Friday, January 25, 2013

My thoughts on Itoi trade, and Fighters


 If you were a general manager in MLB and got a trade proposal that you acquire Evan Longoria and Wade Davis from the Tampa Bay Rays by giving away Joe Blanton, Juan Uribe, and Jerry Hairston, what kind of a response or a reaction would you show? Would you accept the offer, either?

 Everyone -- Yeah, EVERYONE -- , would pounce on the proposition. In no minute would he spend thinking about the offer, as he would be nothing but anxious to think that while he was pondering the relative impact of the trade for his team, the opposing team realized the silliness of their thought and pulled the offer back before your answer. That kind of trade would never occur in real life, ever.

 But the world around you is too small to know the entire world. And where there are more than half a day time difference happened that apparently ridiculous player swapping. Three days ago the Nippon Ham Fighters, a team run by a lot of idiots in the front office, yet somehow considered to be by some dunces, strangely enough, "a smart team whose design and construction of being competitive is rooted in the sabermetric-type thinking", sent Yoshio Itoi, the best player in the league yet carrying with arguably one of the most lucrative contract under his belt, along with a crappy lefty starter to the Orix Buffaloes to acquire starter Hiroshi Kisanuki and shortstop Keiji Obiki, both at most average in their respective position, and one scrap-heap outfielder to give a great shock nationwide. Unlike MLB, the league in which stars are often transferred to another team even during the regular season via a trade, rarely happens the kind of trades which involves at least one MVP-caliber in Japan. Yet it occurred, in too ludicrous, cryptic a way.

 Some reports said that Itoi demanded to be posted this winter, that it ended up in forcing the Fighters to dump him in preparation down the road or as a "disciplinary action". But it misses a vital point; posting system can be ruled out if not for a club's consent. Suppose that to post him this winter doesn't bring sufficient profits they think is sufficient enough, then just keep him in the roster to play for them going forward. Unless he retires from playing in Japan as did Hideo Nomo in early-1990s, he's never be permitted to play in the United States under the current arrangement in the first place. And the likelihood of him making the determination is, of course, none.

 Their decision doesn't make any sense from the monetary perspective, either. The salary of three players they got summed up to 168 million yen. Itoi hasn't actually signed a contract yet, but reports said he was close to signing with 200M, which, in relation to his overall competence, is incredibly profitable to the team. The gap, 56M yen (equivalent to about 600K in dollar-term) can be easily made up for by trading other slightly cheaper but much less talented veterans eating up still bigger portion of their budgets. Not to mention that should they be in desperate need for relieving their sluggish payroll of some salaries, why the heck does such a patchy and crappy closer still see himself in the roster?

 They are the Nippon Dumb Fighters, a team which picked up Tomoyuki Sugano in the first-round at 2011 draft despite no promise of signing, a team which picked up Shohei Otani next year despite his official statement of no will of playing in Japan, little or no promise of reaching an agreement once again, but saw the good fortune in a counterfeit presentation, a team which said after the two straight proclamation of forcing their way through that it was their philosophy to select the best player available, nonetheless ending up in fact in picking up Yuki Saito two years ago. They are those who signed with a obscure softball player (I don't necessarily disparage the decision but just list a set of their deeds), give a high praise to a horrible tactician just because he's a knockout, and archived great success in brainwashing fans into their sect.

 Some dumb jocks are already entrapped into believing that the Fighters did the right thing as they thought they had done, the decision of which was a natural result of the output returned by their bragging operations system implemented some years back with investment of up to one million dollar. They are pretty convinced that that the team has done eccentric actions for the past couple of years is a natural consequence of their sagacious and audacity, ending up in an assumption that they ought to be right again in this trade as they think they have been. Not everyone falls into a crappy gin like this, as those jocks take only up around five percentages out of the population. But they are obsessed, haven't yet apostatized.

 How can the team be so mad and foolish in decision-making process? My assumption is, their goal is not to make themselves competitive and vanquish all the other teams en route to winning a season; it is for the marketing. A set of their peculiar deeds has attracted an enormous amount of attention all over the country, obtaining the great eclat of an advertisement. Shouldn't we be astonished to see them greet Hideki Matsui as a next manager?

 Is it great time to stop using #LoveFighters hashtag and go with #DumbFighters instead? It may or may not.

Friday, January 18, 2013

Catcher Framing Runs, Part 2


 In the previous post, I introduced my methodology of analyzing catcher framing skills using complete Pitch f/x database and published and commented some observations, and cautioned the bias inherent in WOWY. Today's post is built off by the previous article, so if you haven't read it yet, go first to the linked page and read the article (around 2,500 words, a bit long post) entirely and then go back here and start reading this post.

