Rethinking MLE: The role of experience

Bill James famously introduced minor league equivalencies for hitters in his 1985 Abstract. This is among James’ most influential publications because the research debunked the widespread notion that minor league numbers don’t matter in projecting major league performance. There have been some advances since James first wrote about translating minor league numbers; minor league park factors are certainly better than ever, and there are methods for estimating MLE beyond Triple-A leagues.

Still, a lot hasn’t changed. For many analysts, the basic construction of hitters’ MLE still involves adjusting offensive rate statistics by a multiplier that is supposed to work for all players in a certain park or league context. There’s nothing wrong with that, and I suspect that method is good enough for many people. Over the next three months, however, I will examine some of the factors that affect the translation of statistics across professional baseball leagues. I’ll conclude that there is lots of room to individualize MLE based on player characteristics, but I won’t try and sell you on a brand new type of major league equivalency, and I’ll spare you from learning any new acronyms. I don’t believe any of that is necessary. Instead, I hope to simply expand the discussion of what makes up a good MLE and maybe inspire a change or two in some of the systems that are out there.

I have a few new ideas to share, but many of the topics will be based on common sense or widely shared knowledge among scouts and other baseball anlaysts. This is the case in today’s article, where I’ll take a look at the effect of repeating a league. Prospects who struggle in Double-A for a season and emerge as top hitters during their second or third tour around the league often inspire praise with caveats of “…but he’s repeating the league.” This seems like a reasonable concern given that comfort with a particular league—including knowledge of parks and some of the opposing pitchers and coaches—should positively affect the offensive performance of a league veteran. Hitters who are new to a league may be at a disadvantage.

Should this factor into major league equivalencies? I have started to look at the differences in how minor league statistics relate to major league statistics by dividing hitters into two groups. The “inexperienced” hitters have fewer than 300 plate appearances at the Triple-A level prior to the target season in which they play in both Triple-A and MLB leagues. The “experienced” hitters have at least 300 plate appearances at the Triple-A or MLB level prior to the target season. I restricted my sample to hitters between the ages of 23 and 27 with at least 100 plate appearances in both Triple-A and MLB within one season. Here are the results for strikeout rates in the 2004-2007 seasons:

                Hitters' Strikeout Rates 2004-2007
                 n  Age     AAA    MLB  Difference
Inexperienced   54  24.6  16.0%  18.5%  +2.5%   
Experienced     66  25.4  16.6%  20.2%  +3.6%

If you want to think about these differences as multipliers that are commonly used in estimating major league equivalencies, the experienced hitters’ strikeout rate increases by 1.22 times and the inexperienced hitters’ strikeout rate increases by about 1.15. The direction of the effects for walk rates are similar, though the effect of experience is not quite as strong. Depending on how you assess minor league hitters’ power, the effect of experience can be even stronger. These are the kinds of differences that were expected if we bought into the common belief that league repeaters are at an advantage. The implication is that minor league hitters adjusting to a new league should be penalized less than a minor league veterans’ performance when calculating minor league equivalencies.

Note that the average age of the players in the two groups is different; the average ‘experienced’ player is about nine months older than the average inexperienced player. Is age rather than experience the important variable here? Regression analysis suggests age is a non-significant factor once accounting for experience, but given the small number of cases in my dataset it’s difficult to conclusively dismiss age as a factor. It’s likely that many organizations only challenge young and inexperienced players at the highest levels of competition if the players’ tools are exceptionally well-regarded. I excluded super-prospects who reached Triple-A before the age of 23 from this study for this reason, but it’s still a concern that could be influencing the results we’re seeing here.

It’s clear that knowledge of players’ experience should factor into the interpretation of minor league statistics. For example, Jeff Mathis cut down his strikeout rate for Triple-A Salt Lake this season, but 2007 is Mathis’ third tour through the Pacific Coast League. When translating his performance, analysts are justified in treating his contact rate slightly differently than players without the benefit of much Triple-A experience prior to this year (such as Yunel Escobar, Brendan Ryan, and Kurt Suzuki). In fact, both Suzuki and Mathis are catchers of about the same age who both struck out in about 17% of plate appearances in the Pacific Coast League earlier this year. While Suzuki’s strikeout rate has remained about the same after he transitioned to the American League, Mathis’ strikeout rate increased to about 22% after his promotion. This is an exaggerated difference in outcomes at the major league level, but it’s consistent with the results shared above.

Why do these differences matter? Well, strikeouts are one facet of a hitters’ performance that influences the number of balls in play and therefore affects many of the offensive measures we really care about, such as on-base average. An individualized major league equivalency that accounts for experience rather than a “one size fits all” approach to translating strikeout rates might only result in a difference of a couple hits in full season’s worth of translated performance, but if this approach is applied to many facets of players’ performance the cumulative gains could be practically quite significant.

So, experience matters. And while the above results only scratch the surface of translating Triple-A rate statistics, this is a principle that can probably be applied to a number of performance characteristics across many different contexts. In any context, we should not presume to interpret inexperienced players’ performances the same way we would interpret minor league veterans’ performances.