Monday, December 17, 2012

A Possible KBO Projection Using Velocity: Richmond, Rodriguez, Eveland

In just the past couple of days, the KBO has brought in 3 new MLB/MiLB pitchers. Dana Eveland is going to Hanwha, while Scott Richmond is going to replace Ryan Sadowski in Lotte. As I was getting ready to finish this post, it was also announced that Aneury Rodriquez is signing with Samsung. As we have tried to do in the past, some kind of conversion metric between the MLB and MiLB would become increasingly helpful and we have gotten mixed results just from statistical comparisons. I had an idea and wanted to test it out and see if it would work in the transition from the MLB to the KBO (then perhaps try to test it vise versa and even, if we are feeling extremely ambitious, see if I could come up with some kind of velocity metric from the minors to Majors). Basically it goes like this, there is a good correlation between velocity and success, at least in the Majors, perhaps one could do some sort of project based on fastball velocity alone. Of course, there are many other factors that make a pitcher good/bad, but I wanted to see if velocity was a better indicator between leagues than actual statistics.

Here, we will use FIP WAA, Maximum Velocity (MV) and Median (perhaps "mid-range" is more appropriate, as we are adding the lowest number and the highest number and divide by two as they do not have average velocity listed) Velocity (AV) according to Naver KBO Data. I will look at the top 5 in innings for each team, in order to get decent sample sizes and I am using the number 5 as it gives us some semblance of a rotation. Later, I will separate the velocities into "bins", but I'll explain that when we get there. On the original lists, I will keep it in KMH (just for ease), but I will change the averages to MPH later on.


Nippert: -3.83 FIP WAA, 152 MV, 145.5 AV

Kim Sun-Woo: -.79 FIP WAA, 148 MV, 141.5 AV

Lee, Yong-Chan: .733 FIP WAA,133 MV, 127 AV

Noh Kyung-eun: .7298 FIP WAA ,152 MV, 145.5 AV

Kim Seung-hee: -.414 FIP WAA, 146 MV, 142.5 AV


Youman: 1.00 FIP WAA, 150 MV, 144.5 AV

Song Seung-jun: .17 FIP WAA, 149 MV, 140 AV

Sadowski: .311 FIP WAA, 150 MV, 144.5 AV

Lee Young-hoon: .51 FIP WAA, 146 MV, 140.5 AV

Ko Won-jun: -.83 FIP WAA, 146 MV, 139 AV


Jukich: .609 FIP WAA, 146 MV, 141.5 AV

Liz:  1.04 FIP WAA, 162 MV, 152 AV

Kim Kwang-sam: -.06 FIP WAA, 145 MV, 140.5 AV

Woo Kyu-min: .81 FIP WAA, 143 MV, 139.5 AV

Lee Seung-woo: -.814 FIP WAA, 141 MV, 137 AV


Yoon Hee-sang: .2 FIP WAA, 148 MV, 142 AV

Song Eun-beom: -.1 FIP WAA, 153 MV, 144.5 AV

Santiago: -.46 FIP WAA, 153 MV, 147 AV

Park Hee-soo: 1.41 FIP WAA, 148 MV, 141 AV

Kim Kwang-hyun: -.5292 FIP WAA, 150 MV, 142 AV


Lerew: -1.35 FIP WAA, 155 MV,  148.5 AV

Seo Jae-woong: .48 FIP WAA, 147 MV, 140 AV

Yoon Suk-Min: 1.63 FIP WAA, 153 MV, 146.5 AV

Sosa: .66 FIP WAA, 156 MV, 149 AV

Kim Jin-woo: 1.15 FIP WAA, 152 MV, 148 AV


Ryu Hyun-Jin: 2.58 FIP WAA, 153 MV, 146.5 AV

Kim Hyuk-min: .41 FIP WAA, 153 MV, 147 AV

Park Chan-ho: -1.03 FIP WAA, 149 MV, 142.5 AV

Yoo Chang-sik: -1.972 FIP WAA, 150 MV, 141.5 AV

Yang Hoon: -.899 FIP WAA, 152 MV, 145 AV


Brandon Knight: .98 FIP WAA, 151 MV, 145 AV

Van Hekken: -.2822 FIP WAA, 146 MV, 139.5 AV

Kim Young-min: -.85 FIP WAA, 154 MV, 146 AV

Kang Yoon-gu: -.86 FIP WAA, 151 MV, 144.5 AV

Han Hyun-hee: .34 FIP WAA, 148 MV, 137.5 AV


Bae Young-soo: .7 FIP WAA, 148 MV, 143.5 AV

Jang Won-Sam: 1.9 FIP WAA, 146 MV, 141.5 AV

Talbot: -.55 FIP WAA, 149 MV, 144 AV

Gordon: -.11 FIP WAA, 149 MV, 143 AV

Yoon Sung-hwan: .80 FIP WAA 147 MV, 140.5 AV

League Average "Rotation" MV is 92.535 MPH (all speed measurements are now in MPH)

