Scouting stuff: an amateur attempt to figure out pitching, part 2

So now we know about fastballs for the most part, right? No? You've forgotten already? Howsabout another Dave Allen graph:

So right, more run values. Negative is good for the pitcher, positive is bad. Outside of the whole movement/velocity thing, the big takeaway is what happens with and without the platoon advantage. The four-seamer remains a far more viable pitch to opposite handed hitters than does the two-seamer. I'm just eyeballing it, but the difference between having and not having a platoon advantage with a two-seamer is the about the same one we saw in the fastball velocity graph. In case you forgot, it's the difference between an all-star and average.

Given that, it would see we need to figure out some different sorts of pitches to offset the problems of the fastball.

Slider

From a Pitch F/x perspective, the slider is pretty cool. Many of you already know that hitters are able to recognize the slider mid-flight thanks to the light dot that shows up on the darker rest of the ball. The phenomenon is caused by the spin imparted by the pitcher which rotates the baseball like a football. The seams rotate closer to the edge of the ball and the unblemished hide stays in the middle. This kind of spin causes the ball to travel closest of any to the imaginary pitch affected only by gravity Pitch F/x compares all pitches to. The result is just a couple inches of movement vertically and/or horizontally.

While the dot can be a giveaway, the good ones are thrown too hard for the batter to actually make use of the information and change the decision to swing. That is of course unless the hitter has the platoon advantage. Here's what Brad Lidge's slider and fastball look like as a right handed batter:

If you look closely, you'll see Dave has overlaid two small black dots on the red and blue lines respectively. They represent the point in time at which the batter must decide if he will swing. Anything past this point and he won't hit anything anyway. The slider and fastball line up almost perfectly until that moment has past. But what about lefties against Lidge?

Comparatively vast difference if you ask me. And indeed, this is the going theory on why sliders in fact have a very large platoon difference and do absolutely nothing to solve the problem. Once again, the black dots are placed at approximately the point where the batter needs to make his decision to swing. Very clearly, they simply do not get to track the flight of the ball. It's all what it looks like right out of the pitcher's hand. Which means there's an exceptional premium on making everything look the same as long as possible. Hence the paranoia about tipping one's pitches. Anything too different from the fastball has to move so much that that in itself fools the hitter. The other clear difference maker, then, is expectation.

That is to say, count and randomness matters. If a pitcher sequences his pitches too obviously, he will be giving up a huge advantage. If he pitches behind the count or is otherwise forced to throw particular pitches, he's giving up a huge advantage.

Back to the slider specifically. Setting aside velocity, Dave's work has shown that the more horizontal movement away from the batter, the better the slider tends to perform. This may not hold up for sliders without the platoon advantage, but it's something I look for in the numbers in deciding the quality of the pitch.

Honestly though, I don't feel like I can look at the movement of a slider and guess it's quality. I've spent a lot of time looking and I think fastball velocity, disguise, etc. are probably far more critical variables. When it comes to sliders especially, I want to know more than just movement. And what I really want to know are the results. In particular: how often do batters swing and miss at the pitch in question? Fortunately, Texas Leaguers has a web app that accesses the Pitch F/x database and tells me what I need to know.

From there I compare those numbers to what is perhaps my favorite table currently known to mankind. Bojan Koprivica wrote a crazy good and long article that covers much of this territory, but with more of a sabermetric focus and concluded with this. I'd put it in the post itself, but it's so big that the formatting for everything goes to crap. For the record, "it" is a season's worth of pitches categorized by type and divided up by platoon advantage and results of those pitches.

Anyway the table says that with the platoon advantage batters whiff on north of 30% of sliders they choose to swing at. Without the platoon advantage, the whiff rate is 10-15* fewer percentage points. So while we've done nothing to solve the platoon problem, we have shed further light on the White Sox struggles against these random AAAA sinker/slider types. An average two-seamer and an average slider puts the RHB in a serious bind and we'd rightly expect him to do far worse than average facing that combo.

