Wang: The Sultan of Sink?

River Ave. Blues and River Avenue Blues
Checking in on Scotty Bro

This is a guest post by Adam Bernfeld. He is trained as an engineer and likes to apply his analytical nature to baseball to differentiate “what seems” vs. “what is”. His interests include PITCHf/x, DIPS, the concept of clutch, and Laura Posada.

“Statistics are like women: mirrors of purest virtue and truth, or like whores to use as one pleases.”
-Theodor Billroth

It is true that in a vacuum, statistics are almost entirely useless. One can manipulate the numbers to prove any side of an argument. They can however provide illumination in instances where what seems may differ from what is. For example, a few years ago before John Dewan wrote the Fielding Bible, a lot of Yankee fans truly did believe that Derek Jeter was a good defensive shortstop because he both seemed like a good fielder, and our broadcasters told us that he was a good fielder. Similarly, watching a Yankee game, one could be led to believe that Chien-Ming Wang possesses the best sinker in all of baseball. After all, it is 94 miles-per-hour and looks like it is dropping off a cliff. Surely no one’s could be better, right? Well, using PITCHf/x data, I hope to crown the true “Sultan of Sink.”

For those new to PITCHf/x, it is a system developed by Sportsvision in use by Major League Baseball that uses two cameras to measure the position of the baseball between the pitcher’s hand and home plate, which can be used to determine various parameters about each pitch including velocity and break (for a more thorough introduction to PITCHf/x, refer here and here).

Using PITHf/x data compiled by Josh Kalk of The Hardball Times, detailed pitch information can be seen for individual players or in a searchable database. For this study, I will compare the speed, horizontal break and vertical break of the average signature pitch of baseball’s prominent sinkerballers from the 2007 season. Based on reputation and ground ball rates, I have chosen a test group that includes: Chien-Ming Wang, Fausto Carmona, Aaron Cook, Roy Halladay, Tim Hudson, Derek Lowe, Sergio Mitre, Brandon Webb and Jake Westbrook.

Note in the chart above that the values for break represent the number of inches that the pitch moves compared to a pitch thrown without spin, or, compared to a straight pitch under only the influence of gravity. The negative numbers for horizontal break indicate that the pitch moves inside to right-handed hitters. The positive values for vertical break indicate that the pitch crosses the plate higher than a pitch without spin would have. This may seem counterintuitive as we are talking about sinkers, but in reality sinkers are thrown with backspin and do rise (resist dropping actually), albeit much less than a 4-seam fastball, so they do in fact appear to sink. A 4-seam fastball is gripped across the seams resulting in more backspin than a sinker (or 2-seam fastball) which is gripped along the seams; the more backspin, the larger the positive vertical break. For comparison, Joba Chamberlain‘s fastball has a vertical break of 11.42 inches, compared to the average vertical break of 4.66 for our sinkerballer group.

The chart above can be shown graphically by plotting horizontal break on the x-axis and vertical break on the y-axis with pitch speed represented by dot size. In this graph, the lower right hand corner, the point (0,0), represents a pitch thrown without spin.

In hopes to build some sort of consensus as to who possesses the best sinker, I used the three PITCHf/x values (speed, horizontal break, vertical break) to create a z-score (also called a “standard score”) which is a statistical quantity used to combine multiple values measured on different scales. I then ordered the average z-score of each pitcher from high to low to rank the pitcher with the best combination of speed and break. Think of this as an index of the “stuff” on the pitcher’s sinker, or a “sinker-stuff index”. This value appears on the chart below alongside each pitcher’s groundball to flyball ratio (GB/FB), swing and miss (S&M) percentage, fair ball (FB) hit percentage, and fair ball extra base hit (XBH) percentage. All of these values are for sinkers only, except GB/FB which is the rate for all pitches thrown regardless of type.

The results of this study are interesting. While throwing one of the harder sinkers in the league (second only to Fausto Carmona), Chien-Ming Wang’s sinker actually sinks less than any other member of the test group. This is certainly a surprising observation. In fact, his sinker also has the lowest horizontal break, causing his sinker to rank 8th out of 9 in my sinker-stuff index. This certainly manifests itself in his below average groundball to flyball ratio and sinker swing and miss percentage. Remarkably however, compared to the test group Wang has the lowest percentage of hits off of sinkers, and by far the lowest percentage of extra base hits off of sinkers.