League Average "Rotation" AV is 88.6135 MPH

FIP WAA of 40 rotation members (this brings the survivor bias up front, as you would expect that the best pitchers would pitch the most innings): 4.8542

So the baseline is not really zero, as we are using the top 5 innings pitchers per each team. We would expect that if we added the rest of the pitchers in the KBO, they would have about a -4.8542 FIP WAA, unless the math behind the metric is off somehow (more likely, I just miscalculated a few guys numbers, as I did test this against league averages when I originally designed it).

Team Averages (the FIP WAA is the total, sorted by MV):

 KIA: 2.57 FIP WAA, 94.612 MV, 90.768 AV

Hanwha: -.911 FIP WAA, 93.868 MV, 89.59 AV

 SK: .5208 FIP WAA, 93.248 MV, 88.846 AV

Nexen: -.6722 FIP WAA, 93 MV, 88.35 AV

 Lotte: 1.161 FIP WAA, 91.884 MV, 88.102 AV

Samsung: 2.74 FIP WAA, 91.636 MV, 88.35 AV

LG: 1.585 FIP WAA, 91.388 MV  88.102 AV

Doosan: -3.5712 FIP WAA, 90.644 MV, 87.048 AV

We sort of see what we expected here. The hardest throwing team, KIA, is the 2nd best team according to FIP WAA. Doosan, the slowest throwing team, is the worst pitching team. If we sorted it by average velocity, we would change the bottom of the order a little, but there wasn't much difference in team rank between median velocity and max velocity (which I expected). However, Lotte, Samsung, and LG were towards the bottom and were all better than average teams. The bottom 4 teams+ had a 1.9148 FIP WAA, while the top 4 teams had a 2.9394 FIP WAA. So while there isn't a great correlation, and there were obvious exceptions in a small sample size, it does seem that teams that throw harder have a better chance (or are more likely) to be better.

The team averages above were mainly to see that the pattern that harder throwers are usually better held in the KBO. I was looking to see if I could do a projection based off velocity data, so now I will separate the 40 pitchers into 4 different "bins" based on velocity, an elite bin, an above average bin, a below average bin, and a "poor" bin. The Poor and Above Average bins will have 10 players each, but just so I didn't have to break up pitchers with the same MV or come up with an arbitrary tiebreaker, the Below Average Bin will have 11 and the Elite will have 9.  First we will do it for MV (Maximum Velocity) and then AV (Median Velocity):

Poor: 82.46 MPH to 90.52 MPH range. 2.1618 FIP WAA or .21618 FIP WAA per pitcher.

Below Average: 91.14 MPH to 92.38 MPH range. 1.62 FIP WAA or .14727 FIP WAA per pitcher.

Above Average: 93 MPH to 94.24 MPH range. -3.9194 FIP WAA or - .39194 FIP WAA per pitcher

Elite: 94.86 MPH to 100.44 MPH range. 3.56 FIP WAA or  .39556 FIP WAA per pitcher

Obviously the above average section is hurt by Nippert's struggles (mainly that he pitched a lot of innings, one of the league leaders, at a below average/above replacement level). However, it is still negative if you decided to take him out.
The Elite velocity pitchers proved to the be the best, while the poor velocity pitchers proved to be the 2nd best. Here is my theory: While the best fastballs usually make the best pitchers, for a below average pitcher to continue pitching (survivor bias, as we don't see the many soft tossers that are rarely used here or just failed), they must have something else, such as great movement, control, or breaking pitches. Perhaps these are the KBO's Jered Weavers, Shaun Marcums, and Ryan Dempsters (or even, to use an extreme example, R.A. Dickeys). It is not as if pitchers with below average velocity are automatically good, it is just that the ones that survive have to show something else impressive to get over velocity bias in talent evaluators. The worst velocity pitcher, Lee Yong-Chan is a perhaps a perfect example. Let's sort by average velocity and see if we see the same results. For this one, in order to avoid arbitrary tie breakers, we put 11 in the poor bin, 9 in the below average bin, 9 in the above average bin, and 11 in the elite bin.

