)

Note how much more a batter is asked to bunt in the majors. The best Triple-A hitter is just another bench player in the majors and the frequency of sacrificing reflects that. The speed translates pretty well, essentially the same when you consider how less frequently the batter will be on base in the majors.

For pitchers, look at how little Triple-A saves mean. Your 25 save guy from Triple-A becomes a five save guy even if he faces the same number of batters in the majors.

Obviously, you can’t just translate statistics and assume that it tells you exactly how a player would have done in the majors. You need first-hand knowledge of the player and witnessing his talents in person is the ultimate test. However, this tool gives you an accurate forecasting tool that can be used at least as a starting point, in combination with first-hand knowledge, to achieve better results.