Willie and Washburn
JMB · July 9, 2005 at 1:12 pm · Filed Under Mariners
As maddening as it may have been to see Bloomquist in the lineup again last night, you pretty much have to start him any time the team faces Jarrod Washburn. I used to write about this when I did his profile for the Grand Salami, but however you slice it, Bloomquist owns Washburn — he’s 10-for-20 against him in his career with a .545 OBP and one of his two career triples.
Sample size, schmample size. Every hitter has one pitcher — his bizzaro-world match — he simply destroys. Whether or not the two ever actually face each other is a different story, however, and I hear Charles Gipson is still looking for his.
Comments
33 Responses to “Willie and Washburn”
Jason – We touched on this a thread or three or four ago. I agree with you that most hitters have a pitcher they own. However, if you are saying that WFB can be expected to hit Washburn at .500 in perpetuity, I’ll argue that. Sample size is meaningful. Even against his own private whipping boy, a hitter is going to regress to the mean. If he is a “true” .250 hitter, maybe that means he’s a .300 hitter against his whipping boy. But I don’t think it’s going to be much more than that, provided he gets a significantly large enough sample of ABs against the guy.
Somebody else in that earlier thread said they had read something about where 20-25 ABs against a pitcher is the point where you can begin to see a longer-term predictive correlation between past performance and future performance against that pitcher. Maybe so, and maybe WFB is now at the level with Washburn. But I am hard-pressed to believe that WFB is much more than a “true” .300 hitter against even Washburn. Does that mean he should play? Maybe, but rather than looking at it as a pure righty-lefty thing, or the “well, he’s 3-5 against this guy career, so . . .” all I am asking is to consider the fact that somebody like Doyle, despite hitting left-handed, was hitting something like .354 against LHP in Tacoma this year, in way more ABs than Willie has had. Right-handed or left-handed, Doyle is simply a better hitter than Bloomquist, and that may be true whether Bloomquist is facing his own private whipping boy or not.
A little knowledge is a dangerous thing…
but based on my own geekery in the last 24 hours, arguing with some others in an earlier thread about the “hot hand” theory, and on what I remember from my stats Prof Pete Nye at the UW, in a sample of 20 anything can happen. A sample of 60 is much more likely to give useful information.
I don’t know if this has been posted, can’t find it, but ESPN Page Two had an article by David Schoenfield where he puts together the worst line-up of all time… the worst regularly employed hitters of all-time.
Two things about it…
1- it was depressing to realize how poorly some of oiur guys compare against the worst of all time.
2- This fantastic quote:
“No wonder pitchers like Drysdale, Koufax, Seaver, Carlton and Gibson were able to put up such great numbers in the ’60s and ’70s — every team had a guy like Bobby Wine, Hal Lanier, Gene Michael, Dal Maxvill, Don Kessinger or Sandy Alomar who couldn’t hit a lick … guys with low average, no power and no ability to draw walks. It was like a league full of Willie Bloomquists.”
“no ability to draw walks. It was like a league full of Willie Bloomquists.”
While I am not stupid enough to claim that Willie Bloomquist is a great baseball player (even though I do like the scrappy little soandso), his BB:SO ratio is not all that terrible compared to some. He’s got 41 walks to 99 strikeouts lifetime in 509 lifetime ABs for a SO:BB ratio of about 2.1:1. That’s not horrendous, it’s not great either, but it’s not horrendous. His OBP is 50 points higher then his lifetime AVG. If that were consistent, and he was a .300 hitter, his OBP would be about .350. Which, again, isn’t horrendous, but not great either.
But despite Willie’s plate woes, he’s still quite speedy. He could probably eek out a few more basehits then the likes of Alomar and Kessinger just on his legs alone. I think Willie would do better if he played to those strengths; choke up on the bat like That Little Pestâ„¢ David Eckstein and just put the ball in play and see what happens. Instead, Willie keeps trying to hit the ball hard with his hands down on the knob of the bat like Bret Boone. He needs to stop doing that.
