Not You Again! What Pitchers Throw Different Times Through the Order

Relief pitchers have it easy, relatively speaking. Of course when thrust into the game, they may be faced with a high leverage situation not of their making. This is not the easy part. From a pitch selection point of view, however, they have at least two clear advantages.

To start, they do not need to concern themselves with saving energy for two hours from the time they enter the game. This allows relievers to "air it out", which we can see in higher average velocities when compared to starting pitchers.

Another advantage is that short of unusual circumstances, relievers will never face the same hitter twice in the same game. This affords them the opportunity to stick with a smaller number of pitch types, which is in itself beneficial as potential sub-par offerings can be tossed aside given the lack of utility for the ability to present different looks to the hitter.

Starting pitchers, on the other hand, have to square off against the same hitters multiple times on the same day. In addition to generally requiring a more varied repertoire to be successful in such a role, starters must come up with a game plan to decide which pitch types to offer at which frequencies as they work through the order multiple times.

While I have a general idea of how I think pitchers operate in this respect, I realized that I did not recall ever seeing specific data on pitch type usage broken down by times through the order. So this is what I will be investigating herein.

Familiarity breeds contempt - generally attributed to Aesop, circa 600 B.C.

The first point to establish is that as pitchers face the same hitters repeatedly in the same game, it is certainly the hitters who are gaining the upper hand. To demonstrate this point, as well as all others in this article, I will take the group of 128 starters from the 2012 season who threw at least 400 pitches to hitters faced the third times in individual games. Here are some basic stats for the average pitcher in this sample by times facing a hitter in a game:

Times Hitter Faced	K%	UIBB%	K-UIBB%	wOBA Against	ERA	Pitches
1	21.1%	7.0%	14.1%	.301	3.89	1060
2	18.7%	6.7%	11.9%	.314	4.05	948
3	17.1%	6.8%	10.3%	.325	4.51	728
4	16.3%	6.6%	9.7%	.306	4.08	84

Things look worse for pitchers across the board as they are forced to pitch to guys they've already seen earlier in the game. K-UIBB% drops, wOBA against rises and ERA follows suit. Note that the fourth time through the order, things look to improve. This I believe is due primarily to selection bias, as pitchers who last in a game long enough to face hitters a fourth time are likely to be performing better than the average pitcher. As you can see from the number of pitches thrown in a season, there is a very large drop off between the third time facing a hitter and the fourth.

Variety is the spice of life - William Cowper, 1785

Now that we've established the struggles that starters face in this regard, we can break down their performance times through the batting order by the variety of pitch. For this I am using the straight GameDay PITCHf/x pitch classifications, and will only present those pitch types that make up at least five percent of all pitches thrown.

First we can look at pitch performance, using wOBA against:

Times Hitter Faced	FF	FT	FC	SI	SL	CU	CH
1	.340	.331	.298	.338	.236	.232	.271
2	.339	.340	.324	.347	.284	.271	.294
3	.354	.363	.339	.342	.289	.271	.314
4	.319	.359	.313	.310	.280	.261	.314

Offspeed pitches are the most successful. I would suggest though that pitchers cannot just throw all offspeed pitches (except R.A. Dickey) as for one, part of their success lies in the difference of these pitches compared to the fastball that has to stay on the hitters' minds, and secondly, pitchers' arms may not hold up to that many breaking balls.

Next we can look at pitch usage:

Times Hitter Faced	FF%	FT%	FC%	SI%	SL%	CU%	CH%
1	36.0%	15.9%	6.5%	8.9%	11.4%	9.3%	9.6%
2	30.1%	14.3%	6.7%	7.8%	13.9%	12.2%	12.2%
3	29.3%	13.6%	6.7%	7.4%	14.3%	12.8%	12.7%
4	28.5%	14.7%	7.2%	7.6%	14.4%	12.2%	11.7%

This in general lines up with my expectations. The average starter establishes the fastball early, throwing it the most the first time through the order, much like a reliever does in facing a hitter only once. This makes sense to me given that I would assume velocity could be higher at the start of the game when the pitcher is fresh, so why not make use of the extra ticks on the fastball while they are available. As starters work their way through the order the second time and beyond, we see offspeed pitch usage rise, at the expense of the fastball. These are the offerings that starters shift to in a likely attempt to disrupt the timing that a batter may have learned from an earlier plate appearance.

