What’s Wrong With Ichiro
For the first time since May 12, 2004, Ichiro’s batting average is under .300. He’s 7 for 42 in June, following up a poor May, and we haven’t seen Ichiro! be Ichiro! in nearly six weeks. With the offense struggling, all eyes go to the team’s best hitter. And, of course, theories abound, and everyone seems to keep asking the same question. What’s wrong with Ichiro?
Nothing. Absolutely nothing. Since 2002, these are Ichiro’s batting averages by month:
.316/.404/.353/.321/.282/.248/.243/.389/.386/.342/.242/.273/.255/.400/.274/.432/.463/.373/.356/.288/.167
Not including this month, Ichiro has had eight months in which he hit less than .300. That’s 40 percent of the months played. It should be expected that, in any given year, Ichiro has at least two months where he hits .300 or below.
Look at those variances above. From .243 to .389. .342 to .242. .255 to .400 to .274 to .432. Ichiro is what he is; consistently inconsistent.
We shouldn’t be surprised by this. In his major league career, 81 percent of his hits have been singles. That’s an astounding figure. The league average in 2004 was 66 percent. Only a few guys in baseball rely on the single as much as Ichiro. Luis Castillo, the biggest singler in the game, sees the same type of huge monthly swings in his average.
Singles, are, by nature, the least reliable type of hit. The difference between an infield single and a 6-3 putout is usually a tenth of a second. A ball finding a hole might get under the fielders glove by less than an inch. Have a run of balls that just don’t quite find the hole and all of the sudden, your singles are outs and you’re hitting .250. Extra base knocks, however, are almost always hits, and a player who drives the ball will be far less suceptible to the swings of random variation.
Every time Ichiro goes through a prolonged slump like this, new theories arise. Remember when the Yankees had figured out how to pitch him down and in? As human beings, we like answers, so we try to figure out what Ichiro is doing differently, what he’s doing wrong, what he needs to fix. We don’t like that the answer is random variation. It’s not a satisfying conclusion, and it doesn’t make us seem like minds who understand the game. Random variation is very real, though, and while it may not be a sexy explanation, it is the correct one.
Ichiro is going to have months where he hits .270. He’s also going to have months where he hits .450. He’s a player of extremes, probably the most valuable in the game when he’s going well and replacement level when he’s not. You live with months like May because you know a month like August of 2004 is coming. With Ichiro, you take the good with the bad, because in the end, it’s an extremely valuable package.
Comments
76 Responses to “What’s Wrong With Ichiro”
This is a tangential observation… but I’m monitoring the St. Louis – Toronto game and noting that Halladay is mowing down the powerful Cardinal lineup and yet one guy has 2 hits off of him. Little Davey Eckstein. The guy is hitting .304/.378/.388. Just what you want from a leadoff hitter. I’m not suggesting that Ichiro should have been replaced by Eckstein on the M’s roster. What I am saying is that we could have had both of these guys in the lineup instead of the predictably injured Pokey Reese.
Ichiro is a victim of his own success. Every year when he approaches a personal goal he starts to feel a self-imposed pressure. Remember last year as he was approaching the hits record. He was really doing poorly: one hit here, one there, or an 0-fer. This happened as he approached 200 hits in 2002, and 2003. And then on top of that, he likes to be dramatic at times. He will probably want his 1000 hit to be memorable so he might try for a homer. He has done so for his 200 hit in the past. And Ichiro’s marker of his struggle is the fly ball, especially the weak pop-up — he certainly has gotten a lot of those lately. I think once the 1000 hit is done with, look for him to relax but also take a while to shake off the bad after taste of the slump. I think June will continue to be a bad month but he will back to his old good self by July.
To a large degree you are assigning a random walk theory to a specified skill set. I don’t think you can use a stoachistic option model (see: Black-Scholes) to make judgements regarding Ichiro’s hitting average. I think the truly random part you are discussing is the arbitrary monthly average you are forced to use. In other words, the seeming randomness could be removed by the addition of more complete information. Everything is ALWAYS controlled by the bell curve and the 80-20 rule. Any violations of those rules are due to lack of information, not the rules themselves.* Since contributors have frequently noted the concept of the “regression to the mean”, the major topic regarding Ichiro circa 2005, is what is his “mean” this year? *tongue modestly in cheek, but not by a great amount.