Here’s the link: http://sports.espn.go.com/espn/page2/story?page=schoenfield/050708
Funny stuff.
But despite Willie’s plate woes, he’s still quite speedy. He could probably eek out a few more basehits then the likes of Alomar and Kessinger just on his legs alone. I think Willie would do better if he played to those strengths; choke up on the bat like That Little Pest™ David Eckstein and just put the ball in play and see what happens. Instead, Willie keeps trying to hit the ball hard with his hands down on the knob of the bat like Bret Boone. He needs to stop doing that.
Hm. I would find it hugely amusing, in a cosmic way, if he did that and became a league average player.
And about sample sizes….depends on the results. If a player goes 18 for 20 against a pitcher, I’m pretty sure you’re getting a statistically significant result, even with a sample size of 20.
#1 & #2: Do you guys know how long it takes to get 60 AB worth of batter vs. pitcher data? It takes so long that by the end of that span, both guys have become completely different players. Since starting pitchers only go every 5th day, the matchups simply don’t happen that often.
To take the extreme case, I looked up Bernie Williams vs. Pedro Martinez on ESPN. They were in the same division from 1998 through 2004 (five years) and faced each other in two postseasons. I don’t know if the postseason AB’s made it into the database, but in any case there are only 76 AB’s worth of data for them. I guarantee you this: the 2004 edition of Pedro was only a shadow of the 1999 edition, and to a lesser extent that’s also true of Bernie.
There are only 48 AB’s worth of data for Williams vs. Tim Wakefield, and they’ve been in the same division even longer than Bernie and Pedro were. If you think the minimum for a reasonable sample is 60, you might as well say that you should never look at pitcher/batter matchups.
Correct.
The chances of someone starting off like that and then going on to average less than 500 against a given pitcher are pretty low.
10 for 20 you have to be a bit more circumspect, but the odds are pretty good that Judy Bloomquist will hit Washburn at at least a .350 clip forever more.
(Just using binomial distribution, there is a roughly one-in-eight chance of a .350 hitter going 10 for 20 against pitcher X; there is less than one chance in fifty that Judy would hit Washburn like this if he was batting at his established .260 clip)
I think I’m contradicting what I said earlier. I blame sleep deprivation.
While I agree it’s annoying that Port Orchard’s own gets attention all out of proportion to his ability, I do think the condemnation on this board sometimes gets excessive. Bloomquist’s line isn’t all that different from, for example, Mark McLemore during his time here (particularly the last couple of years). I don’t remember people screaming for McLemore’s head: he was a utility guy, he did his job, and was a useful 25th man. And Bloomquist is cheaper and runs the bases better. Now, if they keep him around when he starts earning more than the league minimum, then it starts getting silly.
Re: the sample size of 60 question:
The stats prof talking about the sample of 20 being very unstable was likely referring to a sample with independent observations (i.e., get 20 different people where each person’s performance isn’t affected by anyone else’s), in which case it’s true that 20 would be an unstable sample size. However, in batter/pitcher matchups, you have a paired-or “repeated measures”-sample where a lot of the individual differences that destablize the independent sample aren’t an issue. I’d agree with those who say that 20-25 is a good number for a stable repeated measures sample.
#10: If McLemore ever got an extra base hit it was an accident, but at least he did sport a lifetime .349 OBP. WFB’s lifetime OBP is .319.
Re 8. Agreed. You just don’t get sample sizes of hitter vs batter large enough to prove anything to the satisfaction of a statistician unless you are talking about bizarrely extreme results – 0 for 15 or 18 for 20.
Other than that you have to accept a lot of uncertainty. Statistically, using a fairly low standard of proof, you’d say it is probable that Willie Bloomquist will hit somewhere between .272 and .685 against Jarrod Washburn ad infinitum.
