Saturday, November 30, 2013

Choi Hyang-nam Scouting Report

According to reports out of Korea, Choi Hyang-nam of the KIA Tigers is aiming to come to the United States to play baseball. It wouldn't be the first time the right-handed reliever came to the States trying to pitch in the Majors, as he was famously posted for 101 dollars and pitched for the Dodgers AAA in 2009-2010 and Cleveland's AAA in 2006. He never made it to the Majors, but the now 42 year old (will turn 43 in Spring Training) had a very good year in AAA in 2006 as both a starter and a reliever, followed by a solid year in 2009 in relief, before struggling in 24.2 innings (despite not giving up a homer) in 2010.

In the Korea Baseball Organization, Choi has pitched for Haitai (who eventually turned into KIA in the KBO), LG, KIA, Lotte, and then KIA again after returning from the States from a second time. In 2013, he only threw 30.1 innings, and was above replacement in all three metrics, with FIP favoring him the most and ERA favoring him the least (meaning his strikeout/walks/homers was better than his runs per game average, and he allowed just one "unearned" run). He was less than half a run below KBO average according to runs per game, but that is also not adjusted for relievers, so he was most likely a little worse than that. In 2012, his numbers were a little better, but he threw just over 20 innings. The lack of innings and his age severely hurts his value. However, he was pretty efficient, getting an out every 5.52 pitches, about the same as Lee Jae-Hak

Bizarrely, Choi doesn't have a Naver Scouting Report. Here is a look at his delivery:

I couldn't find any high quality video of his time in AAA, but this is a sideview of what he looks like before releasing the ball


His finishing point in his delivery reminds me a lot of what you see in some Dominican pitchers or sinker ballers, landing hard on the front leg and firing his body forward, shoulders square to the plate. Alexi Ogando's delivery is an example, without as much movement at the first of delivery. He has a high release point for his height, and the overall motion is unique enough that it seems deceptive. At times, the motion looks like he isn't even trying

Here are the pitches I saw him throw in the 2013 season:
81-83 some sink, a little arm side movement
78-79 slider, mostly vertical
77-78 change, looks like the sinker with just a little bit taken off.

Choi struggled with everything staying up when I saw him. I saw him get his fastball up to about 85-86 in 2012. I did not see this velocity in 2013 outings.I also saw a curveball at about 72-74 in 2012 video, but did not see it in 2013 outings.

Ideally, Choi takes advantage of over aggressive hitters by slightly changing speeds and throwing pitches both in and out of the zone. It maybe a skill set that will still work for him in AAA, though returning to a hitter friendly park in the PCL like in Albuquerque would most likely lead to a lot of homers because of diminished stuff. If he didn't have the stuff for MLB teams (specifically Cleveland in 2006) in the past, he certainly doesn't now.

Thursday, November 28, 2013

AFL Pitch F/X Data: Amador and Rua

Eventually what you want to be able to do with Pitch F/X data is to evaluate players. I have been looking at Pitch F/X data in the 2013 Arizona Fall League from several different perspectives, but have yet to look at isolated individual players. In this post, I wanted to look at a couple of notable hitters, not highly rated traditional prospects, but a couple of players I think are interesting, and see what the Pitch F/X data says about their tendencies during the AFL season (and perhaps, their offensive game as a whole).

Japhet Amador is 26 year old former Mexican League slugging 1st baseman signed by the Astros late in the 2013 season. Amador is most known for his weight, but he had some impressive seasons in Mexico. He
saw 160 pitches measured by Pitch F/X in the Arizona Fall League

His 32 swinging strikes is twice as many swinging strikes as you would want (at least compared to MLB league average), and that is going to be a serious problem if he is going to be a MLB player, so let's take a look at the pitches he swung and missed at

Amador is a right-handed batter, but their appear to be plenty of whiffs on both sides of the plate for him. There were also several high inside fastballs that Amador swung and missed at. So perhaps he is struggling with velocity, but let's continue to look at locations and start to focus on the good things he did in the AFL measured by Pitch F/X. Here are his 4 home runs.

 Up and in is also where Amador hit a couple of home runs, as well as missing some pitches. It is worth noting that Amador appeared on the list of hitters that hit a homer in the same at-bat they had a swinging strike. It would seem that pitchers could throw it there effectively at times, but Amador could strike there successfully. How about all of Amador's batted balls?

