Last week, in an article in Sports Illustrated, Tom Verducci put forth an argument that the modern game of baseball has a problem. Hitters, he claimed, have become too passive in their approach at the plate as they attempt to drive up the pitch counts of the opposing pitcher. He mixes together a couple of case examples (Joey Votto, Jayson Werth) with some data that appear to show that hitters have become more passive in their approach over time, and are paying for it in declining run production. Maybe Joey and Jayson, and by proxy the rest of the baseball players out there, should swing the bat a little more.
Mr. Verducci's argument was in part aesthetic, and it's wise advice to remember de gustibus non est disputandum (matters of taste need not be argued). According to Verducci, “What we are left with is a sport in which games keep getting longer but with less and less action… The knottiest issue for baseball is not the stadium issues of Oakland and Tampa Bay or the Biogenesis scandal; it's the increased lack of action in your average baseball game." Of course, one man's snooze-fest is another man's thrilling chess match, so your mileage may vary.
But Mr. Verducci also made several claims about what he saw as the consequences of this development of a "passive" approach, and to his credit, presented data to back them up. His conclusions about the state of the game are interesting. Let's take a closer look at them, shall we?
Are hitters really getting more passive?
Mr. Verducci is correct that the average plate appearance has gotten longer over the years. Retrosheet (put them in the Hall of Fame) has data back to 1993 on pitches and outcomes, and as the graph below shows, there has been a general upward (and significant) trend in pitches per plate appearance. (Note: for all analyses, I'm excluding intentional walks and pitchers batting, as well as plate appearances for which pitch sequence data are not available.)
In 1993, the average plate appearance lasted 3.64 pitches. In 2012, it was 3.79. The difference may not seem like much, but over three trips through the lineup for a starter (27 hitters), it's a difference of four pitches, which represents four percent of the usual 100 pitch limit. If teams really are trying to drive up pitch counts, it's working.
As might be expected, there has been a small decline in the number of outs recorded during the average start. In 1993, starters recorded an average of 18.33 outs. In 2012, the figure was down to 17.67, although the decline was hardly a straight line. If teams now have a goal to get into the other team's bullpen early in the game, then they have been somewhat successful. By about 2/3 of an out over the past 20 years.
Where are those extra pitches coming from? Mr. Verducci seems to believe that they are coming from players being less willing to swing. That's easy enough to measure. Again, we have publicly available data from 1993-2012, and the league-wide swing rate is presented below. Over time, that number has bounced up and down a bit, but stayed almost exclusively between 45 and 46 percent. A one percent change in a rate in baseball is not trivial, but it’s best not to get carried away about it. This is not a fundamental shift in the way the game is played.
But where Mr. Verducci does get it right is that swinging is becoming much less common on hitter’s counts like 3-0. Mr. Verducci, using a case example of a Jayson Werth hitting into a double play on 3-0, wonders whether the greater sin was that Werth hit into a double play (in a two-run game in the top of the eighth) or that he swung on a 3-0 pitch. The graph shows 3-0 swing rates over the last 20 years. Swing rate peaked at a little over 13 percent in 1996 and had been nearly halved at 7.1 percent by 2009. While there has been a bit of a rebound since, there has been movement in the rates at which hitters have offered on 3-0 over the entire two-decade period (although it should be noted that it was never a popular pitch on which to swing). The graph for 2-0 counts (not shown) looks much the same. Hitters really are swinging less in traditional hitter's counts.
So we have a mystery. If overall swing rates haven't changed much, what is going on? Let's first take a look at contact rate. I found that over the last 20 years, there has been some variation, but of late, contact rates (per swing) have fallen. Hitters are actually missing more when they swing. Take a look:
On top of that, there's been another shift. Even when the batter makes contact, there's been an uptick in the number of foul balls. But not just any foul balls. I've shown in the past that not all foul balls are created equal, and that you need to look separately at foul balls that happened with zero or one strike in the count (where the foul ball counts as a strike) and two-strike foul balls. Below are the trends over the past 20 years on one graph (early count fouls are per PA; two-strike fouls are among only plate appearances where the count reached two strikes). Over time, hitters have slowly increased their early count fouls, while two-strike fouls have boomeranged. Because an early count foul is basically the same (although not exactly) as a swing-and-miss for purposes of the count, it means that the effective contact rate has been going down even more sharply over time.
I don't think that the issue is hitters being overly patient. At least, that's not the whole story. They are swinging at roughly the same rate that they always have. They're just missing (or fouling it off) more, particularly early in the count. In my previous research, I've found that early count foul balls are a tell-tale sign of a hitter who is taking an approach that emphasizes power (more fly balls, more home runs), whereas a hitter who fouls off a lot of two-strike pitches prefers a more contact-based (more grounders, more singles) approach.
The evidence points toward a larger magnitude shift of taking bigger (but not more) swings with more misses. You get one, two, three strikes before you're out at the old ballgame, and batters now seem more comfortable using them all. Strikeout rates have gone up quite a bit over the last 20 years, and by a factor that's obvious to the naked eye. In 2012, the number of plate appearances that involved a second strike topped 50 percent for the first time in the 20 years of data that I had available.
