Checking out Dice-K with Peripheral Vision
Diasuke Matsuzaka has only pitched 5 games in the Major Leagues, so we really don’t know all that much about him. But in what we’ve seen so far, a very strange pattern has developed. Matsuzaka has dominated when it comes to peripherals, placing third in the league in FIP at 2.82. And lest you think he’s getting lucky on flyballs, he’s 7th in xFIP at 3.41. He’s done this by putting up an excellent HR rate (10th with .56 HR/G), a good BB rate (2.8 BB/G), and an excellent strikeout rate of 10.6 K/G, second only to Johan in the AL.
Why, then, has he put up the pedestrian 4.36 ERA, (ERA+ of 99)? One possible answer: Dice tends not to scatter his BBs and hits very well. He sometimes has an inning where things seem to completely fall apart for him. Yesterday, it was the 3rd, where he allowed 3 BBs, and 3 singles, leading to all four runs that he surrendered. Vs. Toronto on April 17, it was 3 BBs and 2 singles that did him in in the 4th. Dice has looked awful in these innings, but there have really only been two of them, so that can’t be the entire reason for his troubles.
I think there are two more compelling explanations. First, the Yankees and Blue Jays are darn good hitting teams. The Yankees in particularly seem to have a lot of good hitters who don’t hit that many HRs (Jeter, Cano, Abreu). These teams are going to score some runs off of you even if you keep your HRs and BBs in check, and K a good amount of hitters.
Second, luck (OMG he said luck! he must be a fanboy!). Dice really got himself in trouble with those walks last night, but the hits in that inning were all well-placed. If even one of them is hit at someone, he’s out of the inning with 1-2 runs rather than 4. In the Toronto game, their big inning again had 3 BBs, and 2 singles, which were both ground balls, and one of them didn’t leave the infield.
It’s too early to tell, of course. Maybe Dice will be like Jeremy Bonderman and consistently under-perform his FIP. But if he does, I doubt it will continue to be by a margin of 1.5 runs.
(All FIP and rate stats come from THT. The other stuff is from BBRef.)
Darren
Posted: April 28, 2007 at 07:58 AM |
74 comment(s)
Related News:
Boston
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
In both the bad inning in Toronto and the bad inning last night, he appeared to be trying to use the cut fastball as an out pitch when he got to two strikes - on the outside corner to lefties, on the inside corner to righties - and wasn't getting the call on balls just off the plate. That seemed to frustrate him; maybe he got that pitch called more in Japan and he just needs to adjust with it.
Nice sentence?
Of course having the hightest LD% isn't stopping Jeff Suppan from having an ERA < his FIP and xFIP. (So Matsuzaka is bizarro Suppan?)
MIL is .14 under league avg in DER, Sox are .12 over the league avg.
I think we'll see some adjustents gradually and be impressed.
I think DIPS are overused, and this is a particularly vivid example. Let's assume that pitchers only control their peripherals, which isn't true. When you look at just five games, their ERA will still differ from a DIPS ERA because the batters they've faced haven't averaged out yet. Darren hit on this in the third paragraph of the intro - if two of your five starts are against teams that don't hit a lot of HR but get a lot of hits on balls in play (which is certainly a skill for hitters) then you can have good peripherals and a poor ERA.
This all being said, I'm pretty happy with Matsuzaka so far.
No matter the sample size, DIPS-types numbers are usually a better indicator of pitching skill/performance than ERA is. So when his DIPS numbers are so much better than his ERA, I'd side with his DIPS numbers. It's at least worth investigating the differences.
I can see how component stats would be a better predictor than using ERA, I just don't see why you only use DIPS. Count singles and doubles too. I don't know how that translates to an ERA, but just eyeballing it I think the point still stands that Matsuzaka has pitched better than his ERA. He seems to have clustered his walks, and that's been the cause of a lot of his runs allowed.
Also, I don't understand why everyone uses K/9 and BB/9. K and BB per batter faced is clearly superior, and it's not like that statistic would be hard to calculate. Inertia I suppose.
Because they're a better indicator of how a pitcher will do in the future. So that would still be true with a small sample.
You've taken the most extreme example possible in order to show the limitations of DIPS. Yes, in that particular case, DIPS would do a pretty lousy job. But that's a very, very rare set of circumstances. Most times, DIPS would work better.
Why do I use only DIPS? I don't. I used his ERA and his DIPS (FIP actually), also added in information about his opponents, his propensity for bad innings (albeit only 2 of them), and I even theorized that he might be the type of pitcher who underperforms his FIP. Why didn't I include doubles and singles? Because pitchers have very little control over those things. If you think small samples throw something like DIPS off (and they do), then you must realize that they positively render doubles and singles allowed meaningless.
