pitch classification

Note: This article was originally published at the Statistically Speaking blog at MVN.com on January 3, 2008.  Since the MVN.com site is defunct and its articles are no longer available on the web, I am re-publishing the article here.

Baltimore Orioles ace Erik Bedard had a breakout season in 2007 before being sidelined at the end of August with an injury to his oblique muscle. Prior to his final start, he was receiving strong Cy Young Award consideration with a 13-4 record, a 2.97 ERA, and a league-leading 218 strikeouts and only 135 hits and 52 walks allowed in 176 innings. (His last start, pitching injured, resulted in final season numbers a little worse than these I’ve listed.) Compare his 2007 numbers to his previous career-best season in 2006, when he finished 15-11, with a 3.76 ERA and 171 strikeouts against 196 hits and 69 walks in 194 1/3 innings, and it’s clear he stepped up his game in 2007. How did he do it?

Unfortunately we don’t have any PITCHf/x data from the 2006 regular season, but we can use the PITCHf/x microscope to take a look at what Bedard did in 2007 (in the 701 pitches for which we have detailed data). What does Erik Bedard throw? Let’s take a look at his repertoire by graphing the speed of his pitches versus the direction they break.

Bedard has four pitches, as best I can tell. His famous erstwhile pitching coach will tell you he throws four different types of fastballs, but I can only see a four-seamer and a cutter. Either he throws the sinker and the “comebacker” infrequently, if at all, in game action, or they move too similarly to the four-seamer for me to differentiate them using this data.

I mentioned already that Bedard pitched with an injured oblique muscle in his final start of the year on August 26. Most of the fastballs and cutters with a speed below 90 mph were recorded in that start. He averages just over 93 mph on his four-seam fastball and 92 mph on his cut fastball. In his August 26 start, those clocked at 89 and 88 mph, respectively.

When healthy, his four-seam fastball runs 92-95 mph and breaks away from a right-hander by about 7-11 inches. The four-seamer is one his two primary pitches to right-handed hitters; he throws it 34% of the time. Against lefties, it’s his third pitch, used only 23% of the time.

His cut fastball runs 90-94 mph and breaks away from a right-hander by about 2-6 inches. The cutter is his primary pitch to lefties, used almost half the time (45%); against righties, it’s his third pitch, used 24% of the time.

Bedard also throws an occasional 80-83 mph changeup, almost exclusively to lefties (7%). Probably his best pitch is a 76-80 mph curveball, which he uses equally to righties (35%) and to lefties (32%).

Let’s take a look at how all his pitches move, including the effect of gravity in addition to spin-induced deflection.

The slower pitches like the curveball and changeup drop more because gravity has longer to act on them.

I thought it might also be interesting to show something more in line with what I believe the hitter perceives as the “late break” on a pitch, the deflection of the pitch due to both spin and gravity in the last quarter-second before it crosses the plate. Thanks go to Tom Tango for this idea.

This seems to give a more realistic guess at how a hitter might perceive the drop on a curveball compared to a fastball.

Let’s take a look at how Erik Bedard mixes his pitches in different ball-strike counts.

We can see that he uses his cutter more often early in the count or when he falls behind and his four-seam fastball more often with two strikes. He uses his curveball equally across almost all counts, except for avoiding it on 3-0 and 3-1 and showing some preference for it on 2-1 and 3-2 counts. His changeup shows up mostly on 0-1 and 1-1 counts; he throws it to righthanders 20% of the time on those two counts and only 3% of the time on other counts.

Here’s a table showing the details by count.

Count Fastball Cutter Changeup Curveball Total
0-0 54 62 4 62 182
0-1 20 26 21 33 100
0-2 27 10 0 15 52
1-0 16 22 4 22 64
1-1 18 20 9 23 70
1-2 35 8 1 24 68
2-0 5 10 0 4 19
2-1 11 5 3 20 39
2-2 24 11 0 25 60
3-0 0 4 0 0 4
3-1 8 8 0 1 17
3-2 4 8 0 14 26
Ahead 82 44 22 72 220
Even 96 93 13 110 312
Behind 44 57 7 61 169
0 strikes 75 98 8 88 269
1 strike 57 59 33 77 226
2 strikes 90 37 1 78 206
Ball 0-1 170 148 39 179 536
Ball 2-3 52 46 3 64 165
Total 222 194 42 243 701

Now, let’s examine where in the zone Bedard throws his pitches and what results he gets with them.

