SB Nation User Blog: Dan Turkenkopf

The Need for Speed

Dan Turkenkopf — Thu, 10 Jul 2008 00:27:14 -0000

Let's talk about the fastball. The fastball is what made Walter Johnson the Big Train. The fastball is why Nolan Ryan is NOLAN RYAN!. The fastball made Colt Griffin millions as a first round pick. The fastball is why Daniel Cabrera still has a spot in a major league rotation and it's why Brian Bannister risks his career every start. The fastball makes scouts stand up and take notice and makes fans stand up and cheer. The fastball is the mythmaker of pitching. But how important is it really?

You'd probably say that a good fastball is nearly essential to pitching in the major leagues, and I'd be inclined to agree with you. Barring the handful (fingerful?) of knuckleballers and the occassional Jamie Moyer who float around the bigs, it's hard to find successful pitchers without some redeeming aspect to their fastball - be it speed or movement. But that's not very precise. And if you've read almost anything else I've written, you know I like to try to quantify things. Blame it on the Math degree that collects dust in my closet.

I wanted to measure the value of fastball speed - or how much was an additional mile per hour worth to a pitcher. So I took the full set of fastballs captured by Pitch F/X from 2005 (those pitches identified as a variety of fastball or over 88 mph). For every one of those pitches, I categorized it according to the speed of the pitch and calculated the linear weights run value to find the average run value for each speed category. The results are shown in this graph:

Of particular interest is the linear fit trend line which seems to do a fairly good job of matching the actual results - especially between 87 and 97 miles per hour. It's a little hard to see in the picture, so let me relay it here: y = .0002x - .0292. The key number is that .0002 which means that, in general, each additional mile per hour on the fastball adds .0002 runs per fastball thrown. That equates to .018 runs of ERA (technically, RA) for every mile per hour on the fastball.

Imagine two nearly identical starters. Say they each average 6 innings per start, and throw exactly 100 pitches in those starts. Of those 100 pitches, 60 are fastballs. The only difference between them is that Starter A throws an 85 mph fastball and Starter B average 95 mph. All else being equal, we'd expect Starter B to have an ERA .18 runs better than Starter A.[1].

Obviously nothing every works out that easily in real-life, but there does appear to be a distinct advantage to throwing the ball faster. And while there is likely a selection bias here, I think it might actually dampen the real effect. We've all see pitchers who throw the ball harder get many more chances to succeed than a soft tosser. That would drive down the average value of the faster pitches, thereby lessening the observable effect of speed.

Another possible concern is that the linear weights values I used are slightly off. You'd expect the average run value across all measured pitches to sum to 0, but the sample I have (all pitches, not just fastballs) sums to about -2.75. That suggests that at least some of the weights are off (which makes sense since I didn't calculate them myself for my sample - I just took numbers off the Web). The discrepancy in run values could confuse the results of this study if they don't cancel out across the sample size (roughly 390,000 fastballs).

In some ways, I'm not really surprised there's an advantage to throwing a faster pitch. To think otherwise, you'd have to believe that nearly every decision maker in baseball history has been mistaken. That said, I have to admit I expected a minimal advantage to throwing faster, especially considering the apparent selection bias favoring those pitchers with lesser fastballs. This definitely explains why pitchers like Daniel Cabrera still have jobs, and pitchers like Colt Griffin or Matt Anderson can go so high in the draft.

[1] Starter B gains 10 mph * .0002 runs per mph or .002 runs for each fastball over Starter A. Over 60 pitches, that's a .12 run difference. Since each pitcher only lasted 6 innings, the .12 run difference would be .18 over 9 innings.

Do Relievers Really Throw Harder?

Dan Turkenkopf — Thu, 26 Jun 2008 20:52:31 -0000

For much of baseball history, if a starter couldn't cut it in the rotation, he found himself out of baseball. With the decline in complete games and the rise of the relief pitcher throughout the last few decades, many failed starters have found a home in the bullpen. Pitchers like Dennis Eckersley, Eric Gagne, Jason Isringhausen and John Smoltz, among many others, have thrived as relievers when injuries or ineffectiveness forced them from starting.

In general, relievers have better numbers than starters. They strike out more batters, give up fewer hits, and have a better ERA (somewhat surprisingly, they also walk more batters). In The Book, MGL, Tango and Andy Dolphin calculated that the average pitcher will have an ERA .8 runs better as a reliever than as a starter. Conventional wisdom suggests pitchers perform better in relief than when starting because they're expected to pitch fewer innings and can throw harder. In this post, we set out to test that conventional wisdom using the Pitch F/X data.

Let me pause a moment (as I always do when writing these things) to offer a disclaimer. I know they're boring, but these things need to be said. If you want to, you can read this paragraph in the really fast voice that comes on at the end of bank commercials (you know, the "protected by FDIC" guy). That might make it a little more interesting for you.

I'm relying on the Pitch F/X internal pitch classification system to identify what type of pitch is thrown. Mike Fast took a look at this during the first week of the season and found it less than impressive for classifying pitches. Since then, there have been some changes to the algorithm, and it appears the classification system is more accurate now. However, there is some uncertainty in the pitch types used in this study. Also, I'm relying on a very small subset of information, so sample size will be an issue. We'll get into that more when I start talking about the methodology.

Now that that's out of the way, let's get to the actual study. I identified the 70 pitchers who had made at least one start and one relief appearance this season (as of June 25). From that list, I picked those pitchers who had at least 200 pitches in starts (roughly 2-3 starts) and 100 pitches in relief (2+ appearances). That brought the list down to 26 pitchers. Finally, I split that sample into fastball and junkball pitchers. Fastball pitchers were those pitchers who threw some variety of fastballs in at least 60% of their pitches as starters. Junkball pitchers were those who through less than 60% fastballs. In our sample of 26, there were 17 fastballers and 9 junkballers.

There are two main ways pitchers can throw harder in relief. They can either throw a higher percentage of fastballs, or they could throw each pitch harder. If I had to guess, I'd bet both reasons would contribute to the improved ERA for relievers.

