Over the next two weeks, I’m going to tackle another myth: batter/pitcher matchups. It’s common practice for daily league players and even television analysts to talk about a particular hitter’s career stats against the pitcher he’s facing. It is implied that these stats tell us something about how the batter is going to do in his present matchup with the pitcher, but it’s my contention that they tell us close to nothing. After all, we have thousands of at-bats to tell us how good a batter is that a few against a certain pitcher shouldn’t tell us nearly as much as the batter’s entire body of work. Today, I’m going to run a study to see if this is true.
For my study, I’m going to be looking at all single day batter/pitcher matchups since 2000 where the batter faced the pitcher at least three times. For every matchup, I’m going to first compare the batter’s career numbers against the pitcher to how he performed against the pitcher that day. Then, I’m going to compare his preseason Marcel projection (thanks as always to Jeff Sackmann for his historical Marcels database) to how he performed against the pitcher that day. This will tell us which is better to use – the batter’s career against the pitcher or a simple projection, ignoring the pitcher entirely. When comparing these sets of numbers, I’ll be using average error to study the differences.
I’ll further break things down based upon how many plate appearances the batter has had against the pitcher in his career. After all, while few will argue that going 1-for-3 lifetime against a pitcher is meaningful, some will be satisfied with 4-for-12. Others will still say that’s too small a sample but that 7-for-21 is sufficient. Where we draw the line is a subjective distinction that everyone has their own opinion about, so I’ll look at every possible sample of plate appearances to see how many are needed, if that point ever comes.
Here are the results for hitters facing pitchers. I’ve looked at a hitter’s typical triple-slash line (AVG/OBP/SLG), his wOBA (which is perhaps the best measures of overall offensive value since it weights everything properly), and his FanDuel points per 4 plate appearances (with runs scored removed from the equation).
In the table below, the first column gives the number of career plate appearances the hitter has had against the pitcher in his career. The second column gives the number of batter/pitcher matchups I’m drawing from.
The next five columns give us the difference between the average error for the projection and the average error for the career vs. pitcher. A green highlight means that the projection is better while a red highlight means that the career vs. pitcher data is better.
To give a very simplified example of what the numbers mean, let’s say we’re looking at batters that have between 1 and 5 career plate appearances against a pitcher. If a batter actually hits for a .250 batting average against a pitcher on a given day, the projection might say that he should hit .270 while the career data might say he should hit .336 (a difference of 0.066). Both will be wrong to an extent – after all, we can never guess a player’s performance perfectly – but the career vs. pitcher data will be wrong by 0.066 more than the projection will (on average).
As we can see here, the projection is almost always better to use than (or is at least as good as) the career data.
For batting average, it’s very obviously better up until the batter has 30 career plate appearances against a pitcher, then it’s just a little better up until he has 70 career PA. Of course, that’s also around the time our sample size really drops off. Using this small sample, though, there appears to be little difference between the projection and the matchup data from that point on.
For OBP, we see an even greater reason to use the projection. It’s not until the batter has over 70 plate appearances that the matchup data is even slightly better than the projection, but then it swings back in favor of the projection until we go over 100 PA, where our sample size is tiny.
For Slugging Percentage and wOBA, we see that the projection is significantly better up until around 70 career PA. After that it becomes a toss-up, but again, we’re in small sample size territory.
FanDuel Points shows the most support for the matchup data. The projection is much better for the players without much of a history against a pitcher, but once the batter has faced a pitcher more than 30 times, it basically becomes a toss-up the rest of the way.
Overall, it appears that the projection is far superior (or the very least, no worse) than a hitter’s career versus a pitcher, no matter how many times the batter has faced the pitcher. This is especially true when we consider a few things:
1. I’m using the Marcel system, which is the simplest of all legitimate projection systems.
2. The projections are from before the season. Once we actually get into the season, that projection is going to improve, especially when we’re looking at matchups that occur in, say, August.
3. The career versus pitcher data contains at-bats from the current season; the projection ignores the current season entirely.
Given all this, I think it seems clear that a quality in-season projection and thorough analysis will be far superior than using a batter’s career versus a particular pitcher.
While we may think that a hitter’s .400 batting average in 60 plate appearances against a pitcher is meaningful, it pales in comparison to the near-2,000 plate appearances that have likely gone into his projection, even if those 2,000 PA were against different pitchers. The differences in sample size are simply too large for the matchup data to overcome.
While it’s likely that some batters do hold some special qualities that allow them to hit well against a certain pitcher or simply have a particular pitcher figured out, it’s impossible for us to identify these batters by merely looking at the raw data.
There are, however, certain batter/pitcher matchup considerations that we can give credence to, such as platoon splits (though we need to be careful to interpret these correctly – a topic for another day). It’s also possible that certain batters perform better against certain classes of pitchers – a complicated subject, but perhaps this would make for an interesting article in the future as well.
That wraps it up for this week. Next time, I’ll look at the pitcher side of the equation and see if a pitcher’s career against a particular batter tells us anything about how he will perform when he faces the batter again. As always, if you have any questions, feel free to comment or e-mail me. Also, be sure to add me as a friend on Facebook and follow me on Twitter.
Derek Carty’s work can also be found at Baseball Prospectus, CardRunners Fantasy Baseball, and DerekCarty.com. He has previously had his work published by The Hardball Times, NBC’s Rotoworld, Sports Illustrated, FOX Sports, and USA Today. He is the youngest champion in the history of LABR, the longest-running experts league in existence, and is a graduate of the MLB Scouting Bureau’s Scout Development Program (aka Scout School). He welcomes questions via e-mail, Facebook, or Twitter.