## Archive for the ‘**Roy Halladay**’ Category

## The Importance of the First-Pitch Strike

Upon reading about Roy Halladay’s struggles against some Blue Jays’ minor leaguers, I decided that I should re-consider how important throwing first-pitch strikes is for pitching successfully. (Halladay threw only eight first-pitch to strikes to the 18 batters he faced; he retired only 7 of those batters.) So I am reprising this old piece from 2011 (written in 2010).

If you’ve read this before, go to the bottom where the last several paragraphs contain new links to other writing about first-pitch strikes, writing by people much more competent than I.

I decided to find out how important getting ahead of hitters actually is, so I did a study of first pitch strikes compared to first pitch balls.

Bill Feber*‘s The Book on the Book *got me thinking about this. Felber did a 5000 pitcher/batter interaction study of results in various pitch counts. Thomas Boswell of *The Washington Post* cited Felber’s study in an article about watching baseball while paying more attention to the count, since the count can tell you a lot of things. Boswell focuses on 1 ball 1 strike counts and what results, since 1-2 is a lot different for both pitchers and batters than is 2-1.

However, my focus is just on what happens after that first pitch. That is, what are the results after a 1-0 count compared to the results after an 0-1 count. In short, how important is the First Pitch Strike?

Now, I started to do this study, the really, really hard way. I looked at MLB Gameday for every individual game and began recording results in Microsoft Excel. One day’s games did me in. Too much tedium. Instead, on a lark, I visited Baseball Reference.com and I found this: tabulated data for every pitch count and what happens after reaching a particular pitch count. By parsing this data I was able to find out how results differed between 1-0 counts and 0-1 counts.

First of all, the overall data for 2010 revealed that hitters go .262/.333/.406 with a wOBA of .346. Walks result in 9% of plate appearances, and 18.27% of plate appearances ended with strikeouts, which gives us a walk to strikeout ratio (BB/K) of 0.49.

In plate appearances that are determined by one (the first) pitch, batters did really well: .345/.350/.555 (AVG/OBP/SLG, also called the “slash line“). (Note also that the OBP is higher than the AVG because some batters manage to get hit by pitches) Batters have a wOBA of .413 and an OPS of .905 in these plate appearances. Obviously there is no BB/K ratio since a batter can neither walk nor strike out in one pitch.

After 1-0 counts, first pitch balls, hitters produced the following statistical line: .273/.391/.436, a wOBA of .389, and an OPS of .827. The walk rate was 15.91% and the strikeout rate was 14.47%, producing a BB/K ratio of 1.10.

After 0-1 counts, first pitch strikes, hitters produced at the following rates: .228/.272/.348, with a collective (and woeful) wOBA of .294, with a similarly woeful OPS of .620. The walk rate for these batters was 4.92% and their strikeout rate was 25.67%. The BB/K ratio for these batters was 0.19 (yes, 0.19).

Now, let us assume that every time a plate appearance resolves on the first pitch that the pitch is a strike. This is an obviously untrue assumption since (a) HBP’s don’t happen on strikes, and (b) hitters such as Pablo Sandoval swing at first pitches that are not in the strike zone. But let’s make that assumption for the moment; after all, if the plate appearance resolves in one pitch, the batter almost always has swung at the pitch, and, presumably, batters attempt to swing only at pitches in the strike zone, that is, pitches which are strikes.

If we combine the data from one-pitch plate appearances, which are very good for hitters, with the data from plate appearances that resolve after 0-1 counts (first pitch strikes), we *still see *that hitters performed less well than they did following 1-0 counts. The combined data from one-pitch plate appearances and those following 0-1 counts indicates a collective hitter slash line of .250/.286/.387, with a wOBA of .315 and an OPS of .673. Look at the OBP! .286? That is terrible, even for a backup utility infielder; a .270 OBP constitutes the “Mendoza line” of OBP or it should, at any rate. By contrast, following 1-0 counts, hitters’ OBP figure is .391. Looking further at the first pitch strike data, we find a walk rate of 4.02% and a strikeout rate is 21.02%, which creates a BB/K rate of 0.19.

