Saturday, June 9, 2012


Periodically we grouse about the decline of dimensionality in baseball. This sometimes leads us to create outlandish ideas for how to give the game a larger set of tools for increasing the ways in which runs can be scored. The sabermetric movement, with any and all of the prefix discriminators that we like to slide into the discussion, is--unwittingly, perhaps--an agent working in the opposite direction.

In fact, it may be hastening a kind of philosophical uniformity stemming from its own "empirical" imperatives.

There can be little doubt that the power game (as first manifested by Babe Ruth) transformed baseball. The oscillation in offense since the Babe cast his giant shadow over baseball has mostly devolved around an increasing institutionalization of the power game as the dominant strategy. It isn't coincidental that sabermetrics appeared at a point when home runs were on the decline (late 70s/early 80s). Within a decade, players with high secondary averages (one of the early measures developed by Bill James) were routinely being championed over players with high batting averages.

A decade later, a neo-sabe movement championed a "three true outcomes" construction that stopped just short of suggesting that isolated power (a measure of the number of extra bases over hits--defined separately by Jamesian alter-ego Eric Walker as "power average": total bases divided by hits) was the key to the offensive universe.

Extra bases on hits--and walks. The other "true outcome"--strikeouts--was decriminalized by a series of calculations about run expectancy that showed they were not a bad thing in terms of run scoring.

Bobby Abreu--an increasingly rare throwback
to the old-style "walkman"...
But there are some curious aftereffects from this philosophy--which has come to dominate the offensive structure of baseball over the past twenty years. The greater uniformity of these "active" offensive agents has, over the course of this time, marginalized the "passive" components in offense--singles and walks.

Particularly walks. There was a time when players did not have to be power hitters to draw walks, but such players have become increasingly rare. The lack of historical perspective in sabermetrics (even in the face of the sweeping overview presented by James in the two editions of his Historical Baseball Abstract) resulted in that type of player becoming increasingly marginalized as sabermetric ideas collided with the front office.

Today, that player is virtually extinct. Why is that? Because focusing on the three true outcomes will eventually leach walks from the outcome set, and the game will become increasingly two-dimensional.

Can we measure this? Of course we can. Can we determine when the emphasis on power and lack of contact began to overwhelm the third components? Sure. Can we trace the decline of the classic "walkman"--the player capable of drawing walks at a rate 50% above the league average despite possessing power that is league average or less? Absolutely. Let's take a look:

This chart is a slight recasting of what you'd see if you graphed walks/game over baseball history. One caveat to the title of this graph: this is percentage of PAs for players who walked 13% or more and who had at least 300 PAs in each individual season. We also didn't smooth this chart (no three-year averaging or anything).

What you can see is the Great Walk Spike in the late 1940s, a trend that had first begun in the 1930s when sluggers poured into the game but had been interrupted by World War II. After the percentages had leveled off in the late 50s/early 60s, walking was dealt a serious blow when the strike zone was expanded in 1963--you can see that the percent of PAs from high-walk hitters was instantly cut in half.

A sawtooth pattern then ensured over the next twenty years, as baseball seemed to be settling into a percentage that roughly mirrored its historical average (9%). As power surged in the late 90s, walks followed suit--but only briefly. They would regress in the 00s, and would dip to a level not seen since the 60s in 2011.

This chart doesn't break things out by type of hitter, however. There's more to the patterns here, and we can visualize them more comprehensively by separating hitters into above-average and below-average ISO groups:

Here we can see the overall frequency of high-walk hitters in terms of ISO levels. You can immediately see that the Great Walk Spike in the late 40s was a joint effort--a confluence of walk-taking from both types of hitters. And you can see how the low-power walkman took a dramatic nosedive over the decade of the fifties, stabilized briefly in the years of expansion, but dropped to its lowest level to that point in history in the mid-sixties.

And then you can see how the low-ISO walkman re-established a fairly stable value over the next twenty-five years, only to drop to all time lows during the last decade.

High-ISO walkmen--hitters whose threat of the long-ball forced pitchers to treat them with greater care--took over in the fifties, and the frequency relationship between the two groups remained reasonably similar until the mid-90s, when high-ISO walkmen PA percentages spiked to all-time high levels.

But guess what...the walk spike due to power hitting has proven to be a transitory phenomenon. A countertrend has emerged; pitchers seem to have pushed past the knee-jerk need to issue walks to everyone capable of hitting 20+ HRs in a season.

As a result, however, guys who can get on base without anything more than the occasional threat of the long ball (chicks or no chicks...) have, unlike Abraham Lincoln's vision of America as expressed in the Gettysburg Address, perished from the face of the earth.

The neo-sabermetric philosophical decision to lump walks into the "three true outcomes" hegemony has created a set of statistics that encourage a false context for examining offensive value. Walks are not really part of an all-or-nothing situation the way home runs and strikeouts are--they are not a totality unto themselves, they are as partial an event as any ball in play. The insight that plate appearances that result in the ball remaining out of play are much more valuable is undercut by two facts: a) as noted, walks are a special, partial class of "true outcome" and b) the continuing increase in strikeouts is happening as much at the expense of walks as hits.

Walks reached a 23-year low in 2011--with just under 3.1 per game, they hadn't been as low as that since 1988--a year in which there was a strike zone change after a record year for homers. That is the result of a counter-trend that developed in response to a set of offensive conditions that increasingly concentrated offense in isolated power. The irony for sabermetrics is that, in 2012, the "market inefficiency" in baseball is walks--one of the original rallying points for the movement.

Out of the broadcast booth and back into the
lineup for ya, Joe Morgan!!!
Rob Neyer (not always our whipping boy here, his occasional whelps to the contrary) touched upon this point about six weeks ago: "...what makes baseball so interesting is the variety of things that might happen." True enough, and we still have a great deal of that: we just don't have as much of it as we used to. We need to find ways to add more triples, more walks, less strikeouts--and we shouldn't be afraid to implement some odd-looking rules to see what they do. If anything, sabermetricians should be advocating these things--and many of them would be, if they hadn't become part of a consultancy culture.

We need to get back to a game where Rod Carew and Joe Morgan can exist again--players with unique combinations of skills and strengths that literally cannot exist in the game as we know it today.

[EDIT: In case you wonder what the breakouts for high-walk players look like thus far in 2012...the overall percentage is 8.3%, up from 5.5% in 2011 but still below baseball's all-time average. Currently low-ISO walkmen are up a good bit (2.5%), but several of these players won't receive 300 PAs over the course of 2012, so this figure will regress somewhat by season's end. High-ISO walkmen have also rebounded somewhat, accounting for nearly 6% of plate appearances, up from just 4.7% the previous year. A back-of-the-envelope projection is that walkmen will wind up between 7.0-7.5% of all PAs in 2012, which would still be around 20% below the all-time average.]