First let's explore whether pitchers changed their approach in relief by throwing a higher percentage of fastballs. The average pitcher in the sample increased his fastball percentage 2% when relieving. That breaks down to 1% for fastballers and 4% for junkballers While there were some major swings in pitch selection (Josh Banks increased his fastball percentage 24% in relief), there's no real indication that pitchers throw a substantially different mix of pitches when starting versus relieving.

If it's not more fastballs being thrown, then it's got to be the pitchers putting more behind each pitch, right? The logic makes sense, if pitchers know they're not going to need to throw as many pitches, they can devote more energy to each pitch. But is that really the case?

Before we dive into the data, we need to discuss how we'd identify a pitch being thrown harder. The easiest way is to compare pitch speeds - higher should equal harder. Beyond pitch speed, we can look at movement. Here's where things get difficult. I'll admit, I'm not sure I've got the physics correct, but let's give it a try anyway. The Pitch F/X data gives us information about both the horizontal (pfx_x) and vertical break (pfx_z). For a much better description than I could possibly give, please refer to Dr. Alan Nathan's explanation of what they mean. I compared each pitcher's horizontal break and vertical break as a starter and as a reliever to determine when they had more break. My thought is that, if we hold speed constant, a greater break means the pitch is thrown harder (because it moves more with the same initial velocity).

I think it's time to get into some results. First, pitch speed. There wasn't much deviation in pitch speed across our sample. Most pitchers were within 2 MPH in either direction, which doesn't seem like it should make a major difference in outcomes (fastball speed of 87-88 while starting going to 88-89 while relieving). The pattern holds when looking at the pitcher's primary non-fastball pitch as well.

Because we saw little to no change in pitch speed from starting to relieving, I'm going to make a simplifying assumption that's almost certainly the absolute wrong thing to do mathematically. I'm going to assume speed is held constant when looking at movement. This allows me to apply my rule of thumb from before - where more movement = harder thrower.

Using that rule of thumb, we don't see much of a pattern when we look at fastball movement.

Movement Differences on Fastballs
Category	Pitchers
+H, -V	7
-H, +V	6
+H, +V	7
-H, -V	6

In this table, H represents horizontal movement and V represent vertical movement. A plus indicates the pitcher increased the movement in that direction, while a minus says the pitcher reduced the movement. I'm going to go out on a limb and call this table inconclusive.

But what about breaking balls? We already saw there wasn't much of a speed difference, but what about a difference in movement?

Movement Differences on Primary Breaking Pitch
Category	Pitchers
+H, -V	7
-H, +V	7
+H, +V	9
-H, -V	3

So we do tend to see more movement on breaking pitches (and R.A. Dickey's knuckleball) in relief than when starting. It's by no means definitive, but in this small sample, and according to my assumptions, pitchers do seem to be throwing slightly harder in relief than when starting.

But, if in fact pitchers are throwing harder as relievers, the effect shows up in pitch movement rather than pitch speed - which is somewhat contrary to conventional wisdom. And is this effect enough to explain the .8 run difference in ERA? It doesn't seem plausible to me. But if how hard a pitcher throws is not the cause, what can we attribute it to?

Other possible causes include only facing any given batter once per game instead of 2-3 times (the familiarity effect), not being credited with runs if you allow inherited runs to score (the not-my-fault effect) or possibly even facing more pinch hitters (because of the pinch hitter penalty outlined in The Book).

We started this article attempting to confirm the conventional wisdom that pitchers perform better in relief than as starters because they throw harder in relief. Surprisingly, that doesn't seem to be the case. Looking at the sample of 26 pitchers who have been both starters and relievers in 2008, the evidence they throw harder when relieving is scanty at best. There's a very minor (less than 1 MPH on average) difference in pitch speed, mixed results on pitch movement that may hint at a pitcher throwing a little bit harder on breaking balls, and little-to-no-change in the percentage of fastballs thrown. While these may all contribute to the .8 run ERA difference uncovered in The Book, it seems unlikely they explain the whole effect. Followup work should be done to examine a larger sample of pitchers and to attempt to quantify the run effect of a difference in pitch speed. And someone who has taken a physics class in the past decade should probably either confirm or reject my assumptions.

The data I used for this study can be found here.

Joba's a Starter, Now What?

Dan Turkenkopf — Fri, 23 May 2008 12:31:46 -0000

As of Wednesday night, Joba Chamberlain is back on the path to being a starting pitcher in the major leagues. And this Yankees fan couldn't be happier. For a team that was counting on Phil Hughes and Ian Kennedy to carry their rotation this season, and has been running Kei Igawa out there, the mere possibility of what Joba could bring is exciting.

But what's reasonable to expect from Joba as a starter? In The Book, Playing the Percentages in Baseball, MGL, Tango and Andy Dolphin found that in general, relievers have about a .8 run advantage in ERA over starters. Joba's season ERA of 2.66 would look more like 3.46. Surprisingly, the league ERA for the AL this year is only 4.05 at this point, so that only equates to a 117 ERA+. If we assume the league will end up with an ERA of around 4.40, then Joba would have an ERA+ of 127, putting him in the upper echelon of American League starters.

Of course we can't just assume a straight-line translation of his current stats (well, we could, but then this would be a really short post). First off, we're still dealing with a really small sample size. We can likely expect some regression to the mean going forward (somewhat technical note: the proper mean is likely not the average of all pitchers. If we have some reason to assume Joba fits into some other categorization, we'd want to regress to the mean for that category). Various projection systems have different ways of accounting for that expected regression. Let's look at what some of those systems had to say (all from before this season).

Dan Syzmborksi's ZIPS projected Joba to have a 4.43 ERA in 134 IP. Based on the innings pitched (although ZIPS doesn't predict playing time), I'm going to assume ZIPS either projected Joba solely as a starter, or perhaps in a hybrid role. Sean Smith's CHONE projections predicted Chamberlain to start 10 games, pitch only 65 innings (which works out to around 5 innings per start) and have an ERA of 3.88. Tango's Marcel projections only see Chamberlain throwing 37 innings at a 3.41 clip. Because of Joba's quick rise through the minors, the Marcel system doesn't have a lot to work with, which is reflected in the relatively low reliability Tango puts on that projection. I'd love to tell you what Baseball Prospectus and PECOTA predict, but I'm not a subscriber so we'll have to go without.