LHH Ball CStrk Foul SStrk InPlay Avg BABIP SLG HR
Fastball 0.48 0.16 0.16 0.03 0.16 0.600 0.600 0.600 0.000
Cutter 0.26 0.30 0.20 0.11 0.13 0.250 0.000 1.000 0.250
Changeup 1.00
Curveball 0.36 0.09 0.20 0.16 0.18 0.250 0.250 0.625 0.000
RHH Ball CStrk Foul SStrk InPlay Avg BABIP SLG HR
Fastball 0.35 0.20 0.21 0.08 0.16 0.258 0.233 0.419 0.032
Cutter 0.34 0.26 0.24 0.02 0.15 0.300 0.263 0.500 0.050
Changeup 0.54 0.05 0.12 0.02 0.27 0.273 0.200 0.636 0.091
Curveball 0.34 0.18 0.16 0.21 0.11 0.227 0.227 0.318 0.000
Lg. Avg. Ball CStrk Foul SStrk InPlay Avg BABIP SLG HR
Fastball 0.36 0.19 0.19 0.06 0.19 0.330 0.304 0.521 0.037
Changeup 0.40 0.11 0.14 0.13 0.21 0.319 0.295 0.502 0.035
Curveball 0.40 0.19 0.13 0.11 0.16 0.310 0.290 0.471 0.029

The league average information comes from John Walsh’s article, and I’ve adapted his format in presenting this information. His pitch types probably don’t correspond exactly to mine since he lumps sinkers and cutters in with four-seam fastballs and splitters in with changeups. I believe it’s still helpful to use his league-wide information for comparison since I haven’t established a league-wide baseline on my own yet.

With the four-seam fastball, Bedard mostly works the outside part of the plate, especially to lefties but also to righties. To lefties, he mostly stays up or away with the fastball, out of the strike zone, and when he does get in the zone, he doesn’t have very good results, although the sample size is small. Against righties, he gets very good results with the fastball, holding them to a .233 BABIP and a .419 slugging percentage.

With the cut fastball, Bedard works away from lefties and gets a lot of called strikes and not much good contact, although two cutters in the middle of the zone did go for home runs. Against righties, he’s all around the zone with the cut fastball, and his results aren’t quite as outstanding. He gets a few more foul balls and a lot less swinging strikes, but overall his results with the cutter are still pretty good against righties.

Erik Bedard threw one changeup to a lefty, Lyle Overbay, out of the 701 pitches in our data set, and that resulted in a fly ball out. To righties he works the changeup down and away, mostly out of the strike zone. When he gets it in the zone, they make contact. The changeup looks like Bedard’s weakest pitch.

Bedard throws the curveball down and away to lefties, and he generates a lot of swings with it–foul balls, swings and misses, and balls in play. Against righties he also works down and away but isn’t afraid to throw it in the zone. He gets a lot of swings and misses and when the ball is put in play, it’s hit weakly (.227 AVG and .318 SLG). The curveball is a great pitch for Bedard; no wonder he throws it so much.

I wanted to add a note at the end here about which pitches Bedard used to get his strikeouts. We have PITCHf/x data for 50 of his 221 strikeouts. Of those 50 K’s, 22 of them came on the fastball, 21 on the curveball, and 7 on the cutter. That lines up pretty well, percentage-wise, with his pitch mix with two strikes on the hitter.

Hopefully, we’ve learned a little about how Bedard dominated hitters in 2007–a strong fastball/cutter combo and an outstanding curveball. His changeup could use improvement, but it’s his fourth pitch, so that’s really a small complaint. It will be interesting to see if he can maintain the strong performance in 2008 as well as whether he will be doing so as part of the Orioles or on a different team.


I have two new articles up at the Hardball Times.

The first is a short article on THT Live breaking down Francisco Liriano’s April 13 start against the Kansas City Royals.

The second is an article examining the ways in which ball tracking technologies like PITCHf/x are changing the game and what kinds of analysis are possible with this new data. It’s an expansion on my opening day laundry list of ideas that I posted here.

I posted a brief evaluation of the MLBAM pitch classification algorithm on the THT Live blog. So far I am not impressed with the system, but maybe there is hope for some improvement.

Update 4/11: Dan Fox reports that some improvements have been instituted for the MLB classification system this week. I’m in the process of taking a look at some data for a few other pitchers. This new data set includes a few starts from Thursday, April 10, which I believe should be covered under the improved algorithm that incorporates information about the pitches in a pitcher’s repertoire. I’ll report back if and when I learn something from this study.

I posted a short article at the Hardball Times examining Johnny Cueto’s pitching repertoire from his outstanding debut game on April 3.

I have published a new article at MVN analzying Kelvim Escobar.

I wrote a guest column for Rotojunkie, one of my favorite baseball discussion hangouts, analyzing the pitch repertoire of James Shields.

I have a new article at MVN analyzing Johan Santana’s pitches.

I have a new article at MVN analyzing Erik Bedard’s pitches.

If you’re interested in the raw data, you can download the Excel spreadsheet I used for the Bedard analysis.

I have a new article at MVN on the subject of pitch classification.  I recount a history of pitch classification with PITCHf/x data, offer some lessons learned, and discuss challenges for the future.

I have a new article at MVN on Kansas City Royals closer Joakim Soria.

If you’re interested in the raw data, you can download the Excel spreadsheet I used for the Soria analysis.

Next Page »