Amador was able to put balls in play when they were inside, and preferably up. Pitches away, especially low and away, were almost never put in play by Amador. To illustrate the differences, let's look at his average locations.

While his average home run was on pitches that were harder than the average pitch, just as we saw in the home run graph, his average whiff was on pitches harder than the ones he put in play. His whiffs were much lower as well, and he hit balls inside better than outside. He was pitched slightly inside on average, something I have chalked up to inexperience and lack of control in previous AFL posts, and if he played in the Majors, we would expect to see more pitches outside. This could really hurt Amador's chances of hitting, as he doesn't seem to hit these pitches. Checking on whether or not he struggled with velocity, I checked the 4-seam fastballs he saw, all 90 by Pitch F/X, and saw that the average one he swung and missed at averaged 93.98 MPH, while the ones he put in play were 90.67 MPH, making it seem that he was dramatically affected by velocity in fastballs. The four seamers that he fouled off were 92.69 MPH on average, and the called strikes were 91.54 MPH, while the balls were 92.15 MPH on average.

Moving on from Amador, I wanted to take a look at someone who had a home run explosion in the lower minors (32 homers in all in 2013) for the Rangers system, Ryan Rua. Rua saw 228 pitches measured by Pitch F/X in the Arizona Fall League, 28 of them for swinging strikes. Here are were they were located, labelled with MLBAM tags:

Rua is also a right-handed hitter, so the low breaking balls to the right are actually inside, which is strange. While there are some low balls on both sides, and two pitches higher than the strike zone, almost all of these pitches are actually in the strike zone. Like Amador, Rua hit four home runs that were measured by Pitch F/X. This is what they looked like:

All four are obviously in the strike zone, and are roughly middle height, three of them inside, one outside. Let's look at all of Rua's contact plays in context of location and the opposing pitcher's release point.

Rua seems pretty balanced at making contact on pitches both inside and outside, even making contact on a pitch outside of the strike zone, as well as a pitch low and in, not in the zone. He also made contact on pitches in the normal right-handed pitcher cluster. Interestingly, it didn't seem like he made contact with a lot of pitchers with high release points, though the average AFL release point was lower than the average MLB release point in 2013.

Finally, here are his average locations to give us a better idea of his hitting tendencies in the AFL

Obviously the homers are higher on average, but interestingly they were further away on average as well. His contact came on pitches slightly closer to him than the average pitch, while his whiffs generally came low. While his home runs were on harder pitches than average, both his contact and whiffs were on slightly slower pitches, with his whiffs being slower on average. It would seem that he can hit for power if you get the pitch up and over the plate, while you can still throw inside on him, low preferably, and get him to swing and miss.

Wednesday, November 27, 2013

AFL Pitch F/X Data: Sequencing

Taking a break from looking at batting results, I wanted to take a quick look at another subject that has been very important in recent Pitch F/X research, pitch sequencing.

I think a good way to start is to approach sequencing the same way I did, looking to at-bats that contained both a whiff and a home run, meaning that the pitcher was able to make the batter swing and miss on one pitch, only to give up a home run later in the at-bat. There were 6 such at-bats in the AFL Pitch F/X games:


4 of the 6 at-bats that had swinging strikes and home runs ended with batters getting a different pitch than the ones that he swung and missed at. As far as general location goes, we see the same ratio, two of the six got pitches in the same area. Japhet Amador got the same pitch in a similar area and hit it for a homer, but the difference was that the first one was out of the strike zone. O'Brien got the same pitch in the same area against Pete Ruiz, both in the strike zone, and he didn't miss the second one. 

The Peter O'Brien/Dominic Leone at-bat is fascinating because Leone managed to get O'Brien to swing and miss at two different pitches, one out of the zone and one in the zone, and then threw a third separate pitch in the strike zone, in a different part of the strike zone, and it was hit for a homer. Here is a location graph showing the three pitches:

As the first chart shows, the cutter high and in the middle of the strike zone was the ball hit for a homer. The fastball thrown slightly away was missed at, and the slider was low away, not technically in the strike zone, but close. 