Since it appears that CHONE and Marcel assume Chamberlain is pitching in relief, let's add the same .8 runs to get a rough idea of Joba's projected performance as a starter. Assuming his current performance (plus the adjustment) as the best case projection, what is the range of possible performance based on projections?

ERA Projections for Joba as a Starter
Projection System	ERA
Performance	3.46
Marcel	4.21
ZIPS	4.43
CHONE	4.68

Not much help there, since those range from substantially above average, to slightly below average. Although, at this point, I think the Yankees would be pretty happy if they could be assured of average performance for the rest of the season.

ERA is just one part of the equation, though (and we're not even quite done with that one yet). An even bigger question is how many innings will Joba be able to throw as a starter. He was a starter at Nebraska and the minors (although he averaged less than 6 innings per start in the minors), so he has demonstrated some ability to throws a starter's innings. According to PITCHf/x, there were roughly 97 pitches thrown per team in the first 6 innings of games during the 2005-2008 seasons. Let's use that number as a target for Joba.

In all of his major league appearances, Joba has never thrown more than the 38 pitches he used in the game 3 of the Division Series against Cleveland last year (the game after the midge game). He has four other appearances of over 30 pitches, including Wednesday night. For three of the four appearances of over 30 pitches that we have PITCHf/x data for, Joba doesn't appear to suffer at all from the number of pitches. There's no apparent difference between his fastball speed at the beginning and the end of those appearances. If anything, he's throwing a little harder at the end, which shouldn't be surprising. His long appearances tend to be the ones where he gets himself into a little bit of a jam. As we can guess from the mild controversy over Joba's enthusiastic reactions, adrenaline appears to play a big role in how he pitches, and a tight spot could easily lead him to throw harder. The one exception was the playoff game in Cleveland, where he throws 2-3 mph slower at the end of the appearance than at the beginning. I've got no compelling reason why we might see a dropoff there, except that it's the end of a long season.

In general, it seems appearances of 30+ pitches don't seem to really affect how Joba pitches. But 30 pitches is very different from 90 pitches. Joba's averaged right around 15 per inning, which would say he could get through 6 innings in 90 pitches. But that assumes he's able to not change his style.

The conventional wisdom is that pitchers throw harder as relievers than as starters because they don't have to pace themselves (one of the things that leads to the .8 difference in ERAs mentioned above). It's likely Joba's going to lose a few miles off all his pitches. The 98 mph fastball might drop to 93-94 on average while starting (just a guess, I'll have more research on that sometime next week), which probably makes it somewhat easier to hit. The big question, though, is what happens to Joba's slider. According to Josh Kalk, of the 184 pitches recorded, 66 of them were sliders - over 35%. Those sliders were essentially unhittable. Only 4 batters even managed to put one in play, and not a single batter got a hit over that sample.

But can Joba handle throwing the same percentage of sliders under a starter's workload? At that point, we're talking 30-40 sliders a game. I can only imagine the strain that would put on his arm (not being a pitcher or a doctor, I'm not really sure, but it sounds like a lot). And will those sliders be as effective if he takes a little bit off? Theoretically, Joba would start to mix in his curve and change a lot more, but those pitches aren't quite as polished as his slider, and certainly won't be as unhittable.

Obviously this is all conjecture, but what do I think is a reasonable expectation for Joba as a starter this season? Well, first, I think it's going to take a couple of months before he's "stretched out" enough to effectively start. I don't anticipate real starts (ones where he's expected to go 5+) until late July/early August. The timetable might be faster, if he's been building stamina by throwing on the side. As much as the Yankees have a plan to get him more innings at a time, they still need to win games - and Joba's usage pattern will be affected by that. If a starter is pitching really well in a tight game, will Giradi put Chamberlain in as planned?

But once Joba starts, I expect good things. Perhaps it would be more accurate to say I hope for good things. I do think the projections are all a little high. I'd go with an ERA of around 4.00 as a starter, averaging 5-6 innings per game. It's nothing spectacular, but it's a good start and would bode well for next season. I don't think Joba will be enough to get the Yankees to the playoffs this season by himself, but he should shore up the rotation and increase their chances.

A Nibble Here, A Nibble There

Dan Turkenkopf — Sat, 10 May 2008 03:05:56 -0000

We've already looked at the impact different factors have on whether a ball is called a strike or not. MGL suggested that I try and control for the fact that certain pitchers are much more likely to hover around the edge of the plate rather than down the middle or "just a bit outside"</Bob Uecker>.

Generally, any time MGL makes a suggestion, it's a good idea to at least consider it. So I went ahead and re-ran all the same splits except the one based on how a pitcher started a game, using the same approach as in my previous articles. The change is that I only looked at pitches within two ball widths (just under 6 inches) of the edge of the strike zone (either inside or outside the zone).

I also tweaked my run value of a "mistake" thanks to a comment from Iblemetrician that pointed out that missed pitches are probably more likely to occur at some counts rather than others. The value I'm using for these results is .133 runs per mistake, which is substantially lower than the original value of .161.

There's also going to be some level of trickery here. I'm going to continue to display all numbers in terms of runs/150 pitches just so we can get a better idea of how the old values compare to the new values. But since we're only looking at close called pitches in the second case, we get fewer per game - only about 70 compared to 150. So even if the runs per 150 pitches look a lot higher in the new tables, the actual effect on the game might be less. For a quick conversion to runs per game for the close called pitches, divide by 2.

I'm going to present the numbers from last time again (using the new run value) so we can see how they compare to the close pitches. This should allow us to get a better handle on what the bigger determining factor is - the number of pitches on the edges, or how those pitches are called. I've left out the graphs this time because they don't really show any tremendous differences.

A couple of labeling notes to keep in mind: CP = Called Pitches, CCP = Close Called Pitches

And finally, before we get started, the disclaimer. These splits are not necessarily indicative of skill. They measure less than one full season and include many other factors that should be corrected for. As time passes, we should be able to complete more technically rigorous analysis (that's the royal we, as in someone else who knows more about statistics) that may begin to clarify what percentage is skill and what percentage is unexplained/random variation. Until that time, I strongly recommend not using this information for anything more than entertainment purposes.