 In the last chapter of Part 1, I promised to investigate framing through a variety of situations. For the purpose of this post, I controlled for the identity of yet another causative agent, a catcher, to make things irrespective of a player behind the plate. Since I added the identity of catchers in the form of their estimated skills computed from the estimated skills from their batterymates, basically little or no overlap happened that caused double-subtraction. I also want to state in advance that in order to make numbers represented look more intuitive, I express net called strikes scaled to per 100 pitches in this post.

 [Pitch Type]
 Do different types of pitch make any difference in terms of getting extra called strikes as long as the pitch is thrown to the same location in the zone?

Pitch TypeRuns
FF0.16
FT0.14
FC0.11
SI0.09
SL0.02
KC-0.18
CU-0.24
CH-0.49
KN-0.97
FS-1.32

 One thing worth noting is that right-handed pitchers' two-seamers and sinkers look different from those thrown by southpaws when it comes to getting borderline pitches called to his favor. Righties can benefit from extra called strikes by throwing such pitches whereas lefties can't. Actually, I remember seeing the two sinking fastballs have slightly different impact on whether a successful bunt is expected to occur by a pitcher's handedness in the middle of conducting bunt analysis last December. With a little wrangling, I found out the expectation of fastballs being called strike varies considerably dependent on the identity of opposing hitters' handedness, giving pitchers much more credit if his competitor of the matchup sits in the opposite side of the dish. Likewise, hitters are more likely to see himself in more favorable count after the call of the pitch is done if he stands in the same side of the plate as the ball is thrown. And the split is larger for left-handed pitchers. Nonetheless, pitchers, especially for lefties, throw the disproportionate amount of sinking fastballs to the batters sitting in the same side of the box. This leads to why lefties' sinking balls look below average pitches in terms of framing at first blush. Overall, the above table is a decent representation of the truth however, that I didn't separate it by throwing hand or hitter's handedness to eschew making it look more dirty. As to this pitch type and handedness issue, I would do separate analysis in the future with an expanded application of the analytic field.

 As to other type of pitch, change-ups and splitters are less likely to be called strike, which is no surprise here. So are knuckleballs. Overall, the more vertical movement a pitch has to the downward direction, the less likely it is to be judged strike by a chief umpire.

 [Base-Out Situation]
 We know the fact that different base-out situation has different impact on overall runs transition. For example, we know that on situations where a runner on 3rd and less than two outs, pitchers change his approach to try to defeat hitters by strikeouts to glue the runner to the base, while batters counter the approach to desperately make a contact to plate the runner home and gain a run for his team's scoreboard. When it comes to framing (or more properly, just called strike rate since we don't necessarily focus on catchers' skill, but to avoid confusion and I don't like to fall into such a political conflict with lack of cultural enrichment of English semantics, please take the two expression as an indication of exactly the same thing in this post), how do chief umpires act depending on 24 base-out situations?

OutsBasesRuns
0___0.41
01__-0.51
0_2_-1.1
012_-0.33
0__3-0.92
01_3-0.49
0_23-0.09
01230.26
1___0.11
11__-0.47
1_2_-0.56
112_0.1
1__3-0.83
11_3-0.58
1_23-0.52
11230.42
2___0.09
21__-0.3
2_2_-0.38
212_0.36
2__3-0.27
21_3-0.25
2_23-0.28
21230.55

 It looks like that in bases-loaded or empty, defensive teams can benefit from some extra strikes while offensive teams are more likely to get a favorable call if 1st base is open. Put it more clearly, the divided line can be ruled by whether potential intentional walks could happen. There would be some reasons behind it, such as umpires' unconscious awareness of pitchers' attempt to pitch around. In other words, the relative impact of issuing walks corresponds to the probability of a pitch being ruled strike to some degree. One of my favorite metrics in baseball is boLI, but to take one step further and check out bbLI, it gets itself modest correlation with overall run impact, at r=.45. However, keep in mind that the net called strike is expressed as per 100 pitches, so even the two extreme situation, a runner on 2nd and no out, and bases-loaded and two outs, the effect is mere 0.2 runs per 100 pitches. You may be able to better understand the above output as percentage increase, and even the extreme situation can only affect 1% worth of absolute called strike rate.

 [Park Effects]
 How about the differences between parks? Max actually threw suspicion of some park biases in the final part of his trilogy of catcher framing that some parks may have a different influence than others, citing Brain McCann and Ryan Doumit as an illustration. Actually, there are such bias inherent in how a borderline pitch is judged, and the effect is not so little as to make some catchers look more eminent than others. Taking an example of Turner Field, defensive teams can be given an extra credit to the tune of 0.26 strikes per 100 called pitches, which is extrapolated to no more than 0.03 runs per game. But if you, a catcher, play half of your games there, you would get additional 2 runs to your pocket solely through the context you are in that you have no control over.

