Glossary

As I explore how well a player did on the field I use a variety of statistical techniques. There are many traditional baseball statistics, which are generally well known, and the last thirty years or so have seen the development of what are called “sabermetric” or “advanced” metrics. The latter were created out of a realization that the traditional stats were only fitfully capturing the real value of a player. They’ve become progressively more mainstream, especially in the last decade or so, but depending on how you take your baseball fix you may not know too much about them. I’ll do my best to explain them here.

Slash Line: A traditional method, it consists of three numbers: batting average, on-base average (OBA or OBP), and slugging percentage. A typical line may look like .312/.373/.508. The latter two numbers are two of the best of the traditional stats. Batting average has fallen from grace in recent years, and deservedly so, but I include it for a few reasons. One is that it’s descriptive in the context (a player who hits .250/.350/.450 is of a very different type from one who hits .320/.350/.450), and also because it’s how a slash line has been portrayed historically. I also personally find high batting averages esthetically pleasing, even if they don’t mean very much on their own.

OPS+: This measure acknowledges the usefulness of OBA and slugging, and combines them into a single number for convenience. Simple OPS is just the two added together, while OPS+ involves an adjustment to the simple sum that accounts for the context of the time and place. Some years have higher offense than others due to rule changes or trends in playing styles and officiating. Other teams play in parks where it’s easier or harder to hit. OPS+ adjusts for these.

A higher OPS+ is better, and an OPS+ of 100 means the player was exactly average at hitting for his time and place. The major flaw with this stat is that it undervalues OBA. Two players on the same team, one who hits .300/.400/.400 and one who hits .300/.350/.450, will have the same OPS+, but the first of them is actually the better hitter.

WAR: This is an very advanced metric, meaning “you’re not calculating it quickly”, generally considered by the cognoscenti to be the current best single number that can be hung on a player’s over all value to his team. It includes both hitting and defense, as well as baserunning contributions, or can be broken down into two pieces if you want to discuss solely his value on offense or with the glove. Strictly speaking it’s supposed to represent how many extra wins a team would pick up over a season if:

  • All of their players were exactly replacement level, which is to say the quality of a player who is easily obtainable. Not a quality regular, but a “quadruple-A” player that a team could pick up for next to nothing if they needed an emergency replacement.
  • On Opening Day you then swapped out one of these replacement level guys with the player you’re giving a WAR number to, then let the resulting team play 162 games.

A good number is 5.0 (WAR being conventionally rounded to tenths of wins), 8.0 means you’ve got a really good shot at winning the MVP that year, and anything over 10 is a season for the ages. Babe Ruth got over 10 for nine of his 22 seasons in the majors, while Barry Bonds did it three times. Even a player as good as Albert Pujols has never done it, just to give you an idea of how hard it is.

There are two different standards for WAR calculation, one by Baseball Reference and one by FanGraphs, and they don’t produce quite the same number. I use Baseball Reference’s WAR, for essentially arbitrary reasons.

Similarity Score: A big question on this blog is “What would this player have been like if he continued to play as he started?”

This is a hard question to answer, but my own personal preference for approaching is to compare that player’s star-level career to other players whose numbers were similar up to the same age. To do this systematically, I use the similarity score tool made available on the Baseball Reference website. The details are not that important, just understand that it calculates a rating from 1-1000 for how similar two sets of hitting statistics are to one another, and then shows you the ten players who are most similar to the numbers in question. These players I call “comps”, just for compactness’s sake.

So for example, I could give it Mike Trout’s career up to this point and Baseball Reference will spit back the his comps through age 24 (Trout’s current age). If you’re curious, they are:

  • Mickey Mantle (the most similar player)
  • Ken Griffey, Jr.
  • Hank Aaron
  • Frank Robinson
  • Mel Ott
  • Miguel Cabrera
  • Orlando Cepeda
  • Vada Pinson
  • Al Kaline
  • Jimmie Foxx

Which list goes a long way toward explaining why so many people are excited by Mike Trout—so far he’s hit like eight Hall of Famers, Cabrera who’s only not in the HOF because he hasn’t retired yet, and a very good player in Vada Pinson.

Similarity scores have their problems. There’s no adjustment for era or home park, and it doesn’t consider defense at all. But it does pretty well as a way of staying consistent about evaluating a player’s hitting past and possible future.

High/Low/Average Cut: This is my own twist on the similarity score results above. Occam’s Razor suggests that if a player has, in the past, hit like these players, then he will continue to do so. It’s easy to get Baseball Reference to spit out post-age 24 stats of the ten players listed above, and so from there we can generate a good guess at what Mike Trout will do for the remainder of his career. None of the three methods I outline below are perfect by any means, but they all beat just arbitrarily blurting out “Mike Trout is gonna have 3000 hits! And hit 600 homers!”

The first possibility is to average the ten players remaining careers (using arithmetic mean, if you want to be precise, though median has its good points) and add them on to the actual stats so far of our player in question. The only problem with this is that it tends to produce underwhelming career totals as it includes the guys who haven’t played a full career yet and the guys who tailed off badly (Pinson and Cepeda being the ones on Trout’s list). The whole point of this blog is to explore what these players might have done, not what they were likely to have done. Even so, this technique, with the unretired players removed, is what I call an “Average Cut”.

The next approach would be to just use the one player who had the best career post- the age in question, but this has the opposite problem quite a lot. Some players luck out and get a top comp whose career was almost unattainably long and productive. Trout’s one of these, in that Aaron is his #3 comp, and so adding together Trout-to-24 and Aaron-from-24 produces a career line of 3377 hits, 783 homers, and 161.2 WAR, or (to put it another way) a guy who’s right in there with Babe Ruth as the greatest player who ever lived. I’m as much of a fan of Mike Trout as anyone else, but this is a step too far.

For me the way forward comes, oddly enough, from another of my interests, astronomy. When studying star clusters, you sometimes end up in a situation where a foreground star or one peculiar star actually in the cluster is much brighter than the others. This throws off calculations about how bright the cluster is, and you can’t use median because a cluster might contain ten thousand stars and boy is it a pain in the ass to measure them all and put them in order. Instead what astronomers will do it take the mean of the top ten stars, or of stars five through ten, or something similar.

Applying the same technique to the comps, I use the average of the top three most productive comps to guess what the player in question might do (or could have done) over the rest of his career. I call this the “Average Cut”. It keeps the Hank Aaron types in there, because anything is possible, but throws in a dose of reality in the next two guys.

Take Trout and his top 3: Aaron, Robinson, and Ott. Averaging out the post-age-24 careers of these three and then adding them on to Trout’s career so far produces a line with 2940 hits, 625 homers, and 135.5 WAR. Still absolutely incredible numbers, the ninth best player of all-time by WAR, in fact (slotting neatly between Roger Clemens and Tris Speaker), but one that I’m comfortable with as the answer to the question “If things go pretty well for Mike Trout over the rest of his career, what do you think he will end up at when he retires?” You may think differently, but I don’t think you can argue that it’s unreasonable.

What I call “The Low Cut” is the opposite: take the three worst-performing comps and make their average the remainder of the player’s career. This is less useful, but since a lot of the players we examine did even worse than this, it sometimes informative.