Of course, the at-bat actually lasted 8 pitches, so the three pitches above were just a small, though dramatic part, of the battle at the plate. Here are where all 8 pitches were located, labelled with the MLBAM tags of the pitches, along with the amount of strikes there were at the time the pitch was thrown:

Leone threw fastballs early in the count, both away, one high and one low. After getting a strike on O'Brien, Leone threw the slider that got him a whiff to get the count to two strikes. O'Brien then bore down and fouled off four pitches, a slider and cutter away (perhaps both sliders), and a fastball and cutter high and slightly inside. This is when Leone kept the ball over the plate and O'Brien hit it for a homer.

Of course, we want to look at larger sample sizes, and look at more results than just home run at-bats that also contained whiffs. So here we will take a look at strikeout at-bats, breaking them down by what pitches were thrown by the amount of strikes in the count.

Here are all the pitches in Arizona Fall League Pitch F/X games in 2013 that were either called a strike or swung and missed at in at-bats that ended with a strikeout. First, here are the pitches that were thrown with no strikes:

To spoil the two below graphs, while there are some sliders on the fringe here, we are mainly looking at pitches in the zone, with a stronger concentration. Pitchers are using their fastballs to get ahead in counts fro the most part. 

With 1 strike:

Here we see a lot more curves, with some more changes and sliders as well, we also see a lot more pitches off the plate horizontally, especially on the left side. Middle of the plate and high out of the zone seems to have a higher glut of pitches as well. 

With 2 strikes:

There are more pitches everywhere, whether buried in the dirt, extremely high in the zone, or to the left (still not a great glut of pitches to the right, which is where you should throw your pitches to a right-handed hitter, but again, this could be youthful pitchers). There are also more changeups in the zone.

In case the graphs do not show it, here are the average locations in strikeout at-bats, broken down by how many strikes there were at the time the pitch was thrown:


Not surprisingly, when pitchers in the AFL got to two strikes, they tended to throw more breaking balls, which are lower in the zone on average. Surprisingly, there isn't much of a difference in horizontal location.

Tuesday, November 26, 2013

AFL Pitch F/X Data: Individual Batting Results

In the previous post, I looked at general batting results and locations. In this post, I wanted to look at more individualized results, seeing if the data will tell us something about each individual batter.

First, let's look at how each batter was pitched by average locations:

I am surprised at the concentration of pitches and where they are located. Right-handers were pitched inside on average, which is surprising. However, this matches the average location data we saw with the pitchers, so this could just be right-handers struggling to get the ball glove side. This seems to a mark of a struggling or young pitcher, at least that is what other cases suggest. The White Sox' Micah Johnson, and Alex Dickerson, just acquired by the Padres were pitched extremely low, while Tyler Austin, a top prospect with the Yankees, was pitched extremely inside.

Now that we have seen where the batters were pitched, let's look at the average locations of their whiffs:

As you can see, the graph has shifted to the middle of the plate on a whole. This suggests that the batters were pitched inside on average, but were more likely to swing and miss at pitches away, something we saw in the previous post. Nick Ahmed, the dazzling defensive shortstop with the Diamondbacks, swung at missed at a lot of pitches away, while Johnson and the Phillies' minor leaguer Cameron Perkins struggled with balls low. Wes Darvill of the Cubs is a left-handed hitter, so he also struggled with pitches low and away. The Angels' Jett Bandy bucked the trend by struggling with high pitches.

With this in mind, I wanted to look at the worst pitches that batters swung at, that is, the pitches furthest out of the zone, starting with the highest pitch swung at:

This is not quite a full foot out of the zone. Lamb was a 6th round pick by the Diamondbacks in 2012 out of the University of Washington. He is a 3B/DH that has only played in the lower minors so far, but he has a manageable K/BB and has hit for a high slash line all the way around. AA may be a test for him, especially if he is consistently this impatient, but the numbers are good so far.

Lowest:

Alfaro is one of the Rangers' best prospects, and he has a reputation for being quite the free swinger, here swinging at a pitch that bounced before it reached the plate.

Furthest to the Right (catcher's perspective):

Parker has posted good numbers throughout the minors, including AA, since being drafted out of the first round by the Rockies after a football career. However, there does seem to be some concern about swing and miss, and here the right-hander swung at a pitch a foot out of the zone away.