It is interesting to me that between the first and second time through the order, while offspeed pitches are increased in use, the slider+curveball+changeup combination only increases 8%, often just about one pitch per inning. So it certainly isn't a wholesale departure from the fastball that pitchers make, but enough of a transition that it is certainly perceivable when looking at a larger amount of data.

Best Pitchers vs. Worst Pitchers

Another question that I was curious about is whether pitchers that had more success showed a different pitch use profile than those who didn't fare so well. To look at this, I broke the sample into quartiles based on their ERA and revisited the pitch selection times through the order. Sorry for the barrage of tables...

Times Hitter Faced	FF%	FT%	FC%	SI%	SL%	CU%	CH%	Pitches	K%	UIBB%	K-UIBB%	ERA
1	37.4%	16.9%	5.1%	6.1%	12.2%	8.4%	9.9%	1098	23.7%	5.8%	17.9%	2.93
2	32.0%	15.3%	5.5%	5.3%	14.9%	10.5%	12.1%	989	20.7%	6.0%	14.7%	3.08
3	31.4%	14.5%	5.4%	5.3%	15.0%	11.2%	12.4%	796	18.6%	6.1%	12.5%	3.54
4	32.5%	13.8%	5.8%	4.7%	14.3%	10.4%	12.4%	119	18.8%	4.8%	14.0%	3.33

Top 25%

Times Hitter Faced	FF%	FT%	FC%	SI%	SL%	CU%	CH%	Pitches	K%	UIBB%	K-UIBB%	ERA
1	39.2%	14.0%	6.6%	6.9%	10.5%	9.6%	11.2%	1119	22.5%	7.1%	15.4%	3.72
2	32.0%	12.7%	6.6%	6.5%	12.6%	13.3%	13.9%	1000	18.9%	6.4%	12.5%	3.62
3	31.4%	12.1%	6.7%	6.1%	12.8%	13.8%	13.9%	784	17.5%	6.8%	10.7%	3.98
4	29.6%	13.3%	5.8%	6.4%	13.3%	14.9%	14.0%	85	18.4%	5.7%	12.7%	3.36

Second 25%

Times Hitter Faced	FF%	FT%	FC%	SI%	SL%	CU%	CH%	Pitches	K%	UIBB%	K-UIBB%	ERA
1	35.3%	19.1%	8.8%	5.7%	10.2%	9.9%	9.1%	1006	19.5%	7.4%	12.1%	3.87
2	29.1%	17.6%	8.9%	4.9%	12.8%	12.7%	11.5%	911	18.1%	6.6%	11.6%	4.09
3	28.4%	16.4%	9.0%	4.6%	13.4%	13.4%	12.2%	661	16.4%	6.7%	9.7%	5.02
4	29.1%	18.2%	11.5%	4.9%	13.8%	11.6%	8.3%	65	15.9%	7.3%	8.5%	3.14

Third 25%

Times Hitter Faced	FF%	FT%	FC%	SI%	SL%	CU%	CH%	Pitches	K%	UIBB%	K-UIBB%	ERA
1	32.0%	13.5%	5.5%	16.9%	12.7%	9.5%	8.4%	1017	18.6%	7.8%	10.8%	5.04
2	27.3%	11.7%	5.8%	14.4%	15.4%	12.2%	11.5%	894	16.8%	7.9%	8.9%	5.41
3	26.1%	11.3%	5.8%	13.7%	16.2%	12.7%	12.3%	672	16.0%	7.6%	8.3%	5.51
4	22.9%	13.3%	5.8%	14.3%	16.4%	11.8%	12.0%	68	12.2%	8.7%	3.5%	6.50

Bottom 25%

There are a bunch of quick observations that can be made about the tables. They offer yet another piece of evidence that strikeouts and walks are important, as we see a steady decline of K-UIBB% as we move through the worse ERA quartiles. The selection bias for the fourth time through the order is also now apparent, as pitchers who performed better faced more hitters for a fourth time in the same game than those who performed more poorly. Also, pitchers in the middle two groups start out relatively equal the first time through the order, then diverge rapidly as they are forced to deal with the same hitters again and again.

As far as pitch types go, the top two groups in general threw more four-seam fastballs than the lower two groups. The bottom two quartiles saw many more sinkerballers, as well as pitchers who rely more on the cutter. This finding lines up with the lower strikeout rates in these groups that we tend to associate with such pitchers.

With respect to offspeed pitches, at least in this sample from the 2012 season there did appear to be a higher use of the changeup, and use of it earlier in games in the better performing groups. This brings me to the pattern of pitch use that I had anticipated I would see before starting this study.