I covered this.
"Everyone" uses it because it's very easy to calculate and in some cases readily available. It's also close enough that it rarely makes much difference. Of course, if you'll notice, I don't use either of these stats in my post above. I use K/G, BB/G, and HR/G from THT, which has all of the advantages that K/batter faced does.
I think my post is pretty fair in recognizing the limitations of DIPS numbers while discussing how Dice-K has performed this year. To reiterate: there's probably been some bad fortune on Dice's part that has resulted in the his ERA being much higher than his peripherals suggest; it's hard to tell what has caused the rest of the difference.
That was in reference to Mike Timlin's stats, which are an even smaller sample than Matsuzaka's. Though the point still holds - that small sample size wreaks havoc with both measures - there's already a decent amount of stability around Matsuzaka's dERA.
One way to get a handle on it is by adding or subtracting one from any of the key values: K, BB, or HR. Actually, HR gives the biggest swing, so a decent range would be to look at what his dERA would be with one more (or fewer) HR. I have Matsuzaka at a 2.89 dERA. With one fewer HR it would be at 2.56; one more, 3.22. His actual ERA is a full point higher.
Darren, contrary to what you say in #19, dERA does help explain how he has performed this year. It might not explain the results of the performance, but it does quantify the performance itself. Once the sample size grows sufficiently large it can also help to predict how he will likely do the rest of the year. Despite the sample size - 138 BF - I think we're pretty close to a predictive dERA.
I might test that out in a little bit. We'll see.
I don't think I said anything that would refute this and I certainly agree with it.
Start: 1 2 3 4 5 6 7BF (cumul.): 24 56 84 120 147 175 207
ERA: 1.50 3.21 2.66 4.18 3.31 3.19 2.81
dERA: 1.59 2.18 2.87 3.09 2.71 2.86 2.85
Schilling's 7th start was on May 8. His dERA stayed between 2.75 and 2.96 from the prior start all the way to the end of July, when it moved to 3.10 and never dropped below again. (His HR rate went up; it was 0.55/G in his first 20 starts, 1.0/G in his next 12.) Prior to that, it looks like it reached a pretty stable point in start 6, with 175 batters faced. It could be argued that relative stability was attained as early as 150 BF.
Schilling's 2004 is only one example, so take it for what it's worth. But it looks like dERA is useful around 150-175 BF, and a bit shaky before that. His ERA, on the other hand, still sees some decent swings out at 400 BF.
The 1-HR rule, BTW, produces about a 0.25 swing either way around dERA at 175 BF. It's 0.30 either way at 150 BF.
That is, whether DIPS will work on Matsuzaka like it works on Schilling 2004 is precisely the thing at issue here. We don't really know. I tend to think it will, within a reasonable range - Daisuke has the stuff and command of a very good major league pitcher, and his none on / runners on splits look perfectly fine (650 OPS allowed for both).
I expect his ERA to come down over the next few starts. I'm not particularly worried. Daisuke's going to be one of the top 5-10 pitchers in the league this year, and we won't need to argue for DIPS to show that it's the case.
OTOH, Derek Lowe in 2001 - the year of the Derek Lowe Face, 5-10 with 24 saves in 50 finished games - had a 3.23 dERA vs. a 3.53 ERA. He had 404 batters faced on the season, but his dERA was within 0.6 of his ERA from 133 batters onward. He had a very inconsistent year, but his dERA and his ERA were not as far out of line with each other as we're seeing right now with Matsuzaka.
Personally I think part of the situation with Matsuzaka is a function of the offenses he's faced. His meltdown innings were 4/17 4th vs. Toronto (K, H, W, H, W, W, K, F8), maybe 4/22 1st vs. NY (F6, F9, W, HBP, D, G43), and 4/27 4th vs. NY (W, W, W, H, K, F5, H, H, F7). Those are the best offenses, and as the season goes on those should even out with better results against worse teams. Whether he will improve against those teams as the season goes on, or continue to melt down in the occasional inning against them, or melt down once per game regardless of the caliber of the opponent, I don't know. But like you, I'm not particularly worried about it.
It's interesting that 60% of his walks this year have come in 6% of his innings.