 Year-to-year correlation on called strike rate for each specific park is 0.26, implying there are not, if any, much of a persistent effect in successive seasons. Standard deviation between each park is 0.33, which can be converted to a couple of runs per 81 games played. That is, most regular catchers can be subject to around four absolute runs or fewer playing half of games in his specific park, and even extreme workhorse starters like Felix won't be able to gain more than a fraction of runs dependent on a park he pitches in half of his starts. Comerica Park and Sun Life Stadium is the two most extreme instances, favoring defensive and offensive teams respectively to the tune of six or seven runs per 81 games played. It has to be regressed a bit to estimate true maximum amount of runs caused by parks, which I'm sure is around five runs in reality.

 [Home Field Advantage]
 We all know teams playing in their home park generally wins more games against visiting teams and in today's Major League baseball, the advantage is four additional wins in a hundred games played. Put this figure in perspective, if you play your baseball only in your home park through a entire season, you can be in as good a position as when you would get the best player in the market for free. So how much of this extra advantage is led by the chief umpire's peculiarity to patronage the host team?

Home/AwayRuns
Away-0.25
Home0.24

 As you can guess it easily, pitches typically are more likely to be called strike if they are thrown by home team's pitchers. When visiting teams attempt to score runs during the at-bat, they lose 0.25 potential strikes per a hundred called pitches, equivalent to a couple hundredth runs per game. In other words, we find that the effect of called strikes on home field advantage is only one ninth of the total amount of runs caused by HFA. So umpires slightly favor home teams' players in judging borderline pitches, but there still remain many more causes beyond the match in 18-feet distance.

 [Run Difference]
 Does specific game context have any impact on the likelihood of a called strike? Some assumption is that when you are in a position to win the game if no more runs are scored, you're less likely to be aided by umpires' judgement on each play in the field. Is this true or just an illusion?


 The answer is, yes, umpires in general give more favor to a team behind their opponent at the time when the pitch is thrown. There are a bit of variations in the graph, but if you assume the linear relationship, one run increase corresponds to .06 extra called strikes per 100 pitches, which is not meaningful quantity ever. Put it in another way, there are run differential effect certainly, the amount of which is not we pay a special attention to however. After all, if your team starts the game by a leadoff homer, preserve the run through the entire game, and end up winning in 1-0, the whole batters on your team are not fooled even one hundredth runs during those PAs collectively. It's also worth noting that tie-situation is an exception, where umpires aren't so motivated as to give some favors to the defensive side. Maybe they don't prefer extra innings and want to go home quickly?

 So I took a one more look and demarcated it by innings in order to check out tying-situation. What I did was mapped out average net called strikes per 100 pitches by innings, and connected it to draw a line, but I also separated it based on whether one team's score is identical to the other teams'. Green dotted line represents non-tying situation while red dotted line tie-situation. As you can see the below image, the points in 9th and later innings see further estrangement from each other. Blue solid line represents the magnitude of the difference of the two lines, which shows that umpires greatly favor offensive teams in 9th and later innings if the game is in great competition. Maybe umpires are excellent entertainer? Or are they influenced by the bigger cheers by fans of an offensive team? Or just abhor extra duties?


 [One more thing]
 In my last post, I only briefly correlated individual performers in my dataset to Mike's by comparing in career level. The reason I didn't do it season-wise was just a coding error and now that I fixed it well, let me try a comparison and report the result in this post. For 2008 to 2011, the correlation coefficient of total runs with weights on the number of pitches each catcher caught in Mike's dataset spitted out r=.87. Mean absolute difference of seasonal performance between Mike's and mine is three runs, with only 10% of them seeing more than five runs discrepancy. And it looks like Mike compiled Max's dataset and appended it in his file, so I did the same thing on Max's framing result and found that r=.78 and average absolute difference is five to six runs between Max's and mine, with seven out of 80 players-seasons see more than 10 runs discrepancy. Those seven consist of three McCann and two Yadier Molina, probably suggesting I didn't deal with pitcher-catcher combination bias capably. Keep in mind that Mike and I took the similar approach to estimate the contribution while Max used multilevel modeling, and Max reported only 80 catchers, mostly ranked top and bottom, so it's natural for there to be more deviations found in comparison with Max's estimated values.

 In part 3, I'll take a more detailed look at framing skills by breaking them down to individual level, like the sequence issue I stated briefly in the first post. However, it would not be out online, at least in the near future, since I'm not motivated enough to query at this time. If it were published, the content is focused on why and how some catchers are superior to others in the field of framing, but it's more unlikely to come out than likely right now, so do not pin your hopes on it for the time being.