While some may chalk these low hit percentages up to luck, it appears that something deeper may be happening here. Finally though, the stats are pointing towards something that we already believed to be true. What seems agrees with what is, and that is the fact that Chien-Ming Wang rarely allows a hard hit ball off of his sinker. Why then, is his sinker so successful in spite of the fact that it moves so much less than those of his peers? Maybe my methods are all wrong (I hope not). Maybe pitch speed is more significant than pitch break (possibly). Maybe the ease of his delivery, which also includes a pronounced, varied hesitation, has a great effect on hitters’ timing (possible). Maybe Wang’s defense is better at turning batted balls into outs than the others on this list (unlikely). Maybe the DIPS people are right and hit percentages are no more than statistical variations (possibly, although I believe that DIPS applies more loosely to sinkerballers). Maybe it is a combination of all of these factors or even something that I have not thought of (likely), so please feel free to throw out explanations of your own.

Regardless of the reasons, I am glad that Wang is a Yankee as he gives us a very good chance to win once every five days. While his low ranking in my sinker index does not correlate with his outstanding results, at the end of the day I’ll always take the results over the “stuff”. Though after much machination, I have not definitively proved where Wang’s sinker ranks amongst his brethren, but I hope that this was interesting and informative, and maybe allows you to view the game and the sinker in a different light the next time the Wanger takes the hill.

River Ave. Blues and River Avenue Blues
Checking in on Scotty Bro
  • Ryan

    My completely intuitive guess is that sinker velocity has a tremendous impact. I’m sure that you could figure that out if you ran a multivariate regression with the “stuff-index,” velocity, and the various outcome statistics.

    In your next post, can you statistically explain why David Eckstein has succeeded as a major leaguer?

    • Adam

      ryan, i disagree with your premise, i am not entirely convinced that david eckstein has succeeded as a major leaguer.

      • tommiesmithjohncarlos

        Adam – Ask any Met fan if he’d like to build a time machine to go back and convince Omar Minaya to sign Eckstein instead of Luis Castillo, and you’ll have your answer.

        • Adam

          point conceded, but wouldn’t a time machine be better put to use on not trading scott kazmir?

          • tommiesmithjohncarlos

            What are you talking about? Zambrano’s about to hit his stride, I can feel it!

          • Count Zero

            Wow! Great thread…I can’t wait for the next installment! :-)

  • My Pet Goat

    It’s my understanding that PITCHf/x measures the break from release point through the strike zone or catcher’s mit, or whatever termination point is universal… I suppose my question is one of ‘late’ break. Is this a measurable quality given the information provided? It seems that an unexpected late action break would give a hitter the least amount of time to properly track and drive a ball, perhaps explaining Wang’s success. Or perhaps the myth of the late break is a product of velocity. Either way, I enjoy having him on the team because of his outlier-statistical status. Simply put, he shouldn’t be this good; known metrics simply don’t support the results. But where there’s smoke…

    • Glen L

      I’d like to hear more about this as well … how does Pitch F/X account for and/or differentiate a pitch’s “late break?”

      • Adam

        M.P.G., i agree with you, late break could indeed have a potentially enormous effect on the effectiveness of a sinker. as far as i can tell, there is some debate on the exact definition of “late break”. to a physicist, it may seem that a pitch undergoes constant break from the point of release to contact, and it is possible that “late break” is no more than a perception of the batter. tom tango (aka tangotiger) has hypothesized that late break could be defined as the “amount of break in the last 0.25 seconds of a pitch, prior to crossing the plate.”

        the concept of late break is discussed in this thread, and related to pitch f/x in chapter 5 of this article.

        i guess the point here is that if late break is real, than there still is no perfect way to isolate it yet with pitch f/x.

        • My Pet Goat

          Adam, this is in reference to your comment that physicists would argue against late-break. I won’t pretend to have a grasp on the physics, but I can’t imagine that a pitch begins to break upon release. The break of a pitch shouldn’t kick-in until sufficient drag on the seams causes the ball to ‘tumble’ out of its initial trajectory. A pitch loses both vertical and horizontal velocity as it travels. There must be some threshold of deceleration that kick-starts the movement. And higher speeds shouldn’t diminish this effect (as one could suppose that the shorter travel time allows for less ‘tumble’ time). If I understand correctly, as velocity increases drag increases exponentially. A doubling of velocity quadruples resistance (or some other such calculation. I heard Click and Clack talk about this when discussing fuel efficiency at higher speeds).