Game Specific Factors

Inning

Called Pitches By Inning
Inning	CP	% of Total CP	Runs / 150 Pitches
1	21281	12.8	0.09
2	19165	11.5	0.01
3	19076	11.5	-0.04
4	18093	10.9	0.03
5	18209	10.9	-0.02
6	18028	10.8	-0.07
7	18482	11.1	0.01
8	18482	11.1	-0.03
9	13029	7.8	0.05
10+	2588	1.6	-0.15

Close Called Pitches By Inning
Inning	CCP	% of Total CCP	Runs / 150 Pitches
1	11385	13.1	0.12
2	10395	11.9	-0.04
3	10019	11.5	-0.12
4	9345	10.7	0.02
5	9405	10.8	-0.04
6	9214	10.6	-0.10
7	9464	10.9	0.11
8	9629	11.1	-0.08
9	6783	7.8	0.22
10+	1352	1.6	-0.36

Wow. The obvious thing that jumps out here is how well pitchers do on close pitches in the ninth inning. Following up from my last article, I feel more confident suggesting that the quality of the pitcher plays a role in which way the calls go. Closers appear to get the benefit of the doubt, while the extras appear to be left to the back of the bullpen. I'd still like to think there's some sort of umpire fatigue that explains the extra innings too. It just seems too human an explanation to ignore.

Home or Visitor

Called Pitches By Pitching Team
Pitching Team	CP	% of Total CP	Runs / 150 Pitches
Home	84524	50.8	0.05
Visitor	81909	49.2	-0.05

Close Called Pitches By Pitching Team
Pitching Team	CCP	% of Total CCP	Runs / 150 Pitches
Home	84524	50.9	0.08
Visitor	81909	49.1	-0.09

I changed around the labels this time to try and make things more clear. It's now organized according to pitching team - so the positive number for the Home team is actually a positive benefit for the pitchers.

Beyond the change in labels, there's really no difference in looking at the edges versus all pitches - still a slight advantage to the home team.

Pitcher Factors

Age

Called Pitches By Age
Age	CP	% of Total CP	Runs / 150 Pitches
20	2790	1.7	0.10
21	2944	1.8	0.00
22	4507	2.7	-0.13
23	11582	7.0	-0.23
24	13041	7.8	-0.14
25	15982	9.6	-0.15
26	15387	9.2	0.04
27	17369	10.4	-0.16
28	11809	7.1	-0.10
29	15636	9.4	0.23
30	13211	7.9	-0.01
31	7844	4.7	0.17
32	5177	3.1	0.12
33	8705	5.2	0.01
34	2392	1.4	.16
35	3190	1.9	0.14
36	3187	1.9	0.09
37	1851	1.1	0.60
38	903	0.5	0.49
39	2645	1.6	0.31
40	2184	1.3	0.31
41	1603	1.0	0.33
42	276	0.1	-0.84
43	1498	0.9	0.05
45	720	0.4	0.59

Close Called Pitches By Age
Age	CCP	% of Total CCP	Runs / 150 Pitches
20	1488	1.7	0.18
21	1541	1.8	0.02
22	2340	2.7	-0.25
23	6060	7.0	-0.46
24	6675	7.7	-0.23
25	8367	9.6	-0.30
26	7986	9.2	0.05
27	9230	10.6	-0.26
28	6140	7.1	-0.12
29	8064	9.3	0.47
30	6899	7.9	-0.03
31	4143	4.8	0.35
32	2676	3.1	0.16
33	4571	5.3	0.07
34	1256	1.4	0.31
35	1627	1.9	0.32
36	1672	1.9	0.10
37	1006	1.2	1.04
38	448	0.5	0.89
39	1345	1.5	0.51
40	1222	1.4	0.33
41	928	1.1	0.26
42	134	0.2	-1.35
43	805	0.9	0.06
45	368	0.4	1.13

The most surprising thing here is that the proportions for each age stayed pretty close from all called pitches to the pitches on the edge. I thought (and it was suggested by more than one person) that older pitchers would be be more likely to nibble and therefore we'd see a higher proportion of close pitches going to the elders. That's clearly not the case, as you can see even more clearly below. This means that the difference according to age is mostly composed of selection bias and umpires actually giving older pitchers the benefit of the doubt.

Called Pitches by Age Groups
Age	CP	% of Total CP	Runs / 150 Pitches
Under 25	50846	30.6	-0.12
26-35	97530	58.6	0.03
Over 35	18057	10.8	0.25

Close Called Pitches by Age Groups
Age	CCP	% of Total CCP	Runs / 150 Pitches
Under 25	26471	30.4	-0.25
26-35	50965	58.6	0.06
Over 35	9555	11.0	0.37

Experience

Called Pitches by Experience
Experience	CP	% of Total CP	Runs / 150 Pitches
0	17814	10.7	-0.11
1	20275	12.2	-0.16
2	21048	12.6	-0.07
3	16597	10.0	-0.16
4	14579	8.8	-0.05
5	11218	6.7	0.09
6	10098	6.1	-0.03
7	9716	5.8	0.12
8	13236	8.0	0.04
9	6638	4.0	0.23
10	6063	3.6	0.29
11	1868	1.1	0.24
12	3914	2.4	0.33
13	499	0.3	-0.41
14	2722	1.6	0.30
15	3541	2.1	-0.11
16	863	0.5	0.62
18	929	0.6	0.50
19	1494	0.9	0.56
20	1625	1.0	0.17
21	1522	0.9	0.26
23	139	0.1	0.53