2010 data is as follows:

Total Data shows us 63,516 AB, 16,458 Hits, .259 AVG/.330 OBP/.406 SLG, .346 wOBA, .736 OPS, 9.00% BB rate, 18.27% K rate, 0.49 BB/K.

First Pitch Resolution comes us great for hitters with 7283 AB, 2515 Hits, .345 AVG/.350 OBP/.555 SLG, .413 wOBA, .905 OPS, and nonexistent BB and K rates and a nonexistent BB/K ratio.

First Pitch Balls create the following results: 24.637 AB, 6,735 Hits, .273 AVG/.381 OBP/.436 SLG, .389 wOBA, .827 OPS, with 15.91% BB rate and 14.47 K rate, for a BB/K ratio of 1.10.

First Pitch Strikes are followed by these results: 31,596 AB, 7208 Hits, .228 AVG/.272 OBP/.348 SLG, .294 wOBA, .620 OPS, with a 4.92% BB rate and 25.67 K rate, for a BB/K ratio of 0.19.

Combining the results for First Pitch Resolution with that of First Pitch Strikes gives us the following numbers: 38,879 AB, 9123 Hits, .250 AVG/.286 OBP/.387 SLG, .315 wOBA, .673 OPS, a walk rate of 4.02% and a K rate of 21.01%, for a BB/K ratio of 0.19 (which is a K/BB ratio of 5+!).

Thus *even if *your pitchers are giving up all that good stuff to hitters on plate appearances resolved on one pitch, **even then**, it appears that first pitch strikes are simply awesome for the pitching team. The on-base percentage differential is huge (.286 vs. .391 when the count goes 1-0) and attributable to the incredibly lower proportion of walks that your pitching allows.

2010 data is no aberration. After accumulating the information back through the 2006 season, it proves remarkably consistent.

In plate appearances resolved on the first pitch: .341/.346/.552, .409 wOBA, .898 OPS.

In plate appearances begun with a First Pitch Ball (resolved after a 1-0 count): .279/.394/.456, .396 wOBA, .805 OPS, 4.82% BB, 13.88% K, 1.12 BB/K.

In plate appearances begun with a First Pitch Strike (resolved after a 0-1 count): .236/.279/.361, .301 wOBA, .637 OPS, 4.82% BB, 25.05% K, 0.19 BB/K.

In plate appearances either resolved in one pitch or begun with a First Pitch Strike: .257/.292/.399, .332 wOBA, .684 OPS, 3.88% BB, 20.17 % K, 0.19 BB/K.

The walk to strikeout ratio (BB/K) is the most, er, striking thing. After 1-0 counts, batters walk more often than they strikeout, while after 0-1 counts batters walk less than a fifth of the time they strikeout. That is a huge, huge difference. (Maybe it’s not the actual bases on balls that give managers their gray hairs; maybe it’s the first pitch balls that given them those gray hairs, as they spend the rest of the plate appearance envisioning the walks and other bad things that will follow.)

The conclusion is simply put: **It pays to throw First Pitch Strikes.**

Even though are going to have some guys get on base or hit bombs because they are first pitch swinging, the on-base percentage difference between plate appearances starting with first pitch balls (.394) and those that don’t (.292) is simply huge. This represents a lot of runs saved in plate appearances that don’t start with balls.

In fact, if you look at the summarized data from 2006 to 2010, you will see that the **on-base percentage** in First Pitch Strike plate appearances (.279) is about the same as the **batting average** in First Pitch Ball plate appearances, and the OBP in First Pitch Ball plate appearances is a “robust” .394, which is All-Star-quality good (or bad, from the pitcher’s perspective). Runners simply do not get on base much after a First Pitch Strike.

