Player spotlight: Cliff Lee (Part 2)

If you missed part one of this series on Cliff Lee, check it out here. For part two, we’ll bring in some Pitch f/x data and see how Lee is doing what he is doing this year.

Pitch f/x

So, we know that Lee has been good this year. You probably didn’t need to read all of part one to realize that, but hopefully it gave some insight you weren’t yet aware of. But the question remains: how is he doing it? What has changed? Well, luckily, we now have Pitch f/x to help us try and figure this out. Let’s first look at Lee’s movement charts.


Speed and Movement – 2007

FB	73%	90.3	4.17		11.72
CB	7%	76.4	-4.09		-4.36
SL	4%	85.6	-0.92		6.93
CU	16%	83.3	8.58		8.91

Speed and Movement – 2008

FB	77%	91.3	6.06		11.56
CB	8%	75.6	-6.11		-5.21
SL	4%	83.9	-3.36		5.30
CU	11%	83.3	8.87		9.19

Well, we immediately see differences in the movement of Lee’s pitches. We see that he has added a good deal of horizontal movement to all four of his pitches. We see the least difference in his change-up, but a couple of things have been done to improve that pitch. First, let’s look at the fastball, though.

As I said with Johnny Cueto last week (with his change-up and slider, although I believe it applies here as well), added variability in a pitcher’s movement seems—to me anyway—like a good thing. Even if a batter can identify that the pitch coming in is a fastball from speed or release point or whatever, a wider range of horizontal movement means that they have less of an idea about where it’s actually going to end up.

Before, batters could have expected Lee’s fastball to move from 0 to eight inches (with a few spares ones around 10). Now, Lee is throwing them from -1 all the way out to 12 and 13 inches. That wider range means increased unpredictability. This year, batters could see the fastball coming in and have it break a full 12 inches in at them. Or, they could see it coming in and have it barely move at all. Or, they could see it coming in and actually see it break away from them slightly. Or, it could be somewhere in between. There wasn’t nearly this much guesswork last year.

Furthermore, we see a little bit of additional vertical movement on the fastball. It doesn’t come through in the average tables below the graph, especially when you consider that the grouping of fastballs and changes in 2007 that had vertical movement around 20 likely weren’t tracked correctly and should be shifted down (moving the 2007 average down in turn).

Still, this year, the grouping of fastballs is on more of a diagonal. The pitches without much horizontal movement are “rising” more than they did last year, but the pitches with more horizontal movement are sinking more than they did last year (well, he never actually got that kind of horizontal movement last year; rather, these pitches are sinking more than any of his fastballs did in 2007). While this sink isn’t all that severe, I do think this could be a contributing factor to the increased ground ball rate.

Here, the astute reader might be saying to him or herself, “Well if the pitches without much horizontal movement are rising, and those with a lot of horizontal movement are sinking, then the batter would have a better idea how the pitch is going to move horizontally by observing its vertical movement coming in, right? The whole notion of the wide variability decreasing predictability doesn’t really apply to Lee.” Excellent thought indeed.

Let’s examine the trajectory of these pitches, though, to see if this is really true. In the below graph, I separated his fastballs into those with horizontal movement of six inches or more and those with six inches or less.


As you see, while the thought above certainly could have made sense, it appears that the two follow an identical path to the plate (from the side) right up until the end when the additional movement is witnessed. And because it isn’t very severe at all, a batter would really need to pay attention to notice it. If you’d like to skip ahead to the fastball portion of the “Late break” section, feel free. It ties in well here.

While from the side there is no difference between the two, check out what the top looks like:


A Hardball Times Update
Goodbye for now.

Much different. Here, we see that Lee is releasing the balls that are getting the sink from a different horizontal release point, and the two don’t overlap at any point as they approach the plate. The vertical release point is identical (as seen in the side view graph), but this is something that a batter could pick up on. So while Lee is able to hide this a little, he isn’t doing it perfectly, and he may never be able to. If he shifts the horizontal release point, there’s a chance that the additional horizontal movement would go away, which seems to me would be more important than the vertical.

