Saturday, February 25, 2012


Ever think you are living in a time when the game of baseball is, well, less than it might be? That the game has reached a level of uniformity that is disturbingly akin to that metallic taste that occurs in overprocessed water?

Ever had that feeling, but haven't been able to put a finger (or any other appendage, for that matter...) on why you feel that way?

You've come to the right place. We have that feeling at least three times a week, sometimes more when we go to certain web sites. (Which reminds us of the famous gibe about the patient who went to see his doctor. He's in the exam room, the doctor comes in; the patient exclaims: "Doctor, I'm so worried--whenever I raise my arm like this it hurts". The doctor waits a beat and replies: "Don't raise your arm like that.")

What we want to do is to display this increasing uniformity in a very simple way, one that doesn't get bogged down in "advanced metrics." We've decided to do that as follows:

--Take a group of hitters whose level of performance is very good, but not great. Players who usually the second and third best hitters on their teams.

--That range of performance, after some trial and error, was determined to be an on-base-plus-slugging (OPS) in the range of .850 to .860. Over the course of baseball history, that performance level has resulted in an OPS+ of 130--as noted, very good but not great. Players with a lifetime OPS+ of 130 are often in the Hall of Fame, but players who have an individual season of 130 where the range of OPS is .850 to .860 are  still longshots for induction: only 21% of the 224 seasons in the study were by Hall of Fame players.

--Criteria for inclusion in the study: 600+ plate appearances in the season with an OPS range of .850 to .860.

The first thing that an historical overlay of these seasons shows is that offensive levels have a significant effect on just where an .850-.860 OPS ranks in terms of league-relative achievement. You could draw a serpentine line through the graph that would show the ebb and flow of run scoring levels. The highest OPS+ for a player with a OPS between .850-.860 occurred in 1906, when Nap Lajoie's .857 OPS translated into a 169 OPS+. The lowest such OPS+ can be seen at the bottom of the first offensive explosion: Gus Suhr's 1930 OPS of .860 only translated into a 106 OPS+.

That's the effective range, though the scatter chart shows that the vast majority of these seasons fall between the 120-140 OPS+ middle ground (68% of the seasons, which is what those in the stat biz like to call a "steep bell curve"). The chart above also shows that while we are moving away from a high-offense era where the .850-.860 OPS range is no great shakes, we are far from the conditions where such  offensive production was significantly above average.

The shape of this performance is what should interest us, especially as it has changed over time. The chart below shows the aggregate performance of .850-.860 OPS hitters in twelve decades since 1900.

In that data, we don't see much movement in the OBP/SLG components, but we do see it in stats that help define the overall shape of a player's performance--isolated power (ISO) and what we call the "on-base ratio" (abbreviated here as OB%), which measures the percentage of OBP-BA that is part of the overall OBP.

As you can see, both these stats began to rise in the 1940s, as did the raw total of home runs and walks. The portion of hitting that derives from what classic neo-sabes like to call the "three true outcomes" (a phrase that has led to a series of "insights" that remain problematic in terms of an overall historical perspective on the game), represented as HR+BB+SO, has risen dramatically over time. HR+BB+SO constituted just under 14% of all plate appearances in the decade of the 1900s; in the first decade of the twenty-first century, that figure had more than doubled, to 30% of total plate appearances.

The only point in time where this progression reversed itself was during the 1970s, when isolated power declined among good-not-great hitters even as the overall ratio of walks/OBP showed an uptick, registering its highest overall percentage. These players produced a much higher aggregate OBP--highest since the 1920s, when the .850-.860 OPS hitters had their production driven by batting average.

So what the 1970s had was a kind of offensive balance that registered in its very good but not great players. How do we visualize this balance, however? The aggregate numbers seem mighty subtle, and probably don't get the idea across.

What we can do, however, is display the extreme edges of these hitters as arranged across the decades. The extreme edges are defined by the S/O ratio (the second-to-the-right column in the above table). What is that? It's the SLG divided by the OBP, and this is an average that hasn't fluctuated much.

But there is a range to be found in the individual players, and the diagram below captures the edges by segregating the .850-.860 OPS hitters into groups of high and low OBP. High is defined as .390 or higher; low is defined as lower than .360. The higher OBP guys have a lower SLG/OBP (S/O) ration, and show up in the clustering at the bottom of the chart (1.20 and below), while the lower OBP guys, with a higher S/O, cluster at the top (1.40 and above). The resulting groupings of these extreme edges show how the very good but not great players have been forced to conform to a lower-OBP, higher-SLG configuration:

The sample here represents a little over half of the players in the .850-.860 OPS range; the other half reside in the middle regions of the SLG/OBP ratio and would clutter the chart in a way that would not just be unseemly, it would render the lesson of the extremes to be downright unseeable.

For the first forty years of the twentieth century (1900-39), the great majority of .850-.860 OPS players achieved an OBP of .390 or higher--78%. Over the next forty years (1940-79), the two extremes achieved an almost perfect balance, with the high OBP still prevailing (52%). The trend toward lower SLG/OBP ratios (and higher OBPs) had its last gasp in the 1970s. Over the course of the last thirty years, lower OBPs and higher SLG/OBP ratios have come to dominate the game--70% of good-but-not-great hitters now reside at or above the 1.40+ SLG/OBP ratio, with that figure nearing 80% in the last decade. Good-to-great players with low SLG/OBP ratios (and high OBPs) are becoming extinct.

We hear about how "OBP has been assimilated into baseball's market evaluations." It would be more accurate to say that OBP has been neutralized, but not because GMs are seeking players who can actually draw walks. In fact, the opposite seems to be what's occuring.

Our chart shows that the trend started contemporaneously with Bill James's first decade as baseball's outsider iconoclast, and has only solidified itself despite the valorization of walks and OBP that's been the bedrock for sabermetric theory over most of the intervening years. Though the sample size is much too small with which to draw conclusions, the early returns from our current decade indicate that walks are continuing to decay as a feature of "good hitting" players, and that the uniformity of an approach built around a higher SLG/OBP ratio is continuing to gain ground.

That's sad news for baseball, because the game that emerges from such a trend is far less interesting and a good bit more predictable than what preceded it. Of course, the uncertainties that have been superimposed on baseball by its expanded post-season tend to mask this uniformity, but we have far less of a strategic range in how teams are/can be constructed to score runs now than was the case in the 1970s.

This uniformity is mirrored in the overall direction of neo- and post-neo studies, and in how those studies are defined. ("Value" is an 800-pound canary, while "shape" remains mostly a phantom presence.) While that work operates under a patina of objective science, it stops short of adjusting itself in terms of what it is attempting to study and thus is silent with respect to its truncated historical perspective. A question that current sabermetric study consistently avoids asking is whether the game as currently deployed is optimum. It's a loaded question, one filled with value judgments that cry out for a broader application of scientific techniques--but a mature science should prove capable of answering it. It is becoming more urgent for sabermetrics to at least address this issue, because a game where the SLG/OBP ratio is drifting ever higher is a game that has begun to calcify from within.