Close Called Pitches by Experience
Experience	CCP	% of Total CCP	Runs / 150 Pitches
0	9273	10.7	-0.22
1	10584	12.2	-0.25
2	10874	12.5	-0.08
3	8724	10.0	-0.15
4	7594	8.7	0.22
5	5909	6.8	-0.17
6	5310	6.1	0.00
7	4967	5.7	0.08
8	6846	7.9	0.07
9	3511	4.0	0.38
10	3277	3.8	0.50
11	958	1.1	0.78
12	2041	2.3	0.26
13	240	0.3	0.24
14	1496	1.7	0.41
15	1889	2.1	-0.00
16	454	0.5	1.03
18	443	0.5	0.21
19	769	1.0	0.92
20	863	1.0	-.81
21	878	1.0	-.33
23	69	0.1	0.94

Called Pitches by Experience Groups
Experience	CP	% of Total CP	Runs / 150 Pitches
Under 2	59137	35.5	-0.11
3-7	62243	37.4	-0.03
Over 8	45053	27.1	0.18

Close Called Pitches by Experience Groups
Experience	CCP	% of Total CCP	Runs / 150 Pitches
Under 2	30731	35.3	-0.20
3-7	32526	37.4	-0.04
Over 8	23734	27.3	0.32

Looking at experience shows pretty much the same thing as age - which is what I'd expect since they're so closely correlated.

Handedness

Called Pitches by Hand
Hand	CP	% of Total CP	Runs / 150 Pitches
Left	44213	26.6	-0.07
Right	122220	73.4	0.03

Close Called Pitches by Hand
Hand	CCP	% of Total CCP	Runs / 150 Pitches
Left	23174	26.6	-0.15
Right	63817	73.4	0.05

There's nothing too interesting in this comparison. Righties still have a slight advantage over lefties.

Runs Allowed

Called Pitches by Runs Allowed
RA/9	CP	% of Total CP	Runs / 150 Pitches
Less Than 4.00	30491	20.5	-0.01
4.00-6.00	97973	66.0	0.05
More Than 6.00	20086	13.5	-0.20

Close Called Pitches by Runs Allowed
RA/9	CCP	% of Total CCP	Runs / 150 Pitches
Less Than 4.00	16069	20.7	-0.05
4.00-6.00	51199	66.0	0.09
More Than 6.00	10416	13.5	-0.37

Again, nothing to see here. (I suppose I'm expected to say "Pass.")

Walk Rate

Called Pitches by Walk Rate
BB/9	CP	% of Total CP	Runs / 150 Pitches
Less Than 2.50	21241	14.3	0.15
2.50-4.50	102769	69.1	0.01
More Than 4.50	24540	16.5	-0.18

Close Called Pitches by Walk Rate
BB/9	CCP	% of Total CCP	Runs / 150 Pitches
Less Than 2.50	11524	14.9	0.25
2.50-4.50	53698	69.1	0.02
More Than 4.50	12462	16.0	-0.30

There is a slight variation here in the percentage of total pitches for each category. Sure, .6% isn't very big, and could easily have no identifiable cause, but it does look like pitchers with good control have more called pitches near the edges. That's not very surprising of course, since they're likely going to have fewer called pitches well outside the strike zone (those are largely balls), which means more will be in or around the strike zone.

Summary

Here we are again - the what does this all mean moment. As I disclaimed up top, it's too soon to tell. Before we go any further, let's bring the summary charts from the other article, where we look at the largest possible spread in ERA for a given split.

ERA Effect for All Called Pitches
Breakdown	Good	Bad	ERA Effect
Inning	First	Extras	.03
Pitching Team	Visitor	Home	.01
Age (Buckets)	Over 35	Under 25	.04
Experience (Buckets)	Under 2	Over 8	.03
Handedness	Righties	Lefties	.01
Runs Allowed	4.00-6.00	More Than 6.00	.03
Walk Rate	Less Than 2.5	More Than 4.5	.04

ERA Effect for Close Called Pitches
Breakdown	Good	Bad	ERA Effect
Inning	Ninth	Extras	.03
Pitching Team	Visitor	Home	.01
Age (Buckets)	Over 35	Under 25	.03
Experience (Buckets)	Under 2	Over 8	.03
Handedness	Righties	Lefties	.01
Runs Allowed	4.00-6.00	More Than 6.00	.02
Walk Rate	Less Than 2.5	More Than 4.5	.03

So even though the spread between best and worse is generally a lot bigger for the close pitches, the fact that they only happen about half as often drives the actual effect to be the same or less than for all called pitches. That's not to say that half the effect is from close called pitches and the other half from no-doubters. It looks like most of the mistakes, and most of the variation, occurs on close called pitches. However, there doesn't appear to be a whole lot of accuracy gained on the runs/game level by breaking out only the pitches on the edges.

That said, we did learn a few things. Surprisingly, there appears to be very little difference between age groups in the amount of called pitches that are near the edges. That seems to go against the conventional wisdom of older pitches starting to place their pitches more as they lose a little off their fastballs. I wonder how many pitchers really do make that transition successfully, and how many keep following the same strategy and either succeed (and not change) or fail (and retire). These numbers suggest not too many really do change their approach too much.

To wrap up, again I caution you against reading too much into these results. There's way too many unknowns and, even worse, known issues with the data and the criteria used to determine the splits to make them that useful. They can be used to start discussion and further research, but I definitely wouldn't try to work them into projections any time soon.

A Strike Is a Strike, Right?

Dan Turkenkopf — Thu, 24 Apr 2008 11:07:25 -0000

After my last post about catcher framing, many people (at least two or three) suggested that the pitching staff had a fair amount to do with the results we were seeing. The poor performance of both the Rangers' catchers (Gerald Laird and Jarrod Saltalamacchia) seemed to lend credence to that stance. I started wondering what factors might influence an umpire to call a pitch a certain way. So I retreated into my stats cave (no, it's not my mother's basement) for a while to see what I could I find out. I'll warn you, this is a very long post, so if you want to skip to the summary at the end, feel free, but you'll miss the graphs.

Before I start examining the different factors, let me outline the methodology. It's the same basic approach as the catcher framing article, which in turn is based on Jonathan Hale's article at The Hardball Times. Each called pitch is compared to the strike zones identified by John Walsh to see whether it was "mistakenly" classified. I put "mistakenly" in quotes because Walsh marked the boundaries of the strike zones where 50% of the pitches were called strikes, so it's expected that some calls will not match. Walsh provides values on the height and width of the strike zone for both left and right handed batters and compares them to the rulebook strike zone for the average batter. I used his values for the strike zone widths, but calculated my own for the strike zone height of each batter, based on his deltas from the rulebook strike zone. I used absolute differences (2.2 inches on the top for right handed batters, for example) rather than percentage differences.