Furthest to the Left:

 Gift is interesting because he was signed out of South Africa, but the numbers haven't been very good, and he really struggled in AA in 2012. Here, the switch hitter swung at a pitch that was a foot off the plate, and in the dirt, which I think is indisputably the worst swing of the AFL Pitch F/X games.

Of course, looking at some positive things can be helpful as well, and we looked at homers in the previous post, so let's look at the other two extra base hits, starting with all the doubles in Pitch F/X games in the AFL, marked by location and labelled with the hitter of the double:

Cristhian Adames, a 22 year old with the Rockies who struggled in 2013, hit the highest double, while Robby Hefflinger, a former 7th round pick by the Braves, hit a double on the lowest pitch. Japhet Amador and 2012 Cubs 1st round pick Albert Almora hit doubles on pitches that were up and in and balls. Kyle Parker swung at a bad ball in an above graph, but he also hit a double on a borderline strike on this graph.

Triples are much more rare, so this graph is much cleaner, but there are no repeats, so it is difficult to guage any kind of predictability:

Other than Spangenberg, who hit a low and in pitch for a triple, the triples came on pitches in the middle part of the plate or high. It is interesting to see Alfaro here, hitting a high ball for a triple despite being a catcher. Austin Hedges is also a catcher, and Kris Bryant is hardly known for his speed, making triples as "doubles with speed" a misnomer.

Monday, November 25, 2013

AFL Pitch F/X Data: Battting Results

Moving along from just looking at a pitching perspective when it comes to Pitch F/X data in the Arizona Fall League, I wanted to take a basic look at the batting results in the Pitch F/X game from a hitter's perspective. In the next post, I plan to take a look at the results from a more individualist standpoint, but for now, I'll look at results as a whole. The simplest way to start this seems to be by looking at all the swinging strikes in the AFL Pitch F/X games, labelled by MLBAM pitch type:

There isn't a lot of surprising things here, fastballs are more likely to be swung and missed at high, curves and changeups low, and sliders low and side to side. 2-seam fastballs (FT) are usually low or in the middle of the zone when they are swung at. There are some exceptions of course, as I see a couple of high sliders, a high cutter, and a high changeup on both sides.

When looking at home runs, the pitches are high in the zone, with really only one "low pitch" hit for a home run.


Obviously there is a lot of fastballs, but there is a hanging high curve, some sliders that didn't get real low, and a hanging changeup. Below is the same graph, but labelled with the names of the hitters instead of the pitch types:

Peter O'Brien, a Yankees C/3B prospect in the lower minors, managed to hit a homer that wasn't even a strike, while Ohlman had the low home run. The highly paid Cubs' prospect Jorge Soler had the home run on the highest pitch, while Mexican Leaguer turned Astro Japhet Amador had an honorable mention (as did Peter O'Brien).

Infield fly balls is something that I think has been getting too much attention in the sabermetric community in the last couple of years, but it is something that is interesting since they are almost always turned into outs. GameDay labels them "popups", and here are the locations and the pitch types of each one measured by Pitch F/X in the AFL in 2013.

 Most of them are on fastballs thrown up and in to right-handed batters, which I don't think is surprising. There appear to be some breaking balls thrown away from right-handers that turned into popups, most likely pitches that the batter tried to pull (you wouldn't think a left-hander would pop up a low and inside breaking ball).

Groundballs are much easier to measure, and usually more predictable as a repeatable skill, but the problem is that there are more of them. This makes the graph a little messier, but here are the groundballs turned into outs in the AFL by Pitch F/X data:


There aren't many high pitches, but the concentration of pitches is mostly in the middle of the plate. Not surprisingly, a lot of cutters and 2-seam fastballs show up here, without the concentration of breaking balls that we saw in the swinging strike graph.

Here are the average locations of the 4 main GameDay results:

On average, there is no real average difference in location between outs, no outs, or run scoring plays. This suggest that most of the actual results off the bat were BABIP or randomness driven. Swinging strikes on the other hand, were thrown lower in the zone and a little further away from right-handed hitters.