Overall, though, I think that the net effect of all of this is positive.

Fastball – Basic – To LHB

'07	274	72%	0.12	2.21	20%	4%	34%	21%	21%	.102	-1.29
'08	437	75%	0.23	2.70	22%	7%	32%	22%	17%	.254	-1.90

Fastball – Basic – To RHB

'07	83	73%	0.35	2.61	16%	10%	28%	25%	22%	.333	-1.66
'08	203	82%	0.41	3.16	29%	9%	22%	25%	16%	.167	-0.94

Note: runs100 was invented by John Walsh. You can read more about runs100 here. It essentially measures the numbers of runs saved per 100 pitches, so the lower, the better.

His runs100 improved from -1.29 to -1.90 against lefties (despite a worse BABIP), as we might expect given the extra movement in towards lefties. It decreased a little versus righties (even with a better BABIP), but he is getting a lot more called strikes and forcing the batter to work more to put the ball in play.

While the results aren’t drastic, I think the changes to the pitch had a very positive effect on his change-up as well. Let’s take a look.

Lee’s change-up has changed the least of all four of his pitches, but it has been more effective than it was last year. The pitch is getting a little bit more inward movement on lefties and is rising a little more than it did last year.

Aside from this, though, if you look back at the movement graph, you see the change-ups sitting all alone in 2007. This year, though, they overlap the “sinking” fastballs. This reminds me of the interaction between Tom Glavine’s fastball and change, as looked at by Josh Kalk last year, as well as Bronson Arroyo’s, as looked at yesterday by Josh.

This isn’t exactly like Glavine because some of Lee’s fastballs are off on their own, but the change-ups are completely concealed within the one sect of heaters. Let’s check out the pitch trajectories for the fastballs with at least 6 inches of horizontal movement and the change-ups.


We see that the fastball and change have identical release points and nearly identical trajectories. They are identical from the top, but the two deviate as the ball approaches the plate when we look at it from the side. This is another good time to go check out the “Late break” section, this time for the change-ups.

Overall, there is a lot going on to confuse the batter here.

What we see is that the batter is only given a little bit of information on the coming pitch in the form of horizontal release point (and overall flight path, although if you miss the release point it might be tough to decipher the two paths). If the batter doesn’t pick up on the release point, there are several things that can happen. It can be a fastball that could show a wide array of horizontal movement but coming in will exhibit identical vertical movement until very late. Or, it could be a change-up that is masked by half of the fastballs in its release point, horizontal movement, and vertical movement until very late.

Then, in the last quarter second of flight, the fastball may sink exactly as it did last year or far more than it did last year and is capable of breaking horizontally (in both directions) more than it did last year. If it’s a change-up and the batter hasn’t identified it as such yet or is off with his timing, the bottom will fall out more severely than the vast majority of his fastballs will.

To make things even more complicated, Lee has added an extra miles per hour to his fastball, cranking the difference between the fastball and the change-up to a full 10 mph. And since he uses the fastball so often (77 percent), when a change-up does come it can’t help but be a little unexpected. Oh, and all of this is ignoring that Lee also throws two other pitches.

Change-up – Basic – To LHB

CU	79	21%	0.28	1.94	15%	11%	41%	10%	23%	.267	2.81
CU	89	15%	0.09	1.58	21%	3%	36%	12%	27%	.167	-3.36

Change-up – Basic – To RHB

CU	2	2%	0.00	----	0%	0%	50%	50%	0%	----	2.25
CU	0	0%	----	----	----	----	----	----	----	----	

Lee doesn’t use the change-up against right-handers, but it has been a devastating pitch to lefties this year. According to runs100, he has saved an additional 6.17 runs per 100 pitches over last year. This can be partly attributed to a low BABIP, but that is insane. In John’s original piece on runs100, he found the best change-up of 2007 belonged to Jeff Francis at -2.4. Lee has saved nearly a full run more than that.