Anyway, once I classified all the pitches, I credited the pitcher (or category of pitchers) for all additional strikes and debit for all additional balls to find how many total "misses" happened. I then calculated the average percentage of misses across all pitches and determined how each pitcher differed from his expected value. Those differences were normalized to 150 opportunites, or roughly the amount in one game. Finally, that number was converted to runs per 150 called pitches by multiplying by .161 runs, which is the value of changing a ball to a strike. This allowed me to compare different factors that might influence how a pitch was called and see how important each one might be.

Game Specific Factors

Certain characteristics of the game might play a part in how an umpire calls pitches. Specifically, I looked at which inning the calls happened, and which team was at bat. Another possible area to explore would be day games versus night games, but I didn't have the data handy to look into that.

Inning

Inning	Called Pitches	Runs / 150 Pitches
1	21281	0.10
2	19165	0.01
3	19076	-0.05
4	18093	0.03
5	18209	-0.03
6	18028	-0.08
7	18482	0.02
8	18482	-0.04
9	13029	0.06
10+	2588	-0.18

The most interesting piece of information here is the (relatively) large assist given to batters in extra innings. It's almost like the umpires want to go home and unconsciously give batters the benefit of the doubt so as to increase the chances of scoring a run. Of course the effect is pretty small - a difference of one run every 45 innings or so, but it's almost twice as large as any other inning. The fact that a team is more likely to use the back of their bullpen in extra innings could play a role too.

Home or Visitor

Batting Team	Called Pitches	Runs / 150 Pitches
Visitor	84524	0.06
Home	81909	-0.06

A mild advantage to the home pitchers (when the visiting team is batting) - to the tune of one run every 8 games. I expected the home team pitchers to get some favorable calls, but wasn't sure of the impact.

Pitcher Factors

How could a pitcher influence whether close pitches are called strikes? I separated the possiblities into two main categories - demographics and performance. Under demographics, I looked at age, experience and handedness, and for performance I looked at runs allowed, walks allowed and early-game wildness. I know that age and experience are close analogs but I thought I'd see if there were major differences between the two. More on that later.

Age

Age	Called Pitches	Runs / 150 Pitches
20	2790	0.12
21	2944	0.00
22	4507	-0.15
23	11582	-0.27
24	13041	-0.17
25	15982	-0.19
26	15387	0.05
27	17369	-0.19
28	11809	-0.12
29	15636	0.28
30	13211	-0.01
31	7844	0.20
32	5177	0.15
33	8705	0.01
34	2392	0.19
35	3190	0.16
36	3187	0.10
37	1851	0.73
38	903	0.60
39	2645	0.37
40	2184	0.38
41	1603	0.40
42	276	-1.01
43	1498	0.06
45	720	0.72

The numbers jump around a bit, and the sample size for some of the individual years leaves a lot to be desired, but in general, the older you are, the more love you get from the umpires. The effect is much clearer if we look just at age buckets

Age	Called Pitches	Runs / 150 Pitches
Under 25	50846	-0.17
26-35	97530	0.03
Over 35	18057	0.31

Pitchers over 35 see almost half a run per game benefit compared to pitchers under 25. There are some concerns with this. There's likely a selection bias as the pitchers who make it to 35 tend to be the better ones, so that might influence how the umpires rule. It's also possible that the determining factor isn't age, but experience. Let's look at that one next.

Experience

Experience	Called Pitches	Runs / 150 Pitches
0	17814	-0.13
1	20275	-0.19
2	21048	-0.09
3	16597	-0.20
4	14579	-0.07
5	11218	0.11
6	10098	-0.04
7	9716	0.15
8	13236	0.04
9	6638	0.28
10	6063	0.35
11	1868	0.29
12	3914	0.39
13	499	-0.49
14	2722	0.36
15	3541	-0.14
16	863	0.75
18	929	0.61
19	1494	0.67
20	1625	0.20
21	1522	0.31
23	139	0.64

There are a few strange negative bumps at 13 and 15 years, but the samples are pretty small there. In general the trend is upwards, and it appears more strongly than with age. Let's look at the buckets of experience.

Years of Experience	Called Pitches	Runs / 150 Pitches
Under 2	59137	-0.14
3-7	62243	-0.03
Over 8	45053	0.22

I think these understate the actual value of being a long-time pitcher - mostly because a large portion of the sample is in the set of pitchers with 8 years of experience. I'll admit these buckets don't match up well with the age buckets I introduced above, so the comparison is not as easy to make as I might hope.

Handedness

Hand	Called Pitches	Runs / 150 Pitches
Left	44213	-0.09
Right	122220	0.03

An advantage of 1 run for every 8 games or so to righties. I'm not sure why righties might get more favorable treatment. It could have something to do with the direction the pitches break, or maybe it's related to handedness of the batter - where righty pitchers face more left-handed batters than lefties do. Whatever the cause, I'm going to chalk it up to unexplained variation (Mike Emeigh recently suggested that term as opposed to random variation, and I think it's a good idea).

Runs Allowed

RA/9	Called Pitches	Runs / 150 Pitches
Less Than 4.00	30491	-0.01
4.00-6.00	97973	0.05
More Than 6.00	20086	-0.25

First off, let me make it clear that those runs allowed numbers are for a pitcher's career before 2007 while the runs / 150 pitches arer the 2007 season. There isn't a whole lot of variation between the low RA and the medium RA buckets. However, the high RA bucket loses a quarter of a run per game, or -.03 points of ERA in 2007 to umpires' calls. That suggests that umpires might be buying into a pitcher's bad reputation and reinforcing it. Interestingly, there doesn't appear to be a corresponding positive bump on the other side.