As a meaningful aside, If you actually click through and look at the table, you'll notice a sizable discrepancy between the slider whiff rate of RHB and LHB w/ the platoon advantage. Other such discrepancies exist throughout, not to mention on Dave Allen's charts and graphs. The primary causes, I would guess, are twofold.

One, lefties tend to be better hitters because they tend to also be left handed throwers. Left handed throwers can only play the outfield and first base, which means to make it in the bigs at all, they need to be really good hitters in the first place. So there's that selection bias. The other issue I'm pretty sure is that lefties tend not to face a lot of lefties throughout their careers. This means that the hitters probably aren't nearly as used to seeing lefty sliders as righties are used to seeing righty sliders. On top of that, southpaws have somewhat less velocity than RHP, which makes it all the more advantageous for managers to sit their left-handed batters against left-handed pitchers. This also explains the LOOGY fetish of many managers.

To get back to the slider as a complement to the fastball, it's all the easier to see what White Sox brass saw in Edwin Jackson. Between his elite four-seamer and his solid slider, he really did just need to start finding the zone consistently. Dominate righties and beat lefties with the big fastball and you're above average. Indeed, while Jackson could never have been worth Dan Hudson and David Holmberg, Don Cooper turned him into exactly what the Sox thought he could be.

Of course, Jackson had an outstanding four-seam fastball which makes dealing with the platoon problem a much smaller issue. Having looked at the slider now, we're still stuck with that question. We finally address that in discussing the...

Change-Up

The most common adaptation to the problem of throwing to opposite-handed batters is having a worthwhile change-up. It's often the last pitch added to a prospect's arsenal as the need for a third pitch doesn't tend to crop up until a truly advanced level of competition forces the issue. High school ain't that.

Turning to Bojan's table, we see once again a significant discrepancy between LHB and RHB when it comes to hitting the change-up. Fortunately Texas Leaguers' app also does platoon splits for whiffs, so this doesn't really put me out. RHB whiff on southpaw changes at just under 30% while lefty batsmen are at 20%. What the slider does to RHB without the platoon advantage, the change does to them with it. LHB on the other hand don't actually struggle all that much with the change, but it's still a better option than the slider in terms of whiff rate. The real difference maker is what happens on balls in play. Lefties absolutely crush RHP sliders, changes far less so.

Going back to Dave Allen's work, he's found that changes with extreme movement either up or down make for the most effective pitches. This makes me wonder if in fact what he's found is that the best change ups are just those followed by the best fastballs. Presumably, the guys who put a lot of movement on their changes are able to do much the same with their fastballs. In any case, he also made a fascinating chart comparing run value to the difference in velocity between a given pitcher's change-up and fastball:

As Dave concludes, it's very important that that change sit in that 5-12% slower window. This all gets back to what we said about the slider: it needs to look the same, but it needs to be slower and move off the barrel of the bat as it breaks. Disguise is absolutely critical and hence, the movement and velocity we can find in the Pitch F/x data can be very useful. Even so, while I feel like I have a better gauge for changes than sliders, ultimately what I look at is whiff rates when I'm checking out the numbers.

A quick note on the splitter: I really don't have any good data about it, though I imagine there are a good number of them in the data regarding change-ups since they tend to be misclassified for each other by the Pitch F/x algorithms. They do function similarly, though from what I've seen they move slightly differently. Splitters seem to have less running action and very little vertical movement. From my own experience throwing the pitch, this makes sense. Just from the grip, getting a lot of spin on the ball is difficult. You'll get even less spin if you throw it with a stiff wrist, which I believe is fairly common.

Curveball

The curve is an odd duck. The slider and change are sorta/kinda mirrors of each other. They're the out pitch either with or without the platoon advantage respectively. The curveball on the other hand is pretty rarely a pitcher's big strikeout generator. It's not an uncommon pitch, but it's typically differently used. For instance, regardless of the platoon situation, it's thrown about 10% of the time. Weird, but it makes a little more sense when you check out Dave's data.

The curve has a good amount of movement and it's the rare pitch with a lot of negative vertical movement. Meaning, it bites hard downward. Even the most downward moving sliders fail to break 3 inches, which is well below average for a curve. In addition to the vertical movement, it also breaks toward the glove side arm of the pitcher throwing it. These two components of its movement help determine its efficacy in the platoon dis- or advantage.