#49 Dave:
Why are we so set against the mentality that there are skills in life that are very hard to repeat?
Probably for the same reason that most people believe that they are better than average drivers. (Only 50% can be better than average.) For whatever reason, there seems to be an evolutionary advantage to believing that you are somewhat better than you actually are.
Why is Ichiro struggling?
By applying Heisenberg’s Uncertainty Principle, a dash of Chaos Theory, and a few odd references to the Third Law of Thermodynamics, anyone should be able to come up with an easy answer to this most perplexing question…. 😉
#50 but would you have paid Eckstein three years/$10.25 million? Because thats what it cost St Louis….
I have just killed my goat, and upon close examination of the entrails, I predict that Ichiro! will start to hit better in the summer and fade in September.
So you can just shove all your fancy shmancy equations and science, because goat guts never lie.
Well, I’ve looked all over the Baseball Prospectus non-subscriber site and can’t find the mathematical formula for VORP. In the glossary, the description of what’s being measured is there, and there is a link to “Details” but the link gives the same info as the glossary. Does anyone have a web address to the mathematical formula for VORP?
While I agree Ichiro will “come around,” this doesn’t appear to be a “normal” Ichiro slump. A dramatic difference in his numbers with bases empty vs. runners on persists throughout this season. For instance BA (empty/runners/% of AB with runners):
Apr: .378/.296/26.7%
May: .342/.171/31.5%
Jun: .208/.111/42.9% (way to go Borders/Rivera/Morse)
FWIW.
#58: You have keenly spotted an interesting statistical anomaly, but Ichiro’s average with runners v. average without runners is meaningless with a sample size of only 80 ABs. I think the point you want to make is Ichrio is not as good of a hitter with men on base because force outs are possible, but with a larger sample size, Ichiro has hit virtually the same with runners on base (2003: .310 BA/.383 OBP in 216 AB; 2004: .364/.439 in 228 AB) as without runners on base (2003: .313/.336 in 463 AB; .376/.439 in 228 AB).
Some Ichiro comments from a Japanese News article asking him about going below the .300 mark:
“It never feels good hitting below .300. It’s stimulating, more so than hitting below .400.”
Asked whether he is in a slump:
“I’m keeping that a secret. I might be in a slump or I might not.”
Not much, but some Ichiro comments on his .167 batting average this month 🙂
The article which is in Japanese:
http://www.sanspo.com/mlb/top/mt200506/mt2005061301.html
Ah, but 80 at-bats IS a big enough sample size to conclude that Ichiro’s 2005 runners-on-base performance is VERY unlikely to be the result of “random variation.â€Â
Any BA split is simply the percentage of success of a binomial distribution of (H) successes in (AB) trials with a (BA) probability of success (where “BA†is the “true†batting average, whatever that might be). To test whether the variation is merely random, we compare the split against what we would expect to obtain from a truly “random†sample with the same number of ABs.
If we test the theory that Ichiro’s ’05 BA with runners on (16/80, or .200) is simply random variation from his “true†(lifetime) BA of .335, we find that the probability is a mere 0.6%. As a scientist would put it, we conclude with 99.4% certainty that Ichiro with runners on in 2005 is not the true Ichiro. Comparing it to his 2005 average of .295, we can conclude the same, but with only 96.3% certainty.
A couple of other splits, compared to his lifetime BA of .335:
2005: 75-254, .295, P = 10%, not significant.
May 05: 32-111, .288, P = 17.3%, not significant.
June 05: 7-42, .167, P = 1.2%, significant (98.8% certainty).
As you can see, a large difference cannot be dismissed simply due to sample size.
You can use the Excel BINOMDIST function to test any split against a same-size random sample taken from a population with a given batting average:
BINOMDIST(H,AB,BA,TRUE)
where H and AB are the split, BA is the “true†batting average, and TRUE tells Excel to calculate the cumulative probability (H hits or less).
BTW, use “FALSE” if you want to know the binomial probability of any single result. For instance, a batter with a “true” BA of .335 can randomly be expected to produce:
0-5 13.0%
1-5 32.8%
2-5 33.0%
3-5 16.6%
4-5 4.2%
5-5 0.4%
Yes, Ichiro! is a roulette wheel, one that you can hit on 33.5% of the time, so one that is rigged, fixed, unbalanced, whatever. And it’s his influence/ability that unbalances the wheel.