Walk Rate

BB/9	Called Pitches	Runs / 150 Pitches
Less Than 2.50	21241	0.18
2.50-4.50	102769	0.02
More Than 4.50	24540	-0.22

Again, the walk rate numbers are pre 2007 while the runs / 150 pitches are 2007. In this case we see that reputation appears to play a part in both directions. Pitchers who previously had low walk rates benefit from the umpires' calls, while those with high walk rates suffer. The difference between the two is .4 runs per game or .04 points of ERA. It's important to realize the walk rate is probably not independent from the rate of runs allowed - in other words, a pitcher who has a high walk rate is likely to give up a lot of runs - so these differences can't just be added together to figure out what part reputation plays in skewing the results.

Early-Game Wildness

Wildness	Called Pitches	Runs / 150 Pitches
Low (Less Than 33%)	14704	-0.01
Mid (33% - 42%)	23160	0.08
High (Greater Than 42)	11431	-0.15

This one needs some explanation. What I tried to do was identify how wild each starter was in the first two innings of each 2007 start by looking at the percentage of balls he threw. I then grouped them into low, medium and high buckets, and ran my standard analysis. This shows much the same result (although less pronounced) as the runs allowed case - in that the high bucket is hurt, but there's little effect on the low bucket. Basically, it appears that being wild early tarnishes you in the eyes of the umpire, while having good control early doesn't really do much. Again, there's going to be some causal overlap with this measure and the walk rate and runs allowed metrics, so keep that in mind.

Summary

So what does this all mean? Very good question. Again, I'm not sure. Since we only have one (partial) season to go on, it's hard to be that confident in the data. I think the sample sizes are pretty good for many of the buckets used in this study, but that's a gut feel rather than anything confirmed mathematically. In a moment, I'll share the results of a regression I ran for individual pitchers based on some of these metrics, but first let's look at a summary chart of all the different ways of breaking this down. The ERA effect is the difference in ERA between the "good" group in the sample (low walk rate, low RA, etc.) and the "bad" group (high walk rate, high RA).

Breakdown	Good	Bad	ERA Effect
Inning	First	Extras	.03
Batting Team	Home	Visitor	.01
Age (Buckets)	Over 35	Under 25	.05
Experience (Buckets)	Under 2	Over 8	.04
Handedness	Righties	Lefties	.01
Runs Allowed	4.00-6.00	More Than 6.00	.03
Walk Rate	Less Than 2.5	More Than 4.5	.04
Early-Game Wildness	Mid	High	.03

Again, please keep in mind these aren't independent so cannot be added together to get a total effect. There's nothing that's all that big here, at least compared to what I was finding for catchers (a spread of over a run per game).

Now let's get to that regression. I took all pitchers from 2007 who had over 300 pitches called (174 pitchers) and regressed the runs / 150 pitches twice - once looking at handedness, age, runs allowed and walk rate, and the second time replacing age with experience. Neither case explained more than .10 of the total variation, so I'm not even going to bother sharing the actual values. If anyone really wants to know, ask and I'll share.

I'm sure there are plenty of other things that might influence an umpire to call pitches a certain way for a given pitcher. A few that I didn't get a chance to look at for various reasons were GB/FB ratio and the pitcher's arsenal. Plus it's going to be nearly impossible to untangle the effect of the catcher and umpire until we have more data. But I think there's still value in examining these characteristics and finding the limits of their effects, even if the information is not conclusive.

Framing the Debate

Dan Turkenkopf — Sat, 05 Apr 2008 00:07:42 -0000

Or perhaps, framing, the debate.

The defensive influence of the catcher has long been debated, with sabermetric circles generally valuing it less than baseball insiders. It's possible to identify seven separate components as places where a catcher can contribute defensively: stolen bases, blocking pitches, blocking the plate, fielding bunts (and psuedo-bunts), game calling, pitch framing and pitcher preparation. Many people have made attempts at quantify different pieces of the overall catcher contribution. Ability to prevent stolen bases and fielding ability are included in a number of measures, such as UZR and Win Shares. Keith Woolner from Baseball Prospectus has examined the topic of game calling a few times and has placed an upper limit of the effect at .60 runs of ERA. I've attempted to measure how well catchers block pitches in the dirt. Blocking the plate is very difficult to measure since it occurs relatively rarely and success can be quite subjective (the runner might be out, but the catcher did nothing to impede him from reaching the plate). Determing a catcher's skill at blocking the plate is probably best left to visual analysis. Pitcher preparation, which I'm using to mean any interaction the catcher has with his staff, from going over game plans, to calming the pitcher during visits to the mound, is another area that would be nearly impossible to objectively measure, but almost certainly has some effect on the game. One area that hasn't been explicitly covered (I suspect many would roll it into game calling) is pitch framing.

I've attempted a study to measure the effects of framing pitches. I say attempted because, to be honest, the results seem wrong to me, but I can't figure out where the issue lies. I present the rest of this article in the hopes that someone can take the information and make sense of it. Some may say I shouldn't publish this until I'm more comfortable with the outcome, but I'm a big fan of collaboration and knowledge sharing, so I'm going ahead even though I don't know the answer. Ok, enough honesty, let's move on.

Back around Thanksgiving, Jonathan Hale broke down the strike zones of various umpires over at The Hardball Times. He used the PITCHf/x data and the strike zones figured by John Walsh to determine how many strikes above or below average an umpire called in the course of a game (for his purposes, 150 called pitches). The results ranged from -5.2 strikes for Gerry Davis to +4.81 strikes for Jeff Nelson, where negative numbers implies a smaller strike zone.

A similar methodology can be used to examine whether catchers have any effect on whether the pitch is called a ball or strike. Using Walsh's strike zones, empirically defined as the areas where at least 50% of pitches are called strikes, I determined for each catcher how many balls should have been strikes and vice versa. This allowed me to calculate an average rate and then figure out how many strikes above or belowe average each catcher was. The results for those catchers with more than 2000 called pitches (roughly 120 innings using Hale's 150 called pitches per game number) can be seen below. The run value is calculated using the run value for changing a ball to a strike - .161 runs, which I explain here.