With the platoon advantage, the horizontal break is the biggest variable in its success. The more the pitches moves away from the batter, the better. Without, the downward bite is what matters more. This meshes with what Mr. Allen found with success of the slider and its horizontal movement. Even more interestingly, this finding meshes with what the typical curve looks like from RHP and LHP. Lefties, whom we know face very few LHB, throw an average curve with 6 inches of drop, but not quite 4 inches of horizontal cut. Righties on the other hand are at just over 5 inches for each component, more of a swiss army style pitch.

It's also worth mentioning the kinds of curves that show up. If you were looking at the distribution of curveball speeds, I'm guessing you'd find two sorts: fast and slow. The best kind is definitely the fast, the kind that makes Gavin Floyd who he is as a pitcher and has helped Josh Beckett dominate when he's on his game. But the most common variety is what I often refer to as the show-me curve. It's slower and move obvious. There's no disguise intended, but rather it's very difference from every other pitch--it's the slowest pitch in majors outside the knuckleball--is what makes it worthwhile. The show-me is an early count pitch that steals a strike without a swing. And given that it's thrown 10% of the time, a batter can't expect to see it even once per PA. Gavin, on the other hand, threw his twice as often over the last two seasons. Remember: how often a pitch is used says a lot about its quality.

Cutter

The pitch that launched Esteban Loaiza's Cy Young candidacy and made John Danks into the excellent pitcher he is today. Unfortunately, there's not a ton of data on it since it's still a fairly rare pitch. It's definitely on the rise though and it's my favorite pitch in the game. It's a true equalizer. Here's that graphic that opened the post with the cutter added back in:

You'll notice the error bars are much bigger for the cutter. That's a sample issue. Once again we see that LHB are a real bear to deal with, but that it's an excellent alternative to a two-seamer. That's potentially where it's real value lies. A pitcher without excellent velocity, but can throw a two-seamer and a cutter has the best of both worlds. Additionally, I'd wager the cutter isn't that difficult to learn to throw, as it's more of a re-imagination of the slider than an actually new pitch. If you've already got a two-seamer and a slider, you're well on your way.

Though as Dave says, it's more of a cross between a four-seamer and a slider. But I like to think of it as a slider thrown harder. You subtract horizontal movement for some gain in vertical movement and velocity. The ideal cutter has some amount of rise to it and, hopefully, little to no running action to the pitcher's throwing arm. The ideal location, if you've ever watched Johnny Danks pitch, is right under the hands.

Out of the hand, a batter reads fastball, presumably with a good amount of horizontal movement on it. The resulting pitch, however, is 5-10 inches farther inside than expected. The result, if not a whiff altogether, is very often a weak ground ball. And for pitchers trying to cut it in the big leagues, it can solve a lot of problems. More grounders from the guys with the platoon advantage? Who wouldn't want that? Especially if you're a AAAA sinker/slider RHP.

That said, if it were so easy to learn, it'd already be in more arsenals. Still, I expect the pitch to increase significantly in popularity, the unfortunate side effect of which may well be making Don Cooper less of a competitive advantage for the White Sox. Fortunately, we're not there yet, largely because adding pitches is a repetition-intensive activity. And if you're a pitcher who struggles repeating his mechanics or are already in need of fine tuning without the addition of another pitch, you simply might not have the time to waste trying to add yet another thing you're not good at.

Conclusions, Errata, Apologies

Oof. I'm already past the 2500 word mark, so it's time to wrap this sucker up. Sorry, knuckle- and forkballs. Maybe next time, Eephus pitch.

All it really comes down to is answering the question: how are you going to get guys out? For me, and perhaps this has been a point of over-emphasis, I like to break that up by platoon. It puts things in relief. First, check out the fastball. Can he get grounders? Can he throws strikes? Then check out how this guy's going to get whiffs from lefties and from righties. It's a pretty straightforward list and the end goal is projecting three things. Ground balls, strike outs and walks. If you can avoid the latter and get the former two, you're set. It's the specifics that get the word count up.