1) ERA based on defense independent pitching statistics over a whole season is a better predictor than actual ERA for actual ERA the next year, fine. That does not mean that a) It's the best predictor (though I realize the small number of inputs required is an advantage) nor that b) It does equally well no matter what the sample size is. This is especially true because when we talk about a few starts by a pitcher we're not just talking about small sample sizes, we're talking about a large portion of the sample being against a few batters. Therefore any characteristics of the batters are going to make a big difference. One part of that is obvious - Matsuzaka faced the Yankees twice so we expect his ERA to be somewhat higher than against average competition. All I'm saying is that for DIPS ERA, or whatever else like it you use, the tendencies of those batters to hit HR, walk, and K can introduce systematic bias. In just five starts, it's easy to construct a scenario where this would have an impact. Two of the five starts against the Yankees, for instance.
2) I'm not convinced that not considering singles and doubles is more accurate. It may work great for the portion of pitchers that give up singles and doubles at the league average, but we don't know ahead of time if a given pitcher is in that category or not. I'm willing to be convinced, but stating it as fact will not do that.
3) The stats I've looked at pretty much always include batters faced, so it seems just as easy to calculate. I'm not saying everyone here should do it themselves, just wondering out loud why people who were doing these things seriously didn't. I fail to see how looking at stats per game isn't even worse, unless batters faced by a pitcher in a game tends to be constant. Seems like you should naturally strike out more batters if you pitch deeper in the game though. That being said, I agree that it probably doesn't make much difference.
Again, I say that I'm optimistic about Matsuzaka specifically. He has a league average ERA against some better than league average offenses. He has pitched well the vast majority of the time. It's possible that the couple innings of control problems he's had are a real problem, and not just bad luck, but even that can presumably be addressed.
On another note, I read Sox Therapy a lot, though I rarely post because what I think is often already being said. In this case I thought I had something to add, and even commented on the common ground I had with Darren. I don't see why after my second response I should get a line-by-line response that consists of seemingly angry assertions of fact without an attempt to actually explain or discuss anything, other than that I was wrong in each point. Maybe I'm overreacting, but it doesn't make me want to increase my posting frequency, and I'm told I'm someone who does like to argue.
As to your other points, its seems that you are trying to say that DIPS/FIP have their imperfections, which is certainly true. I still think that they are more accurate than actual ERA, or at the very least that they add another layer of understanding and are worth including. I think that in most cases adding singles and doubles is going to cause more harm than good, so it's best not to (except as a sort of side note or as additional information). I don't think you've made a strong case that Dice-K is is the small minority of pitchers for which DIPS/FIP are not a good metric.
On the K/G, BB/G, etc., I think you're missing my point. They are not per "game" stats, but per XX number of batters, where XX represents the number of batters in average game during the season in question, which is usually around 38.
Bad call flustered Dice K?
BB, BB, BB, FC, HBP, 2B - those were the first six batters in his last start. Against the Yankees 4/27 - BB, BB, BB, 1B, K, F5, 1B, 1B. Toronto on 4/17 - K, 1B, BB, 1B, BB, BB. His walks have been amazingly concentrated, and a few weak hits (or errors) have been devastating in those innings - 11 runs (out of 23 on the season) on 9 walks (out of 15 on the season). It seems quite likely that this tendency is real, and it's not just luck that these walks have been strung together. I still think it can be fixed.
he used to throw lots lots more between starts right? and now he doesn't seem as sharp
Crossed my mind also, maybe he's out of his comfort zone for lack of a better term.
Is that true or is he not pitching as well with runners on base? (small but important distinction) I don't know, I am really asking. I haven't seen all his starts. When I have seen him, it looks more like jitters. That he is getting nervous in his first few major league games. He is locating his outpitches way out of the zone and getting too much plate with his other pitches. That very well could be because he loses performance in the stretch, but I haven't seen enough to know if that is systematic at all times during the stretch.
Nerves will go away after awhile. It will not be a major press event every time he pitches. If there is a problem off the stretch, it must be caught and corrected, which can be an iffy proposition.
That type of information is far more important than all the DIPS garbage.
I haven't noticed how well he does with a man on second vs a man on first so I can't answer the question whether it's the stretch or it's a command issue.
I honestly have no clue, but I can logically understand how switching from Japan to the U.S. or vice-versa would be a significant adjustment, while switching from AAA to MLB or NL to AL might not be as big. Maybe I'm misreading it.
IMHO, it depends on the type of adjustment being made.
For instance, let's take one example of your starting hot pitcher -- Hideo Nomo. Two important criteria for his early success was the dominance of his split finger fastball and his delivery. There is a limit to how much the league is going to adjust to that pitch. They may learn a little about when he likes to throw it, and they may recognize it better, but if he continues to throw it for strikes, its still going to have an impact. The delivery can be adjusted to, and its a matter of the timing. It may still have some effect, but the large majority of that effect will be later lost. A current example of a pitcher that I'd think could suffer from league adjustment would be Okijima with that trick delivery. It will still have value, but players will learn how to time it better.