          • Adam

            your response much more mathematically sound than my casual casting aside of “late break”, but here is why i think that the effect of late break is over stated:

            if home plate is 60 feet 6 inches from the pitchers mound, is it purely coincidence that this sudden “threshold of deceleration” happens to occur somewhere during the last quarter of a second before the ball crosses home plate? and if home plate were 70 feet away instead of 60, would this break continue to the point where breaking pitches could not be consistently thrown for strikes? it all seems far to chance to me. i think a scenario easier for me to understand would be if the batter subconsciously extrapolates the initial trajectory of any pitch, and any variation to that movement seems to them to be late break. after all, a 90 mile per hour fastball gets to home plate in about 0.46 seconds, so to me at least, any break would seem late.

      • Adam

        My Pet Goat and Glen L: i am not ignoring you, i have replied to your “late break” comments, but it is lost somewhere in the interweb. if it doesn’t come up in a little while i will try again.

    • Jon

      Late break is my “guess” as well.
      Definitely a combination of delivery, velocity and late break are what make Wang great.

      Now let’s just hope he gets to, and performs in the playoffs this year!

  • barry

    His success could have something do with release point or deception of delivery.

  • Dan

    Great article. My biggest feeling as to why Wang is having as much success this year is because he’s become more of a ‘pitcher,’ for lack of a better term. With the majority of the pitchers on your list, their primary pitch is a sinker. However, this year, Wang has rediscovered the slider that got him signed in the first place, as well as improved the location of his four seamer. From the perspective of a hitter, they now have three things to worry about when they face Wang – the heavy sinker, a sharp slider, or a straight fastball. Wang’s sinker and slider are perfect complements – the sinker breaks in to righties, while the slider breaks away. Since Wang has a seamless motion, the ball looks the same when coming out of his hands, making the distinction between these two pitches even harder. As a result, Wang is generating more swings and misses with his slider, and has become just as effective in terms of his Game plan. Now, as to why hitters never make great contact against Wang’s sinker. After watching all of Wang’s starts, it seems that he likes to jump ahead with the straight fastball and slider. As the count progresses, or if he is in a situation where he needs a ground ball, Wang just throws an effective sinker inside. Since the hitter is more likely to take a defensive swing later in the count, a weak hit ball is a common result.

    • Brian

      Dan, your analysis definitely seems sound for this year, but to clarify, Adam’s post uses only data from before this season.

  • JT

    Can you post the formula used to calculate your sinker-stuff index? Unless I’m misinterpreting the data, Wang has the greatest vertical break. How could he have second lowest score in your stuff index?

    Also, what is a negative horizontal break? I’m assuming that for a right handed pitcher, the negative represents a break to the right.

    • Brian

      It is counter-intuitive because we know a “sinker” should break down, but as the post describes, pitches with backspin actually resist falling. The amount of vertical break is compared to a pitch thrown without any spin, that would fall due to gravity. More backspin means less falling (simplistic explanation). As a result, fastballs break “up”, with sinkers breaking up less than 4-seamers. Since Wang’s ball has the greatest amount of vertical break, it means is “resists falling” more than all of the other sinkers analyzed.

      You are correct about negative horizontal break, as Adam said, “The negative numbers for horizontal break indicate that the pitch moves inside to right-handed hitters.” The handedness of the pitcher does not matter in this case.

  • A.D.

    It could come down to Just enough sink, sure Wang statistically might not have as much break, but that can also be an asset, sometimes you see a sinkerballer have trouble throwing strikes if theyre is too much sink, thus Wang can be successful throwing the ball down, have it sink, and still be/look like a strike where other sinkerballers have pitches that consistently sink out of the zone so they have to start them higher, or hope for chasers

  • tommiesmithjohncarlos

    To sum it up, in an inappropriate way:

    Wang is so hard that his vertical pitch doesn’t break.


  • Bo

    My head hurt after the first 4 words.

  • Mark

    Does this pitching system rate control of each pitcher, since that could affect the hitting statistics. FB%

    • Adam

      if by control you mean “how well the pitcher is able to locate his pitches”, then no, that is not tracked because it is almost impossible to tell where the pitcher was aiming, although having good control could certianly be a factor in wang’s favor. if you mean where the pitches were located, then yes, that is tracked by pitch f/x, however i left it out of this analysis for the sake of simplicity.

  • Josh

    The assumption seems to be that speed * break is the measure of the “stuff” on a pitch. The problem is that “stuff” is relative compared to the other pitches the pitcher can throw. Joba’s slider is so devastating because he can throw his fastball at 97 for strikes. I confess that I don’t know the arsenal of many of these pitchers, so I can’t make similar arguments for the others. I think statistically this a much more complex problem than trying to look at a single pitch in isolation when you are using dependent measures that are reflective of the overall effectiveness of the pitcher.