Fastball, then Changeup, then Breaking Balls

This was the order that I had expected to see most often. Establish the fastball while velocity is highest, and then mix in the changeup next while the difference in velocity between the two is likely to be the furthest apart. Then rely more on breaking balls the final time through the order.

If I look at starters who have peak changeup usage in the second time through the order, there are 41 out of the 128, or roughly one third. So this pattern appears to be no more common than an equal distribution would yield. Interestingly, if I look at these starters by team, there are certain teams with many pitchers using this pattern. The Rays (4), White Sox (4), Nationals (3), Athletics (3) and Rangers (3) perhaps believe in this game plan? Note that these are all teams with a perception of really getting pitching. That said, many of these pitchers do not use the changeup very often. If we restrict our sample to only those pitchers who throw at least 10% of their offerings as changeups, only one team remained with at least three starters. The Tampa Bay Rays, with four - David Price, James Shields, Jeremy Hellickson and Matt Moore.

"Saving" a Pitch

Another avenue that I wanted to investigate is which pitchers, if any, "save" a pitch for later in the game. In this case, I am thinking of a pitch type that is used infrequently the first time through the order (say <10% of all pitches) and then suddenly broken out with reasonable regularity the second time through the order (say >10% increase).

There are only two pitchers who did this in 2012: James Shields and Jered Weaver.

Shields had the most drastic jump of any pitch type, offering his curveball just 7.5% of the time the first time through the order, then ramping it up to 20.5% usage in the second round.

Weaver barely utilized his changeup at the start of the game, throwing it just 6.5% the first time through the order. He then jumped that up to a more robust 16.7% of all pitches the second time through the lineup.

Here is a table of all pitchers who experienced a ten percent increase in a pitch type between the first and second times through the order:

Pitcher	Pitch	Increase
James Shields	CU	12.9%
Ervin Santana	SL	12.9%
Justin Masterson	SL	10.8%
Wandy Rodriguez	CU	10.4%
Chris Sale	SL	10.2%
Jered Weaver	CH	10.2%

Ervin Santana essentially just slowly exchanges four-seam fastballs for sliders over the course of a game. The point in the game where he begins to throw more sliders than four-seamers has moved earlier and earlier in the game over the years, from never occurring to the 9th inning to the 7th inning last season and now as early as the 6th inning in the early part of 2013. This can be seen in the "Seasonal Trends" section of his Brooks Baseball player card.

Justin Verlander

As I'm guessing you already know, Justin Verlander is a pretty special pitcher. Looking at pitchers in this light, he also stands out, as he was the only pitcher to actually increase fastball usage the second time through the order as a result of his four-seam fastball. In general, if we group all of the GameDay F* pitch types together as fastballs, there were a few others as well (Bronson Arroyo and Carlos Zambrano).

A couple of honorable mentions here who used four-seam and two-seam fastballs just less the second time through the lineup than the first were Alex Cobb and Johnny Cueto.

Mike Leake

Finally, Mike Leake stands out as having the most peculiar pitch use pattern. While his overall fastball usage does drop slightly every time through the order, he organizes his game plan such that he shifts from sinkers and changeups early into more cutters and curveballs as the game progresses.

Summary

Here is the full pitch use table for the 128 starters used for this study.

In the end, we confirmed that batters gain an advantage as they face the same pitcher multiple times in the same game, and that pitchers in general establish their fastball the first time through the lineup and gradually introduce more offspeed pitches throughout the game.

I did run some regression analyses with respect to certain pitch use patterns and results, but nothing incredibly significant turned up.

As should maybe come as no surprise, there are many ways of putting together a pitching game plan, which I'm sure is due in no small part to every pitcher having his own particular pitch repertoire and relative strengths of each pitch within it.

Did anything surprise you about the information presented?

You can follow me on Twitter at @MLBPlayerAnalys. <a href="https://twitter.com/MLBPlayerAnalys" class="twitter-follow-button" data-show-count="false">Follow @MLBPlayerAnalys</a>
<script>!function(d,s,id){var js,fjs=d.getElementsByTagName(s)[0];if(!d.getElementById(id)){js=d.createElement(s);js.id=id;js.src="//platform.twitter.com/widgets.js";fjs.parentNode.insertBefore(js,fjs);}}(document,"script","twitter-wjs");</script>

Credit and thanks to Baseball Heat Maps for PITCHf/x data upon which this analysis was based and Baseball Reference for some split data.