For Dice-K, I think the adjustment being talked about above is where he locates the fastball. If he is not getting a call, he is used to getting, he should adjust to that.
In short, I think the league can usually catch up to something that is gimmicky such as a delivery or one special pitch. Those things may still have value, but the value is diminished.
A pitcher that has ML skills can usually catch up to a change in leagues, or an aggregate profile in hitters.
What I presume tends to balance itself out are the adjustments made on pitching patterns and hitter weaknesses. As long as a player doesn't have a fatal flaw that precludes them from the league, those types of adjustements may oscillate with one another.
The only recurring concern mentioned is the "pitching from the stretch" flaw. If he does lose effectiveness when pitching from the stretch, it will adversely effect his performance. Right now, that is far more important than some flawed designer statistics in a small set of performances.
We should be comfortable that DiceK will be good, because we have seen his array of pitches, and because of his success in the Japanese leagues. There should not be many flaws that would allow success in that league but preclude success in MLB (as opposed to a pitcher that can have success in the minors but not the majors). If he does not have success, we may be able to isolate said flaw. But those macro statistics aren't going to do that. In addition to what is already mentioned, the small space of performance is also more dependent on environmental conditions, injury, adjustment period, and just generalized mood. Using any of macro stastic is not far removed from saying, well he struck out Giambi therefore he is going to be a good pitcher.
I agree. I probably didn't articulate that well. I mean that when those runners get on (particularly after the other team may have scored), he may lose the command b/c of nerves (although not necessarily because of the stretch). As I mentioned, I am mainly thinking about the Yanks game, where during the big innings, it did see to be a command issue. He was getting a lot of the plate on strikes, and throwing those other pitches way out of the zone. (but that might not be true if just a single runner gets on with a big lead, etc.) I really don't know which is why I am asking.
You really think Okajima is really going to fool MLB hitters with that delivery alone? The Changeup and Curveball are quality pitches, otherwise he'd be just another pitcher in a long line of "krappy left handers that don't last more than 5 outings with the Red Sox"
Not from the delivery alone. Mike Myers couldn't fool players with his delivery alone, but it is still going to add something to effectiveness, because its a pretty big change in timing from what players are going to be in a rhythm of seeing.
okajima has 3 quality pitches that he can throw for strikes--a high 80s fastball that he can spot well, a change-up that he can spot on either half of the plate, and the breaking ball that he can throw for a strike or bounce to get a swinging strike. that's legit major league stuff ... i have no doubt that he'll continue to be successful if he maintains command of his pitches.
Slider what?
He's got the huge 11.5-5.5 curveball at 70ish mph, the 80mph changeup which drops, and a fastball that he can relatively locate (dones't move a lot, isn't all that fast). Supposedly he's got some forkball thingy in Japan, but Ihaven't seen him throw it here yet. Haven't seen a slider?
Okajima's delivery's effect is overrated. It's not THAT weird. Myer's has a weird delivery. Fuentes has a weird delivery. Even Foulke's delivery I would classify as weird. Okajima? Unorthodox, but not weird.
Could he not get it through customs?
I haven't seen his Japan games, but could that "changeup" be his forkball? As crispy mentions, he has some late downward movement on that pitch.
And the rest is history.
At this point (I'm only counting guys who stuck in the majors for more than a year), the only real Japanese import to have lasting success in the bigs is Ichiro. You could make an argument for Sasaki and Nomo, but I think both of their MLB careers qualify as "disappointments".
If it only the leap between the quality of Japanese baseball vs. American? Or is there something else there at work? I know that people here pooh-pooh the "softer", not quantifiable factors. But I've done a bit of touristing in my life; there is a certain mental strain involved in living in a country where your countrymen are few and far between; where something simple like food shopping is really complicated because of the language barrier, where the food is strange to you. The Sox (and other MLB clubs) try to assuage this as best they can, but the results thus far are not too encouraging.
That said, I think Matsuzaka will be fine. I don't think he'll settle in to "top ten in baseball" status, but there's significant value even in his realistic downside.
Racist Tariffs
oh come on. nomo won over 100 games, sazaki came over late and had 3 good years (at age 32, 33, and 34) before getting hurt, iguchi has been solid, and otsuka has been solid. those are just off the top of my head.
