Are Baseball’s Fundamentals Changing?

It’s easy to see that baseball has changed over the last couple of decades. Walks, strikeouts, homers, and stolen bases have all seen their ups and downs, and we’re currently experiencing a valley in terms of offense. Games are longer. There’s instant replay.

But there’s evidence that players are getting similarly better and worse at these things — the distribution hasn’t changed, the graph has just been shifted. It’s possible that the relative value of certain events in baseball as a whole could still be about the same. A stolen base’s relationship to a win could be unchanged if the distribution of stolen bases is similar, and there are just fewer of them.

Is that what you find when you look back empirically? If you relate strikeouts, walks, stolen bases, and home runs to winning, is that relationship steady over these turbulent times?

In order to try and begin answering this question, I broke baseball into four pieces since 1974. I took a glance at this annotated history of offense and tried to pick some obvious groupings, but I wanted them to be roughly similar in size and some randomness was alright. Then, I simply correlated team performance in some key categories to their winning percentage. The “plus” stats below are all indexed to league average since I was asking “Has performance relative to the league in certain factors of the game become more or less important over time?”

The table contains the correlations for each stat by r. All p values < .00001 unless otherwise mentioned in the writeup.

Strikeout Rate
Not once in this sample did indexed strikeout rate explain more than 8% of the variance in win percentage. It just doesn’t seem very important. The relationship bordered on statistically insignificant in three out of the four years, with p values ranging from .001 to .025. It’s tempting to say that a low strikeout rate is slightly more important these days, but since it’s only explained 5% of the variance in win percentage over the past seven seasons, it also is folly to put too much stock into this small, potentially insignificant change.

Walk Rate
Walk rate is a little different. It was even more important than power, relatively, during the ‘steroid era’ even. That seems to be the case despite evidence that higher run environments make walks less important by linear weights. On the other hand, there’s been a noticeable jump in the correlation between walks and wins since the seventies and it’s held even as the offensive environment has slowed recently. It’s hard to say that we should move on from walks, however. They remain fairly important.

Isolated Slugging Percentage
When power was at its most scarce in this study, it was the most important. Even when indexed against the league, it was very important to have power in the late seventies. Almost twice as important as it was in the two eras that followed, when there was a power explosion due to expansion, rule changes, and/or steroids. As we return to baseball that looks a bit like the 1970s, it looks like power has once again become more important, relatively. Where it fell behind walks in the early aughts, power is now once again the best single-stat predictor of wins.

Batting Average on Balls in Play
Though the stat has changed — from .277 in 1974 to .295 in 2014 — batting average on balls in play *seems* like a stable stat. I wasn’t expecting much of a correlation, and to be fair, the relationship in the last decade has been tiny (.0278 r2) and possibly insignificant (.016 p value). Maybe that’s because of this:

The fewer balls there are in play, the less those balls in play can impact the game. Seems relatively simple.

Baserunning Runs
Since the mid-nineties, baserunning runs have described less than 5% of the variance in win percentage. In our current era, a team built on patience and power would have a decent statistical head start on a team built on baserunning and contact. This ignores defense, which has been a factor in some team-level resurgences in the past decade, and of course we’re focusing on offense over pitching. But there’s little evidence that baserunning is in the midst of a renaissance in the game.

Weighted Runs Created Plus and Offensive Wins Above Replacement
These stats are component stats, built on the back of many of the stats we’ve already investigated. But it is instructive to see how stable they are, especially over the last twenty years. Since 1996, wRC+ has explained about 40% in the variance in wins, and offensive WAR has explained about 54%. If someone wants to tell you that baseball is all about pitching, this is an easy empirical finding that suggests, no, baseball is mostly still about offense. Even as offense seems to be harder to find with every passing year.