An additional thing to consider is that many plate appearances that are resolved after one pitch are swings that occur when the batter thinks he’s gotten “his pitch,” that is when the pitch is in the location where the batter is looking for a pitch to hit. This means that those first pitch swings resolve plate appearances often happen when the batter thinks he has the best chance of doing something with the pitch. I can’t quantify how often this happens, but I *can *point out that batters are said to approach the first pitch in such a way.

Results obviously vary from batter to batter. After all, some guys don’t mind hitting with two strikes (Joey Votto, we are looking at you). Others mess up in 3-1 situations. But, *as a general rule*, the First Pitch Strike seems to be the best pitch to throw.

On the other hand, James Shields’ 2010 provides perhaps the most obvious counterexample/counterargument to what I’ve said, but still…the numbers are pretty conclusive.

Are you interested in more reading about first-pitch strikes? If so, here are links to some of the more competent writing about the topic.

First-pitch strikes are so important, that *FanGraphs *tracks it. You can find the MLB leaderboard here. Note that the MLB average is ~59%. (I should do a study with an IP minimum, establishing the standard deviation and then listing pitchers from top to bottom so we can see where they lie along the spectrum. Included should be K%, BB%, ERA, FIP, xFIP. Then we could test linkages between F-Strike% and other peripherals and see if we can predict peripherals on the basis of first-pitch results. That’s what I *should* do, but I am terrible at following through.)

Considered from a coaching standpoint by Jack Dunn, a youth coach. So a general theoretical discussion of inducing more pitchers to throw first-pitch strikes.

Brian Oakchunas of *Baseball Prospectus* makes an extremely data-driven argument, as always, and they conclude that extreme strike-throwers are more successful pitchers than others, so you might as well throw strikes. At a minimum, first-pitch strikes move the “peripherals” (K/9, BB/9, ERA) in the “right” direction (higher, lower, lower).

Additional coaching considerations on first-pitch strikes motivate Phil Tognetti to consider the mindsets of both hitters and pitchers, which, when combined with data, leads to the conclusion that pitchers should generally work to get ahead of the hitter on the first pitch.

When Craig Burley, writing for *The Hardball Times*, did his study in 2004, he found that *fewer than 8% of first-pitch strikes turn into base hits*. Once again, then, it pays to throw them. Also, even if the spread between the slugging percentage and batting average on the first-pitch strikes put into play is higher than the spread between them on other pitches put into play, they happen at a low enough rate, and, remember, generate a sufficiently low on-base percentage, to make throwing them worth it.

Peter Ellwood uses OPS to make his point, rather than the more elegant, comprehensive, and accurate wOBA, but he makes exactly the same point I made above (which makes me happy…confirmation-bias considerations aside).

The Rockies pitching rotation sucked in 2012, and while there are other factors explaining that, it is no coincidence its’ members also threw the lowest percentage of first-pitch strikes in the major leagues.

Remember that study I said I should do to determine the relationship between first-pitch strikes and other peripheral measures of pitching performance? Well, writer natstats over at *Federal Baseball *did a version of that study and provided a nice visual aid.

Rob Biertempfel of the *Pittsburgh Tribune-Review*, discussed the problem in 2011 in terms of the Pirates pitching staff. Note that Charlie Morton got off to a great start in 2011. Note, however, he isn’t great, and thus not the poster boy for my proposition I wish he could be. (Confession: We here at Twisting Blade Productions cheer for the Pirates, though we also like the Twins, are developing a fondness for the Mariners, have a crush on the Rays out of sheer respect for their organizational competence, and are intrigued by the Diamondbacks’ front-office maneuvers of this last off-season.)

*The Flagrant Fan* (that’s a good blog title) nicely provides data from 2001 through 2012. It illustrates a negative correlation between F-Strike% and OPS, which is expected. (Of course, correlation is not causation, through the underlying mechanics explained by game theory–the mindsets of pitcher and batter–demonstrate an at least theoretical structural linkage; if we could quantify the degree to which batter and pitcher mindset factor into the outcome we would be better off on establishing causation.)