11. The pitcher and hitter are the same, but there is a lot of noise in there (park, defence, atmospheric conditions, day/night, how much either of the players drank the night before etc…) to be considering these “repeated measures”.
The info is worth something, but take it with a big pinch of salt.
i could see not playing willie if he wasn’t hitting well, or didn’t hit washburn well, but it’s hard to counter 10/20 against a pitcher + 9/21 w/ 5 2B over the previous few games. and it was the right move, he got on 3 times last night.
McLemore is only similiar to Bloomquist in his last year here. The year that got him sent packing was pretty much Bloomquist’s career line.
Before that his OBPs were .353, .384 and .380. WAAAAY better than Bloomquist.
In 2001 his sb/cs was 39/7, so I’m not sure Willie “runs the bases better”.
In 2001 and 2002 he also slugged .406 and .395. Again, Bloomquist’s career slugging is .348. Even in the minors it was .375.
I would say McLemore was twice as valuable as Bloomie as the 25th man, myself.
I kind of agree on the Bloomquist love (the M’s broadcasters) hate (the USSM posters) thing. Willie’s a below average hitter making $385,000; he’s a good basestealer, and probably the second or third best defensive player the Mariners have at any position on the field (which is a roundabout way of saying there’s at least one better defender than Willie at every position).
This isn’t Lake Wobegone; some kids have to be below average. Bloomquist is actually slightly better than, say, Christian Guzman, and Guzman makes $4.2Million a year.
Hey Daaan, you’re justifying decisions based on results again.
It was the right decision based on that 10 for 20 and that is the only bit of info that should have influenced the manager.
9 for 21 will happen to almost every hitter in the bigs at some point this season if they manage to log 500 AB. It does not mean they are likely to go 9 for their next 21.
And saying Hargrove was right because because Willie went 1 for 3 with two walks is using the same logic as saying you were right to play the lottery because you won a prize.
#8 – I never said you need to have 60 ABs (or any other number somebody believes to be statistically significant) before you can use batter vs. pitcher info in making a line-up decision. What I said was, I disagree with the increasing tendency to rely on 20 ABs (and many times, much less than that) as virtually the sole factor considered in making that decision — maybe that and a lefty-righty split. I am lamenting the overuse of “the hot hand” and lefty-righty platoons, and the decreasing consideration of the far more important factor of who is the better hitter, regardless of whether he has a (usually small sample) history against a pitcher or is opposite-handed.
#15 – I’m not necessarily disagreeing with you, except that I still value longer term performances over a 20 AB sample against a pitcher, or a 21 AB string over 5 games. That tells me that WFB is not a very good hitter, where we have reason to believe Doyle will be a good ML hitter. The larger samples have more predictive value, IMO, than do either of those 20 AB samples, and the fact that WFB got 3 hits, after the fact, does not make the decision to play him, before the fact, a wise one.
#5 – I was just coming here to link to that story, good stuff.
11 – The fact that there is noise in measurement is not related to whether or not 2 measures are considered independent or paired. An independent observation is one in which no other observation in the sample affects the probability of any other one. If that condition is not met, then the measures are not independent. Whenever you take more than 1 measurement from a single individual, that assumption is violated, since that person’s skill/disposition/whatever at time 1 also impact their performance at time 2, regardless of whether or not there is noise in the data like parks, weather, etc. Those factors provide the variance/error in your data, but I can guarantee you hitter/pitcher matchups are paired/repeated measures, not independent.
Also, for what it’s worth regarding post #9, an assumption of the binomial test is that measures are independent. Perhaps you were using a Poisson distribution? I don’t know much about the Poisson, but I understand that it allows repeated measures. At this point it should be obvious that my statistics training isn’t of the type often used in baseball (my graduate training is in statistics for the social sciences), and I just may not be familiar with the distribution you’re using.
Cheers,
Oly
Re 19. Pete, I brought up the figure 60, and I realize I was overstating my case, but I agree with your point.