Slider and Curveball
Overall, Lee appears to be much more consistent with the movement on his curveball and slider this year. They are clustered closer together (possibly one reason for the improved control), and he is also getting additional horizontal movement on them.

Slider – Basic – To LHB

'07	13	3%	0.00	1.50	8%	0%	46%	15%	31%	.333	5.85
'08	21	4%	0.86	6.50	5%	29%	33%	24%	10%	.500	-1.68

Slider – Basic – To RHB

'07	9	8%	0.25	3.00	22%	11%	44%	11%	11%	1.000	6.04
'08	13	5%	0.20	2.00	15%	8%	38%	15%	23%	.333	-1.66

Curveball – Basic – To LHB

'07	13	3%	0.00	1.67	8%	0%	54%	15%	23%	.333	2.28
'08	33	6%	0.07	1.44	15%	3%	45%	9%	27%	.556	4.92

Curveball – Basic – To RHB

'07	19	17%	0.13	5.00	5%	5%	42%	37%	11%	.500	-0.09
'08	33	13%	0.44	1.77	3%	12%	27%	18%	39%	.154	-7.45

We see in these numbers that the slider has improved dramatically this year, although it’s still his least frequently used pitch. Many of his sliders looked pretty flat last year, not really exhibiting much horizontal movement. This year, I think the combination of the added movement, late break, the identical release point and similar path to the plate as the fastball and change (here if you’re interested in seeing it), and the fact that it is unexpected when it comes (being used just 4 percent of the time overall) has made it a very effective pitch for Lee. He’s getting 29 percent swinging strikes on it against lefties, and runs100 drastically improved against both types of batters.

The curveball looks good against righties in terms of runs100, though the .154 BABIP is surely helping it along. He’s allowing it to be put into play a ton against righties, so once that BABIP evens out the curve doesn’t look all that great. The movement is certainly a little better this year, so this all looks a little strange.

Release point
This added horizontal movement across the board might lead us to believe that Lee has altered his mechanics. I first thought that he might have lowered his arm slot to get that added side-to-side movement, but that’s not the case. Check out his release point graph.


Aside from the fastballs we observed earlier, it doesn’t look like Lee has changed his release point much at all. I can’t explain how he’s added all of that movement to his pitches. He’s obviously throwing them differently, but my guess would be as good as yours as to how he’s accomplishing this. Any thoughts on this matter (or on any matter, really) would be welcome.

Late break
Now let’s check out his late break. In case you didn’t get a chance to read the Cueto piece, here’s Mike Fast’s description of late break again:

The goal is to show something close to what the hitter perceives as the break or movement of the pitch. I calculate the deflection of the pitch due to two forces, spin and gravity, in the last 0.25 seconds of its trajectory before it crosses the plate, an idea I got from Tom Tango. I chose a quarter second because that’s roughly the reaction time of a batter executing a swing. I chose to include the effect of gravity because I believe that more accurately reflects what hitters see. Hitters don’t attempt to hit a gravity-less pitch; they attempt to hit a pitch that’s being affected by gravity and being deflected by spin.


As with his movement, we see that Lee has added a good deal of late break to his pitches.

We first notice that, like with the overall movement on the fastball, this pitch has a wider range for its late break as well. Extra variability overall is good, and extra variability in the last quarter of a second before the ball crosses the plate is even better.

We saw earlier that some of Lee’s fastballs have picked up some additional downward movement. When we looked at the pitch trajectory graph, though, we saw that this added vertical movement didn’t come until the ball was almost at the plate. Late break! We see in Lee’s late break graphs that he is indeed getting additional late sink on his fastball this year and that the fastballs roughly follow the diagonal pattern that they showed in the movement graph.