Catcher	Called Pitches	SAA	SAA / 150 Pitches	Runs / 150 Pitches
Gregg Zaun	4518	159.70	5.30	0.85
Jeff Mathis	3430	93.43	4.09	0.66
Yadier Molina	3246	84.49	3.90	0.63
Russell Martin	7061	183.16	3.89	0.63
Yorvit Torrealba	2867	58.03	3.04	0.49
John Buck	3443	64.93	2.83	0.46
Brian McCann	5762	105.98	2.76	0.44
Javier Valentin	2013	36.79	2.74	0.44
Jose Molina	2313	41.28	2.68	0.43
Brian Schneider	2119	31.72	2.25	0.36
Ivan Rodriguez	3881	55.13	2.13	0.34
Brad Ausmus	2576	35.19	2.05	0.33
Johnny Estrada	3072	36.16	1.77	0.28
Carlos Ruiz	2471	11.22	0.68	0.11
Benji Molina	3909	14.06	0.54	0.09
Jorge Posada	2616	3.66	0.22	0.03
Paul LoDuca	2569	-0.54	-0.03	-0.01
A.J. Pierzynski	5932	-3.54	-0.09	-0.01
Victor Martinez	3233	-2.01	-0.09	-0.02
Josh Bard	5701	-15.68	-0.41	-0.07
Kurt Suzuki	4269	-15.75	-0.55	-0.09
Jason Varitek	3543	-14.90	-0.63	-0.10
Jason Kendall	5770	-25.32	-0.66	-0.11
Michael Barrett	2499	-15.86	-0.95	-0.15
Chris Snyder	3342	-23.19	-1.04	-0.17
Ronnie Paulino	2343	-16.87	-1.08	-0.17
Rob Bowen	2057	-26.90	-1.96	-0.32
Mike Napoli	3106	-45.14	-2.18	-0.35
Jamie Burke	2270	-35.07	-2.32	-0.37
Ramon Hernandez	2361	-39.56	-2.51	-0.40
Jarrod Saltalamacchia	2467	-41.63	-2.53	-0.41
Joe Mauer	2434	-44.36	-2.73	-0.44
Dioner Navarro	2436	-64.44	-3.97	-0.64
Kenji Johjima	7120	-201.11	-4.24	-0.68
Gerald Laird	6129	-317.09	-7.76	-1.25

I'll be the first to admit this is a much larger effect than I expected to see. In fact it's so large that I have to think there's something wrong in the analysis. Over the course of 120 games (a reasonable estimate for the number of games caught in a season by a starting catcher), the difference between Gregg Zaun and Gerald Laird is over 250 runs or 25 wins. The per game numbers, though, are similar to what Hale found for umpires. So either this style of analysis is lacking, or my run values are off.

But what about the effect of the umpire on this? After all, my study is based on Hale's work with the umpires. I did a weak correction for the umpires by using his numbers and determining what the expected strike delta would be for each catcher / umpire pair and subtracting that effect from the catcher's individual numbers. Since he only published umpires with a sample size greater than 20 games, I assumed the other umpires had 0 effect. This obviously gets into more drastic sample size issues than does the initial study, but I thought it would be interesting to look at.

Catcher	Opportunities	SAA	SAA / 150 Pitches	Runs / 150 Pitches
Gregg Zaun	4518	167.81	5.57	0.90
Yadier Molina	3246	83.25	3.85	0.62
Russell Martin	7061	167.61	3.56	0.57
Jeff Mathis	3430	77.43	3.39	0.55
Jose Valentin	2013	42.55	3.17	0.51
Jose Molina	2313	46.95	3.04	0.49
John Buck	3443	63.28	2.76	0.44
Brian McCann	5762	98.41	2.56	0.41
Brad Ausmus	2576	41.80	2.43	0.39
Yorvit Torrealba	2867	44.57	2.33	0.38
Ivan Rodriguez	3881	55.67	2.15	0.35
Brian Schneider	2119	29.76	2.11	0.34
Johnny Estrada	3072	32.23	1.57	0.25
Jorge Posada	2616	21.64	1.24	0.20
Benji Molina	3909	22.23	0.85	0.14
Paul LoDuca	2569	2.11	0.12	0.02
Carlos Ruiz	2471	1.56	0.09	0.02
A.J. Pierzynski	5932	2.11	0.05	0.01
Victor Martinez	3233	-4.29	-0.20	-0.03
Kurt Suzuki	4269	-6.32	-0.22	-0.04
Jason Varitek	3543	-5.81	-0.25	-0.04
Jason Kendall	5770	-21.46	-0.56	-0.09
Josh Bard	5701	-31.55	-0.83	-0.13
Ronny Paulino	2343	-15.23	-0.98	-0.16
Michael Barrett	2499	-20.02	-1.20	-0.19
Chris Snyder	3342	-34.53	-1.55	-0.25
Mike Napoli	3106	-42.51	-2.05	-0.33
Jamie Burke	2270	-33.45	-2.21	-0.36
Rob Bowen	2057	-30.50	-2.22	-0.36
Ramon Hernandez	2361	-35.80	-2.27	-0.37
Jarrod Saltalamacchia	2467	-48.45	-2.95	-0.47
Joe Mauer	2434	-48.64	-3.00	-0.48
Dioner Navarro	2436	-63.65	-3.92	-0.63
Kenji Johjima	7120	-189.77	-4.00	-0.64
Gerald Laird	6129	-314.00	-7.68	-1.24

Adding the influence of the umpire seems to dampen the effect somewhat, which is what we'd expect. Still, the spread between best and worst is huge. We're no better off than we were before.

What to make of all this? There are obvious quality and sample size concerns from the PITCHf/x data, which calls some of this into question. Beyond that, the results just seem too outlandish to be correct. Plus, there was only a weak correlation between these values and the team's runs allowed per game (r=-.30 based on those catchers who caught for the same team all year). I find it hard to believe pitch framing can have this big an impact and not be more noticeable. I have the same concern about the results of Hale's analysis of umpires - where the impact is nearly as big. Two other possibilities are that my run value number is wrong (likely, but I think it's in the ballpark) or that there's some underlying issue that affects both of our studies. Finally, maybe we've been wrong the whole time and catcher defense is really that important.