20 AB against a certain pitcher has some predictive value of what a hitter will do in his next 20 AB against that pitcher. You just have to accept a large degree of uncertainty – like the 413 point spread in Willie’s probable future batting average against Washburn as outlined in 13.
20 random AB during the season has ABSOLUTELY NO PREDICTIVE VALUE WHATSOEVER as to what a hitter will do in his next 20 AB. Just let it go.
21. No, binomial. I can’t remember a lick about Poisson Distribution. My only training in stats was for business grad school. I’m reaching around in my memory to make any sense of this.
Re-reading my last post (#19), I realize it may sound sorta like I’m taking both sides of this. To clarify, I think Willie’s 20 AB history against Washburn is worth considering in making a line-up decision. I also think the platoon advantage is worth considering. Both would be factors lower on my decision tree ladder, however, than his longer-term history as a hitter, period. A hot streak over 21 previous ABs would be something I would also consider, but it would be the lowest factor of them all. The most important factor would be the fact that, over 500+ career ABs (a disproportionate amount of which have been given to him to give him a platoon advantage against LHP), Bloomquist has proven himself to be at best a league average hitter for BA, and under league average for both OBP and SLG. That means, in most cases, I won’t start him over a guy whose performance (either minor league, major league, or some combination thereof) suggests he will be better than league average.
Pretty much ditto, except I’d just ditch that 21 AB “hot streak” altogether.
Let’s face it: the best reason to play Bloomquist at all is he is on your roster, and he has to have some amount of playing time to keep him fresh enough so that (God forbid) you have enough injuries where you have to play him, he’s somewhat ready. Given that, it probably makes sense to start him against guys he appears to have some comfort level against. But giving him 5-7 starts in a row? No.
#26. Agree.
To put it another way, if you want to keep your players fresh (a good idea, even if Tony LaRussa does run it into the ground a bit), why wouldn’t you want to keep a better player (Doyle) fresher than a worse player (Bloomquist)?
Here’s the reason: the standards are different for Bloomquist. Consider Doyle and the people WFB most often subs for (Morse, Reed). Say they are asked to do something and, say, they fail. Management + announcers will forever remember that and not all times they succeeded.
Bloomquist: Say he’s asked to do something and he succeeds. Management + announcers will forever remember that and not all the times he failed.
We’ve been assuming that the same perceptual/performance standards were being applied to everybody. Obviously we were wrong and we should feel a little foolish about not figuring it out a long time ago.
#18, no, i’m not. i said wfb should play until he 0-fers
i never said that he would continue to hit well. obviously he will stop at some point, and when he stops hargrove should stop playing him. but until then, why would anyone bench a player that is hitting well in a game against a pitcher they hit well.
lastly, the lottery comment is really off base. there was justification for the decision prior to the game.
#28 Really is odd because both Morse and Rivera were hitting well when they stopped playing them. Willie just gets held to a different standard.
morse went 1/18 before he was given any real time off
rivera started around 40% of the games (decent for a backup C) the entire time he was up.
i do agree that i’d much rather see those two play than willie though.
Yeah. Benching Morse when he was still hitting well was inane. It’s not just that they persist in playing WFB (or, say, Dobbs). It’s that they do this while utterly screwing actual prospects with actual upside out of playing time to do so. Morse, Doyle, Lopez might all be somebody some day, and they’re sitting down so WFB can play….
People keep saying Morse was benched when he was still hitting well. When Morse was benched, he had gone 3 for his last 25. He had stopped hitting and was falling quickly back to earth. Just because his overall average was still .330 doesn’t mean he was “still hitting well.” Some of you won’t be satisfied till he hits the earth with a thud you can feel in your teeth.
Is Pete Nye any relation to that Bill Nye the science guy character that used to do all those science vids back in when I was in school and looked like he was using his “lab” (set of the show) to make speed or supercaffine or something? I do remember he was from Seattle. Was he an actual scientist (or teacher or something) or just an actor?
Sorry to interrupt, just curious.