While it doesn’t move (in general) or break late as drastically as a true sinker might, Lee has indeed added a good deal more late sink than it had last year, and this seems to be another reason for the higher ground ball rate in 2008. I don’t know if this could cause an additional 13 percentage points on his ground ball rate, but it certainly should be adding some. Not only do batters now have to deal with the additional late horizontal break on the ball, but the bottom is also falling out more than it used to.

As we also observed in our pitch trajectory graph comparing Lee’s fastball to his change-up, we see that the change has a good deal more late vertical break than this fastball. This will become especially confusing to a batter who hasn’t yet identified that the pitch is a change-up, which is entirely possible given the fact that it has an identical release point to the fastball, the same horizontal and vertical movement, and follows an identical path to the plate (from all angles) up until this point.

In addition to the obvious benefits of this, I would have to think that this late sink is another factor driving the higher groundball rate this year.

Slider and Curveball
The slider had very little horizontal late break last year, but this year’s he’s getting not only some horizontal break but also a lot more vertical break. He’s getting just a tiny bit more horizontal and vertical late break on the curve.

So while this explains the additional strikeouts, I don’t know if it necessarily explains the control.

Let me first note that control is not the easiest thing to pin down, even using Pitch f/x. In fact, one could make a case that it is actually the most difficult to pin down. You see, control is all about intention. If a pitcher throws the baseball and it just clips the corner of the plate for a called strike, this is a good thing. But is that what the pitcher was trying to do? Who’s to say he wasn’t aiming for the complete other side of the plate and the pitch got away from him? The result was a good one, but if he wasn’t trying to do that, his control wasn’t truly good.

So while we may now be able to see the final result, we still have no idea about a pitcher’s intentions, and we never will with absolute certainty. We’d need to be inside the pitcher’s head on every single pitch. One thing I would love to see Pitch f/x track at some point is where the catcher is setting up. Is he setting up inside? Low and away? This could at least give us a rough estimate as to where the pitcher is likely aiming. Wouldn’t be perfect, but it would give us more information than we currently have.

Using the information that we do currently have—and keeping in mind that this is far, far from perfect, so take it for what it’s worth—let’s see if we can’t find anything out about Lee’s stellar control thus far.

The first column in the following table shows how often Cliff Lee hits the edge of the zone with one of his pitches. I defined the “edge” as 1/6 of the width or height of the rulebook strike zone added to either side of the given boundary (vertical boundaries mapped to average as given in this John Walsh article). The immediate outside corners were also included.

The second column gives the percentage of balls that were either on the edge (either in the zone or out) or outside the zone completely. The balls that are out of the zone but don’t hit the edge—I’m assuming for the sake of this analysis—are balls that the pitcher was trying to locate on or near the edge but that got away a little bit and traveled further outside the zone. Of course not all of them will be—some pitchers were probably actually meant to be outside to work around certain hitters or something like that—and of course there will be some that land fully inside the zone that were meant for the edge. Hopefully this will serve as a fair enough estimate without going through pitch-by-pitch.

The third column gives the number of balls that were outside the zone but didn’t hit an edge divided by these same balls plus the number that did hit an edge. This will hopefully give us a rough idea of how many balls that were meant for an edge got away from the pitcher. Again, not perfect for the same reasons mentioned above, but let’s take a look anyway.

Year	Edge%	Edge+OOZ%	OOZ/(Edge+OOZ)
2007	21%	56%		62%
2008	21%	56%		63%

We see that Lee has performed almost identically this year to last year. He hit the edge at the same frequency as last year, “attempted” to hit the edge at the same frequency, and let just about as many get away from him.

Now let’s look at the percentage of the pitches that hit the edge that were actually in the strike zone. Since pitchers really don’t have pin-point control, I would think that pitches hitting the edge would be at the mercy of chance as to whether it falls just within the strike zone or just out of it. Let’s take a look.