Let's return to the issue of the 3-0 count, though. Because the batter has the advantage, he can afford to sit on one pitch in one location and if he gets it, take a big swing. If the pitcher misses the zone, the batter gets a walk, and even if it's a strike or he swings and misses, it's 3-1 and the pitcher is still in a hole. And yes, the pitcher had to use another of his 100 pitches for the day. In a 3-0 count, the cost-benefit analysis is fairly obvious to favor adopting a "swing real hard in case you hit it" mentality more often. It's one thing when Casey says "That ain't my style" on 0-1, but you could forgive him for it on 3-0.
Maybe what we're seeing is that the average MLB hitter is more often looking for "his" pitch, and on more counts than 3-0, at the cost of perhaps letting a pitch go by that was in a good spot for hitting if he'd been looking there. It's a high-risk, high-reward strategy from an individual plate appearance point of view that has a nice side effect of driving pitch counts up a little bit. The problem is that if hitters are trying to be more selective and in doing so get higher rewards, they've not been doing a good job. Slugging percentage on balls in play that were hit on 3-0 counts has basically gone up and down over the years, and is probably just vibrating around the mean. Even looking at slugging percentage on balls in play during plate appearances that at some point passed through 3-0 (not shown), we can see the same trend.
Is forcing extra pitches helpful in winning games?
If hitters aren't getting any extra bases for their patience, you could make the argument that teams are at least breaking even. And perhaps they derive some benefit from making the other guy throw a lot of pitches. However, Mr. Verducci suggests that there is no correlation between extra pitches per plate appearance and winning games, although his evidence on this matter is a little wobbly from a research methodology standpoint:
Last year there were 13 teams that ranked above average in most pitches per plate appearance. Nine of those 13 teams did not make the postseason. The two pennant winners, San Francisco and Detroit, ranked 25th and 27th in pitches per plate appearance.
Let's clean that up.
In 2012, the correlation between pitches per plate appearance, at the team level, and team winning percentage was .14. In 2011, it was .07. In 2010, it was .02. For all team-seasons from 1993-2012, it was .14. Mr. Verducci is correct. High pitch counts are neither a harbinger of success, nor of failure at the macro level. It's not the length of the at-bat that matters. It's what you do with it.
This actually cuts against sabermetric orthodoxy at the micro level. One common-sense reason to try to drive up a starter's pitch count is that because teams generally pull their starters after 100 pitches, but are loathe to pitch their good relievers more than an inning at a time, evicting the starter after five innings means that the other team will have to put in a few of their less-than-stellar relievers to cover the extra innings. In fact, a study done by BP's Colin Wyers shows that the earlier a starter exits, the higher the bullpen ERA is for the rest of that game. Teams can try to get at the soft underbelly of the other team's bullpen if they force the starter out early. Assuming that the underbelly remains soft.
Mr. Verducci essentially makes the argument that teams have responded in recent years by fortifying their bullpens. It's hard to tell whether he's right on this one or not, but as he points out, offense is down over the last few years. Maybe this is a case where teams have seen a long-term trend and made a sensible (and successful?) counter-move. If starters are on a pitch count, and relievers are allowed to go only one inning, then it leaves open the "work the starter's pitch count" option. To cover that, teams may have put more resources into developing relievers, and now it's starting to bear fruit. The game of baseball is subject to evolution, just like everything else. Mr. Verducci's main argument seems to be that the "grind out an at-bat" strategy is not silly, just outdated. It would need more investigation (in a different article), because it's hard to tell whether, for example, strikeout rates are going up because the pitchers are better at striking hitters out or the hitters are worse at not striking out. But he might be right.
Are we evolving too?
I've had my disagreements with Mr. Verducci before, but in this case, he indirectly brings up an important point. Any strategy that teams employ, whether intentional or not, can have a counter-move that renders it useless. And so, when the sabermetric movement observes some inefficiency or strategic edge and then it becomes widely known, there will be a race to develop some response to it. Ten years ago, it was new-ish information that OBP was undervalued relative to batting average and that teams should look for high-OBP, low-average guys. Today, you don't see people making that argument. It's common knowledge, and the ecosystem has adjusted. Perhaps sabermetricians (and teams?) are guilty of thinking that the strategic edges that they have found are grand truths. Far from it. Baseball is a game of move and counter-move, both within the game and at the broader level.
Yes, I could nitpick Mr. Verducci's methodology to death, and that would be wonderfully entertaining, I'm sure. But I think that to do just that misses the chance at a bigger learning opportunity. It's not that our theories are wrong in the sense that when we came up with them, they didn’t accurately reflect the state of the game. It's that the game may have changed under our feet, and we might still be peddling strategy fit for a different set of assumptions. Mr. Verducci's piece has a lot of value in that regard. It's a reminder that the game can evolve, maybe faster than we would like it to, and we have to evolve with it.