The fixation on June and the insignificant sample size therein is just, and remains, insignificant.
Another quarrel: Why is Randy Winn hitting over .300 this early in the season?
You can accept the mathematics, or you can ignore it, but the fact is Ichiro in June 2005 clearly is not himself. Just because we can’t identify any reason for the slump does not mean it’s not a slump. The small sample size tells us only that the duration of the slump is short (so far).
Mathematics? Mathematics tell us that June 2005 is both insignificant over a season and far from over.
64 – I have yet to see another 2 week sample offered by anyone to contrast with Ichiro’s June-to-date. It’s entirely possible he routinely does this in his down months.
It appears that there are two different questions being addressed here.
Brain Rust is answering the question: “Can Ichiro’s BA in June 2005 be attributable to random variation under the assumption that his hitting can be modelled as a memoryless process.”
I believe he has answered this question correctly at the 99% level.
Others are asking an entirely different question: “Is Ichiro’s BA in June 2005 significantly different from our observations of Ichiro over the past several seasons.”
The trouble that I see is that if Ichiro’s AB sequence is NOT memoryless, which seems to be what people are implying, then it is also true that the outcome of his “next AB” wouldn’t be independent of his previous N AB’s.
I’m slightly puzzled …
Evan, I haven’t looked for a comparison period and you could be right. If he hit the rest of the month at .295 he’d match his worst previous month, and if he hit the rest of the month like his best previous month he’d finish June at .345. But he better start doing it soon or he’ll have a new worst month.
Look, it’s not as if I’m trying to prove Ichiro has become a bad baseball player (although the data shows he IS in a slump). I’m just tired of seeing “random variation” so carelessly and incorrectly used to explain any variation for which no other reason is posited.
Actually, Brett Farve, I’m DISPROVING the notion that hitting is a “memoryless process.” Those who attribute performance variability to “random variation” are treating it as if it is.
Brian … My comment was intended for those who suggest that hitting is memoryless. I understand your point and agree with you.
However, I am having trouble getting the 0.167 that you claim in post number 61. Could you please elaborate?
Brian, it may be you and others are arguing a bit at cross purposes?
I think the data used (BA per month) clearly shows that recently Ichiro has not been hitting well, but OTOH this does not suggest that a predictive trend due to some (as yet undisclosed) reason.
Strictly, as you suggest, the data is not truly representative of a classic random variation, but meets an everyday usage criterium.
Another way to look at this: there is no such thing as randomness because everything happens for a reason. And free will is an illusion.
Brett Farve #70 — In the first 10 games of June, Ichiro has 7 H in 42 AB, a BA of .167. If we had 1,000 lottery balls, 335 printed with “H” and 665 with “O”, and drew out 42 balls (representing AB)the probability of drawing 7 or fewer with “H” is only 1.2%. Given all the factors that go into hitting in the major leagues, I’m going to conclude that some combination of those factors is more likely to be causing his slump than just “random variation.”
Granted, this may or may not have predictive value (congratulations on that nice liner to right for hit #1,000), but it does tell me Ichiro should be looking for some adjustments to make. It makes the game more interesting for me as well if I try to look for factors that might be causing his slump.
Then there’s also the matter of his runners-on-base performance. While 80 AB certainly doesn’t represent his total performance (and I never said it did), it is enough to conclude that his performance in this situation this year is significantly different from his performance without runners on base. Whoever says it isn’t can’t back it up mathematically.
This thread may be dead … I misread your original post. Not sure why I asked about 0.167, brain fart I suppose.
You’re right, the probability of a 0.336 batter getting 7 or fewer hits in 42 AB’s is small … but …
Ichiro has 2976 AB’s in his career. A typical “Ichiro” will have about 84 different stretches in his career in which he gets 7 or fewer hits in 42 AB’s.
I’ll have to side with randomness.
Ichiro’s problems stem from the curse of the Cubs. Don Baylor, one of the most atrocious managers we’ve seen here in recent years and now your batting coach, is messing with Ichiro’s mechanics. He’s probably telling Ichiro to alter his swing, and the cumululative effect of all the losses has made him “press” more at the plate. Why hasn’t anyone picked up on this?