Maybe this happens to you. You're enjoying the ebb and flow of a baseball game on a summer afternoon. In the home third, Joey Grind takes two fastballs for strikes. Then he takes three sliders in the dirt, fouls off a couple of fastballs, and finally walks on the eighth pitch. As sure as night follows day, the announcers go crazy congratulating Grind on a great at bat. A little later, the same pitcher starts Mickey Hack with the same first-pitch fastball that he threw to Grind. Except Hack isn’t taking. Instead he pounds it 410 feet for a game-changing home run. Between sips of your beverage you're thinking: ``Wow, what a great at bat.'' But you will never hear those words in reaction to Hack's heroic deed. An industry insider, whom I deem reliable, recently confided that the Broadcaster’s Handbook stipulates that an at bat can only be described as great if it includes at least six pitches and two foul balls.
Given the modern obsession with pitch count, the idea of making a pitcher work seems admirable. After all, once the starting pitcher departs you get to feast on your opponent’s soft middle relief corp. In 2012, ML teams played only .450 baseball when their starter left with at least 100 pitches before recording an out in the 7th. Add in the twenty-first century realization of the importance of on-base percentage and the take-and-grind approach seems to make even more sense. The numbers support this point as well. Major league plate appearances that lasted at least six pitches last year yielded an OBP of .370 which was well above the ML average of .319. On top of all that, your local announcer will happily weigh in with the opinion, perhaps drawn from the aforementioned Handbook, that long at bats favor the batter because the more pitches a guy sees the more likely he is to square one up.
How does this relate to the American League Champion Detroit Tigers? Last year Tiger batters saw the fewest pitches per plate appearance in the American League at 3.74. On the other hand, Tiger pitchers threw the second most pitches per plate appearance in the AL at 3.90. This would seem to give opponent batters more opportunities for great at bats than Tiger batters. But Detroit made it to the World Series. Are we missing something?
Perhaps we are. Major league batters hit .330 and slugged .531 when they put one of the first two pitches in play last year. This is pretty good, of course, and the only hitter who exceeded a .330 batting average for the full season won the NL MVP. For pitches beyond the second, however, the possibility of the strikeout puts a significant damper on batting performance. And strikeouts are on the rise. For the 5th consecutive season, MLB set a new record for strikeout rate in 2012 with a mark of .198 K/PA which obliterated the .186 record set in 2011. This increasing tendency of batters to strike out raises the incentive for hitters to make something happen early in the count. If they don’t, they suffer the consequences. Major leaguers hit .260 and slugged .411 on pitch 3 and we see another big drop to .213 and .333 on pitch 4. Performance even gets a little worse for pitch 5 at .204 and .327. Maybe there is an argument to be made for swinging early.
One job of a hitting coach is to determine the appropriate level of aggressiveness for an offense on a given day. Each offense has its own personality and, in addition, a well-formulated game plan must account for a number of external variables. If you’re going up against Ubaldo Jimenez it will typically pay to be patient while against Cliff Lee trying to grind your way to walks is usually a losing proposition. If the other team has a weak bullpen you may want to try driving up the starter’s pitches. If the opposing starter has a wipeout slider, then you don’t want to risk falling behind in the count. There’s also a gamesmanship element. If you expect the opposing starter to attack the zone, then you want to come out swinging. If he expects you to come out swinging, then he’ll work more off the plate. There’s a reason they call it a game of adjustments.
We've seen that grinding and hacking each have their benefits. But what’s the best approach? It might seem like there’s too many variables in play to evaluate the tradeoffs. But let’s try something simple. The goal is to win games. To look for general trends, therefore, we can ask how a team’s win rate varies with the number of pitches that the opposing starter throws per batter. The figure below plots this data for two seasons. I selected last year as well as 2005 when the current streak of eight consecutive seasons of increasing K-rate began. In particular, the 2005 strikeout rate was .165 as compared to the record-breaking .198 in 2012. For reference, the league average for P/PA was 3.73 in 2005 and 3.82 in 2012.
The red curve in the plot is the cumulative winning percentage for ML teams in 2012 as a function of opposing starter pitches per batter faced. For example, teams won at a .557 rate in games last year when the opposing starter threw fewer than 3.5 pitches per plate appearance. Note that the curve is monotonically decreasing which indicates that teams do better, on average, when their batters see fewer pitches per PA against the opposing starter. The black curve is the corresponding data for 2005 and shows the same decreasing trend. By comparing the two curves we also see that, as expected, completing a plate appearance with a small number of pitches leads to a higher win rate in a higher K-rate environment like we had in 2012. In summary, teams do better when they see fewer pitches per plate appearance and the benefit becomes more significant as the league strikeout rate goes up.
Now back to the Tigers. By now, you should be willing to accept that short plate appearances on offense are not necessarily a bad thing. And this was more true for Tiger batters in 2012 than for the league at large. We can generate the curves plotted above for individual teams, but they lose their monotonicity and smoothness due to sample size effects. Instead, we'll split the curve at a location of interest and compare the Tigers with the rest of the league. As we pointed out earlier, Detroit batters saw the fewest pitches per plate appearance in the American League last year. It's not surprising, therefore, that the Tigers saw fewer than 3.5 P/PA versus the opponent starter in a fraction of their games (29.0 percent) that was significantly larger than the overall ML average (22.1 percent). The Tigers embraced these opportunities by going 31-16 (.660) which was significantly better than the overall ML average win rate of .557 for these low P/PA games. In their remaining 115 games (P/PA >= 3.5), the Tigers went just 57-58 which gives a slightly better win percentage than the ML average of .484 for these games. Thus, the division was won in the small fraction of games where Tiger batters saw the fewest pitches per plate appearance against the opponent starter.
What about Tiger pitchers who threw the second most pitches per plate appearance in the AL in 2012? For pitchers, the advantage occurs for games on the right end of the curve with high values for starter P/PA. In particular, Tiger starters threw at least 4.25 P/PA in 19.8 percent of their games last year which was well above the ML average of 14.8 percent. Just like the batters, the pitchers made the most of their opportunities. Detroit went 21-11 (.656) in these games (P/PA >= 4.25) which was much better than the ML average winning percentage of .526. In their other 130 games (P/PA < 4.25) the Tigers went 67-63 (.515) which was somewhat better than the overall ML winning percentage of .495.
So what's the message? Should the Tigers try to reacquire Delmon and install Boesch as the everyday left-fielder to reduce team pitches per plate appearance? Unfortunately, it's not quite that simple. Anyone reading this who can swing a piece of wood can enjoy a P/PA of 3 or less in the big leagues. But maybe, with the season approaching, we can be a little more patient when a Tiger flies out on the first pitch or enters the 4th with a high pitch count. After all, great at bats aren't as great as people think.