Year		In Zone%	Out of Zone%	
2007		56%		44%
2008		56%		44%
Lg. Avg.	47%		53%

Identical both years, though a little better than league average. Could be random chance, but a pitcher with good control might be able to do a little better than league average. Now let’s look at the balls and called strikes Lee is getting on these edge pitches.

Year		Ball%	Called Strike%	
2007		67%	33%
2008		58%	42%
Lg. Avg.	68%	32%

We see that Lee is getting more called strikes on edge pitches than he was last year and is now doing significantly better than league average. This could explain part of his good control.

If I had more trust in these numbers, I would say that Lee was due to regress here, but there are so many things to consider that I really don’t know how much weight I would really put in this. These numbers are far from perfect. Umpires are human and don’t call the exact rulebook strike zone that we are using (as John Walsh pointed out in his article cited above), the Pitch f/x data isn’t complete, the data we do have isn’t perfect, plus all the disclaimers I gave earlier about intention and everything.

Honestly, I almost took this entire section out because of all this uncertainty, but I had it written up and ultimately decided to leave it in as at least something to think about. Make of it what you will.

Here’s one more thing that I think might have something to do with Lee’s decreased walk rate this year. The following graphs show how he’s using his pitches in different counts.


While Lee obviously uses his fastball a lot, you’ll notice that he is now using it more in two-ball counts and almost exclusively in three-ball counts (though he did that last year as well). Fastballs, in general, are easier to control, and Lee’s is no exception. He gets the least number of balls using his fastballs and throws it in the zone the most often (along with the slider).

Pitch		Ball%	In Zone%
Fastball	29%	59%
Curveball	36%	52%
Slider		35%	59%
Change-up	36%	42%

Therefore, it makes sense that if he is using the fastball more on two-ball counts, he will get to three-ball counts less frequently because he isn’t getting as many called balls. And when he is using it a lot on three-ball counts, he’ll be walking fewer batters. Of course, once batters know that they can wait on a fastball in these counts, he’ll be getting fewer strikeouts and giving up more hits. To this effect, he has started using the change-up on three ball counts a little bit, but he may end up needing to start using his off-speed stuff in two-ball counts more often going forward.

Overall, I don’t know if we can really draw anything meaningful out of this discussion of control. As I said, I almost didn’t include any of this. Even if all it does is get people thinking, though, perhaps some new ideas will show up down the line.

Final outlook on Cliff Lee

Overall, Lee has looked very good this year. He’s made some nice changes to his repertoire and increased his strikeout and groundball rates. I’m not sure if the changes will allow the groundball rate to stay this high, but the increase (in general) certainly seems legitimate.

Lee is also displaying stellar control this year, although I don’t think I’m really in a position to say whether or not it’s for real. He has shown good control at times in the past, and others have kept walk rates as low as Lee’s before, so perhaps it is for real. Just based on the fact that we’re looking at a 66 inning sample and regression to the mean, I would expect it to climb at least a little.

Overall, Lee definitely appears to be a better pitcher than he was a year ago. I don’t necessarily view him as a sell high candidate, although there are certainly cases where it would be a good move. If you can get a top pitcher, do it. It’s not as if he’s putting up elite strikeout numbers. Or, if you’re dealing with someone who trusts peripherals more than surface numbers and believes that Lee’s control is for real, that might also provide a good trading opportunity. The control may very well be for real, but if I can get someone with a longer track of similar performance, I’m doing it.

Concluding thoughts

I put this note at the top of part one as well in case some readers didn’t make it this far, being that (I’m pretty sure) this is the longest article I’ve ever posted here.

I’d be interested what you guys think of the Pitch f/x stuff so far. I think it adds a lot to the analysis, but I’d love to hear any comments or suggestions you guys have. Requests are also welcome.

Furthermore, these posts seem to be running kind of long. I’d also be interested in hearing your take on this matter. I ultimately decided to split this one into two posts. Should I leave them as one going forward? Should I write less and let the numbers and charts speak more for themselves? Or is it okay the way it is? Let me know what you think.

Comments are closed.