H. Matsui has been solid as well. In fact, the disappointments aren't that long:
Irabu - definately underperformed
K. Matsui - He was a met so I didn't expect much.
Yoshii - decent years, did we expect much more
Komiyama - Supposedly the Japanese Greg Maddux, wasn't even the Japanese Mike Maddux
I think Sasaki has some; I think Nomo has some.
Matsui was a 50-home run monster in Japan - one of the inner-circle greats of the outfield over there, from what I understand. I don't know what the Yankees thought they were getting exactly, but I think it's fair to compare Hideki to "Paul O'Neill lite".
There are a lot of teams that would be very happy to have had Matsui production. What were you expecting -- Barry Bonds.
The expectation is so high on Dice-K because you are getting him for his prime years. Moreover, you don't have to look at Japanese league equivalency or other junk stats, you can see the array of what he can do. Many people saw him pitch in the WBC, and saw his repitoire of pitches.
it's been established that power numbers do not translate well.
There are different levels of success. Hideki Matsui has been a solid ML left fielder, which is success, but he's a failure if you were looking for someone hitting 50 home runs a year.
Shigetoshi Hasegawa came in at age 28 with no pedigree or expectations. He threw 9 years with a 124 ERA+ and an All-Star appearance in his brilliant 2003. He was never a star, but he was a good foot soldier. That's success.
That's a bit misleading, as Matsui hit 50 once, and hit 50 exactly. But he was a monster over there and it probably wasn't all Yomimuri-style hype. His HR totals from age 20 on:
22, 38, 37, 34, 42, 42, 36, 50
Those were in about 80% of the AB's he had in his 1st NYY years.
imes Facing Opp. in GameI Split G PA AB R H 2B 3B HR BB IBB SO HBP SH SF ROE GDP SB CS Pk BA OBP SLG OPS BAbip sOPS+ tOPS+ Split
+-+------------+---+----+----+---+---+--+--+--+---+---+---+---+---+---+---+---+---+--+--+-----+-----+-----+-----+-----+-----+-----+------------+
1st PA in G 7 63 54 8 11 3 0 0 6 0 15 2 0 1 1 1 0 0 0 .204 .302 .259 .561 .275 60 72 1st PA in G
2nd PA in G 7 63 53 4 10 1 0 0 8 0 16 2 0 0 1 1 0 0 0 .189 .317 .208 .525 .270 48 62 2nd PA in G
3rd+ PA in G 7 68 64 11 19 4 0 3 4 0 16 0 0 0 0 2 0 1 0 .297 .338 .500 .838 .356 110 156 3rd+ PA in G
Now, batters on average do better against starters in the 3rd PA. But the sOPS+ is adjusted for the split; Matsuzaka has been doing worse than the average on that split.
This indicates to me that most of the damage done against him is crowded into that third time around the order. Whenever you crowd all that damage in such close quarters, a guy's gonna give up more runs than his peripherals would indicate.
I would suspect that the distribution of hits against him will even out, and his ERA and Component ERAs (DIPS, FIP, etc.) will converge as the season proceeds.
Games with a HR allowed: 30 IP, 2.70 ERA, 30 K, 5 BB.
Games with no HR allowed: 24 IP, 6.00 ERA, 22 K, 13 BB.
Mostly just for fun, but perhaps it indicates that he is more successful when going after hitters than when nibbling. Tonight's win was also great because it came against one of the league's top offenses.
After his next start, last week vs Toronto: 4.80 ERA, 3.45 dERA.
After his start this evening vs the Tigers: 4.17 ERA, 3.46 dERA.
Hmmm... I see a 3.50 ERA in his future... which would be around a 125 ERA+ right now. It'd be nice if he improves on that, naturally, but that depends on whether he'll continue with the meltdown innings.
Toronto was once one of the league's top offenses. Then they played Boston.
Seriously, twice I've led off Game Chatter for a series (once TOR, once NYY) saying that it was a good test for Boston's pitching because they were going up against a good offense. I've grown tired of saying that. It's now worth saying that Boston's pitching is a good test for whatever good offense du jour is facing them.
It's interesting, though, that the Sox have decided they're perfectly comfortable letting Daisuke throw 120 pitches. Not every night, certainly, but occasionally.
There's a discussion on SoSH of the Globe report that the Sox decided, after the Seattle start, to let Daisuke increase the intensity of his workout pattern such that it's closer to what he did in Japan, even having him toss 100+ pitches in a side session. Dunno if that's responsible for the recent turnaround, but it's worth noting.
You must be Registered and Logged In to post comments.
<< Back to main