Poor: 78.74 MPH to 87.11 MPH. 1.8568 FIP WAA or .1688 FIP WAA per pitcher

Below Average: 87.42 MPH to 88.35 MPH. -.6162 FIP WAA or -.06847 FIP WAA per pitcher

Above Average: 88.66 MPH to 89.9 MPH. .472 FIP WAA or .05244 FIP WAA per pitcher

Elite: 90.21 MPH to 94.24 MPH. 1.7098 FIP WAA or .15544 FIP WAA per pitcher

So here, the poor velocity pitchers are actually the best. The average and below average pitchers are the worst again (seemingly validating the idea that the pitchers closest to the mean are the worst, while the slower and faster pitchers are the best), but at least it is in the right order this time. Nippert appeared in the elite this time, bringing down the results big time. If you remove him, and he was tied for the lowest velocity in the elite group, the elite group is the best group easily. Only 4 of the 11 elite pitchers in median velocity were below 0 FIP WAA and 4 of 10 elite pitchers in maximum velocity were below 0 FIP WAA.

So now, let's bring Dana Eveland and Scott Richmond into this. We will use both Fangraphs (fMPH) and Brooks Baseball (bbMPH) for average fastball velocity and Fangraphs for a max MPH. Using career for averages, just 2012 for Max MPH.

Scott Richmond: 91.89 bbMPH, 91.2 fMPH, 92 Max MPH

Dana Eveland: 90.64 bbMPH, 89.1 fMPH, 92.2 Max MPH

Richmond clearly will fall into the elite category in average velocity, while he actually falls into the below average category when it comes to maximum velocity.

Eveland is an elite velocity pitcher according to Brooks Baseball, while an above average velocity pitcher according to Fangraphs. Of course, this is not an insignificant difference and depends on which one you trust more. If you average the two, he falls under 90 MPH and into the "above average" category. Like Richmond, his max velocity falls into the "below average" category, which may introduce a classification/data difference problem. This makes actual projection difficult. However, just like in the Majors, it does seem there is a correlation when it comes to elite league velocity and success.

Let's see the effect when it comes to velocity in replacing Sadowski with Richmond for Lotte and Ryu and Park for Eveland (so we will just average the 4) for Hanwha.  

Lotte: 91.684 MV (From 5th to 6th in the KBO), 88.424 AV (Using FanGraphs' MPH. Improves them from 5th to 4th)

Hanwha: 93.575 MV (stay at 2nd place), 89.4675 AV (stay at 2nd, but like MV, the number drops slightly).

Hanwha obviously needs to use a different pitcher more to make up for losing Park (since he was terrible, he should be easy to replace, and an average pitcher replacing Park would also help make up for Ryu, as there isn't much from a statistical point of view to suggest that Eveland will be near as good as Ryu), and those innings could very likely go to their other foreign pitcher Denny Bautista, who threw 86 innings in 2012 (of course, another pitcher would have to take his innings).

Evidently, Samsung is not bringing Gordon back (so Rodriguez would replace him as one of their two foreign pitchers), so we can project how their rotation will look velocity wise.

Aneury Rodriguez: 91 fMPH, 92.9 Max MPH. Elite in average velocity and below average in max velocity, pretty much like Richmond and Eveland.

Samsung adjusted: 91.74 MV (Slightly goes up, but not in rank), 88.818 AV (This would jump them up to 3rd best in the KBO)

As far as I know, there is no real way to compare pitchers using Brooks Baseball or Fangraphs' Pitch F/X leaderboards by Maximum Velocity (the only way I could think of is to sort by average velocity, go to each individual's page, say the top 30 or so, to get the hardest thrown pitch, then compare which ranking worked better against FIP -. This would take quite a bit of time, but I may do it in the future), so it is really hard to see whether average velocity or max velocity has a better correlation to success. Just rationally, it seems to make more sense that it is more important what your velocity is on average than your best pitch, as obviously you are throwing your average fastball more than your best fastball. When scouting, people usually say that a pitcher sits at XX-XX and touches XX. Considering the data we have, I would trust the average velocity more. This would have Eveland, Rodriguez, and Richmond as elite velocity pitchers, with a 63.6 % (each) empirical probability of being above average KBO pitchers with a .46632 expected FIP WAA between the three of them. Unfortunately, for now, we have to sit and wait and see how they do in order to get more evidence and see whether this is predictive or not.

No comments:

Post a Comment