    • Adam

      right josh, stuff is a term used more to describe the arsenal of a pitcher, than any individual pitch. maybe my “stuff index” is poorly named, although i thought that it got the point across. each of these pitchers are however primarily sinkerballers, with about 65-75% of their pitches thrown for sinkers. in this article i was merely trying to describe who’s sinker had the best combination of speed and break, and in the second table, with the exception of GB/FB, all of the percentages are for sinkers alone. i think in 1000 words i would be being overly ambitious to quantitate the “overall effectiveness of the pitcher”, so instead i focused on a thought exercise involving only one pitch.

  • Bronx Cheer

    Wang is a big guy who throws down hill, thus having a very high release point and making his pitches have a sharper natural downward plane (w/o any sink) than one thrown by an average height pitcher. Think this has anything to do with the numbers? I know Carmona is a fairly tall guy, as is Westbrook. What about the others? If we factor height into the S-S index, does that help things shake out a little more intuitively?

    • My Pet Goat

      Halladay is 6’6″. Lowe is 6’6″. Webb is 6’2″. Cook is 6’3″. Hudson is the smallest at 6’0″. While serious height is a common trait, it certainly can’t be a distinguishing factor.

  • Geno

    He’s an ace, baby! He’s who we build our pitching staff around. Dude’s 28 and in his prime. Now if we could just get one of the “Big 3” to emerge as a front-line starter, I’d be able to relax. Seriously, just think of what this rotation will look like when Hughes & Kennedy get their acts together. We’ll be like a runaway freight train – no stopping us.

  • Dan H

    I could be totally wrong on this, but is it possible that you didn’t account correctly for the difference between vertical break and horizontal break? It is much harder for a hitter to adjust to vertical break than horizontal break (as is evidenced by everyone’s desire to throw a 12-6 curveball as opposed to an 11-5). I could be reading this completely wrong but it seems to me that you said Wang has the best vertical break of anyone on the list. If that is true, then it might explain why his sinker is harder to hit than some of the other guys you have ranked above him.

    • Adam

      this was a point that i tried to get across, but the nature of the sinker makes it difficult, but wang actually has the worst vertical break on that list. the convention of pitch f/x is that positive vertical break values denote pitches that end up higher in the zone than a pitch thrown without spin (under only the influence of gravity). a sinker actually fits into this category, because its backspin prevents it from breaking downward like a curveball which has topspin. for example, a curveball would have a vertical break of -10, compared to wang’s sinker which is 6.6, further compared to joba’s 4-seam fastball which is 11.4. everyone on this list has a lower vertical break value than wang, thus their balls “sink” more. incidentally, he has the lowest horizontal break as well, which brings us back to the conundrum of this study. while his sinker is very fast, it has less horizontal and vertical break than anyone else of the study group, but still manages the lowest rates of hits and extra-base hits on sinkers put in play.

      as far as your point on hitters adjusting better to horizontal break than vertical break, i think mariano rivera and his cutter would beg to differ.

      i hope this clears up the confusion some.

  • Chip

    Alright, as a physicist/engineer, I feel like I can shed some light on this. It is physically impossible for two pitches thrown at the same speed and with the same spin would break at the same time. The faster the pitch, the later it would appear to break to a batter because before that it breaks so little that it is nearly undetectable.

    Let’s put it this way, a ball pitched essentially has 3 forces working on it (there are more but they’re too small to worry about), those forces are gravite, the initial force from the throw and the force from the spin (Bernoulli’s principle). Gravity is constant (well almost) so let’s throw that out. The force from the spin I would say remains fairly constant so that leaves the “speed” of the pitch that we see. Now fastballs lose the most velocity (due to air resistance) on the way to the plate. The faster is starts out, the more speed it loses (because resistance works as a ratio rather than a constant force in this example). So a pitch that starts out at 96 mph probably ends up at about 88 where one that starts at 88 might end up at 84 (these are guesses). Well if an equivalent force is applied with spin to both pitches, the 88 mph pitch will “break” more because the force will have more time to act on the ball before the catcher catches it. To a batter, the first pitch is probably going so fast that he can’t see the smaller movement it would have say in the first 30 feet. So even though it moves less, the batter has much less time to react and adjust to it and may miss the ball by an extra maybe .1 cm which is the difference between a double to the gap and a ground ball to short.

    So there we go, faster pitches appear to break later, it’s an optical illusion created by the faster ball. So is this why Wang is so effective? I don’t think so, I think it’s about location. That’s the name of the game in pitching. I don’t care how good your sinker is if you leave it up it will get hit

    • Chip

      *It is physically impossible for two pitches thrown at the same speed and with the same spin to break at different times.*

      I need to think before I type