Sunday, November 24, 2013

AFL Pitch F/X Data: Release Point Data

As I continue to look at the Pitch F/X data from the Arizona Fall League, in this post I'll look at the release point data from the AFL. First, here is a graph showing each pitch thrown in Pitch F/X games in the Arizona Fall League regardless of pitcher:


It would be nice to have labels for those pitches, so here are the average release points for all 110 pitchers that pitched in Pitch F/X games in the Arizona Fall League.

Even the above graph isn't all that helpful without some context, so below is a comparison of each of the AFL pitchers' average release points to MLB pitchers in 2013 based on the release point leaderboard (sorted by height of release point of the AFL pitcher). Some comparisons are better than others obviously, and it has nothing to do with stuff or any other part of pitching, just average release point:


Of course, like with pitch movement, release point data in Pitch F/X is subject to park factors, as it is measured slightly different from park to park. The AFL parks showed significant bias in movement data, so we would expect to see the same in release point data. So using the same method, we can see if there is biases from the individual pitchers that pitched in both the AFL and the MLB this season.

For all three right-handers, we see basically the same height, but the AFL data brings them closer to the center of the rubber.  For the lefties, 2 of them are brought further away from the center of the rubber, in the same direction, while Vidal Nuno stands out as an exception. Of course, if just one of six are different, it could have been a conscious decision by Nuno to slightly move on the rubber. The difference doesn't seem to be drastic, but there is probably some shift to the right (catcher's perspective) caused by the AFL Pitch F/X measurements. 

Saturday, November 23, 2013

2013 KBO Position Player WAR Leaderboard

In case you missed it, I posted the pitcher WAR leaderboard for the 2013 Korea Baseball Organization here. This post is the position player version of the 2013 KBO WAR leaderboard. Before the list below, some quick notes on methodology.

For ease, I kept the same simple speed score method from last time. This time, I did note that league average simple speed score was 5.96, so I began by grading harder. So to keep it somewhat simple, I calculated the first method of speed score for each player, then subtracted it by .96. Then, to take account of playing time like we did with batting (I didn't do this last year), I assumed that a "full time player" had about 300 plate appearances (70 had that much in the KBO). So while I kept the run scale of doubling speed score from 5 to equal 1 run, I assumed it equaled 1 run over 300 plate appearances. So I used the percentage of plate appearances over or under to adjust the run value. But before I did that, I had to give everyone 6 extra runs to make league average 0, something I didn't do last year, just so we make sure we are weighing each player against average. Just like with last year, I am keeping the defense and speed values and the batting values separate, so if you want to ignore them, feel free, but hopefully this gives us some idea of how valuable the players' ability to steal (and by extension, run the bases) was. 

For defense, this year's method will be a lot simpler. One reason is the lack of defensive stats easily available (like with the pitching numbers, I used the numbers from MyKBO.net). So I will just use positional adjustments, the same I used last year. I will just use the rosters' position labels on homeplate.kr to determine what position a player usually is. For the purpose of this article, we are assuming that all players at each position are equal defenders, which is obviously wrong and overly simplistic, but will have to work for this article. Again, you can ignore the adjustments all together.

With the pitchers' WAR, I put every single pitcher that threw an inning in the spreadsheet. For the hitters, this isn't practical. There were 60 players who played in a game as a position player but didn't have a plate appearance. 106 of them had 10 plate or appearances or less. So I decided to post the WARs of the 129 players that had 100 plate appearances or more. This is more than I did last year.

For defense and baserunning, I only posted the league average rates, but for the batting, I used three different levels of measurements: replacement, average, and "foreign level expectation". For the batting version of WAR, I used the same simple runs created method, but as the run environment did change in 2013, I adjusted the baselines:

RC/PA League Average: .117

Replacement level (.320): .075

In 2014, KBO teams will be forced to use at least foreign roster spot on a position player (each team gets an extra foreign player as well), so I thought we needed some projection for how we expect those new players to perform. This isn't very mathematic, but without looking at past players for translations, we don't really have any idea what to expect from foreign position players. So I just used the same number baseline (which was gleaned from runs allowed per inning)  that I used for foreign pitchers in the pitcher WAR post, which is 1.1 better than league average:

Foreign Player Expectation: 1.1 * .117 = .129 runs created per plate appearance

The sum of the batting runs above average, speed score runs, and defensive adjustment (divided by 10 to convert from runs to wins) creates the Wins Above Average at the end (like last year, I didn't create an actual WAR for position players). Here is the leaderboard containing every player with at least 100 plate appearances in 2013, sorted by WAA:

Thursday, November 21, 2013

AFL Pitch F/X Data: Movement

So far in the look at Pitch F/X data in the Arizona Fall League on this blog, I have looked at the location of pitches, velocity, and even how balls and strikes were called in the league. In this post, I want to look at the movement of pitches, and first, let's look at a graph showing all the pitches thrown measured by Pitch F/X in the league, in the context of horizontal and vertical inches of movement:


Some of the more notable pitches (catcher's perspective):

Far to the right but no real vertical movement: Kyle Simon SL

Far to the left with no real vertical movement: Noe Ramirez FT

Positive vertical movement, some horizontal movement to the left: Matt Ramsey FF

Positive vertical movement, some horizontal movement to the right: Miguel Pena FF

Most negative vertical movement, some horizontal movement to the left: Jason Adam CU

Most negative vertical movement with horizontal movement to the right: Henry Garcia CU

Significant negative vertical movement and horizontal movement to the right: Brandon Maurer CU

Significant negative vertical movement and horizontal movement to the left:
Jason Gurka CU

Significant positive vertical movement and horizontal movement to the right: Mike Montgomery FF

Significant positive vertical movement and horizontal movement to the left:
Jonas Dufek FF

Of course, the biggest problem with using Pitch F/X movement is that it is often measured differently from park to park. That is, it is hard to compare pitchers from different teams using Pitch F/X movement because of all the biases in the data. The Arizona Fall League/Spring Training facilities are notorious for having some movement biases different from "normal" MLB parks. With this in mind, and using the MLBAM tags, this graph compares the average movement of each pitch type (just the major ones, I left out sinkers, splitters, etc.) in the AFL versus the MLB:

 We see that the majority of pitches are shifted up and to the left in the AFL, meaning there is a higher vertical number and more negative horizontal number in the AFL data on average than in the MLB.

Of course, this may be because of the pitchers' themselves. Perhaps there is a different ratio of righties and lefties (which affects movement data) or the pitchers in the AFL on average just have different movement than an average MLB pitcher. Luckily, there were a handful of pitchers in the AFL in 2013 that also have pitched in the MLB. Using just their FF (I used the FA from FanGraphs for their MLB data) tags, we can compare how their fastballs "moved" in the AFL versus the MLB:

Almost without exception, we see the same trend, to somewhat varying degrees. It seems safe to say that AFL vertical movement numbers are too high above 0, and too close to negative numbers.

However, I think we should look at movement data from the individual pitchers on the AFL's Pitch F/X data's own terms, so below are graphs on the four seam fastball, slider, curveball, and changeup split between horizontal and vertical movement, labelled with some notable pitchers.

Johnny Barbato is a Padres' right-handed reliever that was drafted in the 6th round in 2010. In 2012 in A-ball, he was very effective, and matched the peripherals in 2013 in A+. His fastball had better vertical movement than Alex Meyer. What about vertically?


 Ramsey and Adam show up again. What about sliders?

Malcom Culver is a 23 year old reliever in the Royals' system coming off a nice year in A+. 

It is interesting that Kyle Simon's slider appears a couple of times here, getting very little vertical movement at times, and a decent amount at other times. Will Roberts also seemed to vary his slider as well, getting moderate movement to good movement, while his horizontal movement was mostly moderate. Alex Sogard seems to vary his horizontal movement with his curve slightly:

Ryan Harvey is a Rangers' 2012 18th round pick that, despite a large volume of walks, pitched relatively well in relief in A+. Vertically, we see that some curves, perhaps those of Mike Morin, were misidentified sliders.

Matt West shows up again, showing good movement both horizontally and vertically. West is another Rangers reliever, missing most of the last two seasons with injury. West shows up again when it comes to changeup movement:

When it comes to vertical movement, I am not entirely sure why some, with Kevin Vance's name appearing around some of them, have negative vertical movement:
Trevor May, the former highly rated Phillies prospect that was a part of the Ben Revere trade appears with Royals' Noel Arguelles, a high bonus pitcher that has struggled with command.