David Justice. He was probably the best veteran we’ve ever had in our clubhouse. Unbelievable. He was worth every dime we paid him. ........... He was everything you would want. ....... There’s been other guys that people thought were team leaders that were not at all.


damn--he wouldn't name names

I think the best chance of winning the series in a five-game series, no matter who you are, is 55 percent.

I'm sorry, but Billy Beane is now an official idiot.

One of MGL's rules is that if you say something really stupid (and incorrect) and it is not extemporaneous (I assume that he had time to compose his responses), you cannot recover - you are officially a moron.

Beane shouldn't have to worry about how much he gets paid by the A's. He could just go to Vegas each year and bet on all the 2-1 and 3-2 and 2.5 to 1 underdogs in the 5 game series. For example, if the Yankees (or Bosox) play the Twins in the first round (I don't know if that is possible), they will be at least a 2-1 favorite and that would probably be about right. Ditto for the Cardinals versus the Giants or even the Dodgers.

The other moronic thing about that statement is that he is implying that the 5-game series is so much more of a crapshoot than the 7-game series. Of course that isn't so. If a team is a 55% favorite in each game (say, on the average), they will win 59% of the time in a 5-game series and 60.6% of the time in a 7-game series, hardly any difference at all...

Wow mgl, isn't that a little harsh on Beane. Idiot? How do you know that he had a lot of time to generate a response. The transcript reads like a conversation, with Beane laughing at one point. In a conversation one usually doesn't have time to look up stats. beane may be wrong here and in fact you are probably right, but damn. I think you are being a little harsh here.

Again no offense as you are smarter than I, but does anyone here see mgl as this dude with a really big head you looks around at everyone and gets frustrated by their lack of genius? Kinda like that guy in Princess bride?

Again I dont' really mean any offence.

You have your own set of rules? What are some of the others?

I agree that 55% is low. Should be easy to figure out though. Get a decent baseball simulator, set up a team with the Yankess stats, one with the Twins stats, allocate the playing time correctly, and let them have at it a few thousand times.

I'd say about 70-30 Yankees.

One of my rules is that people who refer to themselves by name, rather than the using the common or garden "I", "me" or "my" etc, are suffering from a form of mild insanity or delusion usually resulting from an excessive sense of self-importance and grandeur. It is particularly common among sports people. In fact, some pitcher did it yesterday after losing: "I'm not quitting. Jose Lima never quits. Tomorrow I'm going to come out and be a cheerleader for my team." I can't remember who it was though.

Actually, MGL doesn't do that. Only with respect to "MGL's rules".

Seriously, I don't mean that literally (that Beane is an "idiot") but it is quite illuminating that Beane would not know, even off the top of his head, that there are plenty of playoff matchups where one team is favored my more than 55%.

These interviews are usually conducted by e-mail, aren't they?

Actually, MGL doesn't do that. Only with respect to "MGL's rules".

Philistine is very glad to hear that but it still transgresses one of his rules.

Isn't there a picture there with Beane and the guy from AN? Also, how would the interviewer know that Beane was laughing? I guess he could email LOL. Plus he mentioned that he got a lot of 'time' with Beane. These are just hints that make me think that it was a sit down interview. I could be wrong.

And I must say I figured you would be really mad at me following my post. Kudos to you.

Why would I be angry? You're pretty much right! I couldn't stand to read the rest of the interview, so you are probably right about the interview being live as well...

I agree that 55% is low. Should be easy to figure out though. Get a decent baseball simulator, set up a team with the Yankess stats, one with the Twins stats, allocate the playing time correctly, and let them have at it a few thousand times.

I'd say about 70-30 Yankees.

Is this in the universe where Johan Santana wouldn't be pitching two of the games?

Can someone who knows how to do these things produce a logarithm or whatever it is to show us what the real chances are of teams winning five-game series? The Yankees beating Minnesota 70% of the time sounds absurd to me, but maybe it's right.

mgl, keep in mind that Beane isn't a neutral observer, and perhaps if he has convinced his boss that its just chance that he has lost 4 years in a row, that ever saying anything to the contrary might endager his job?

Of course, he'd still be wrong, he'd just be lying instead of being an idiot.

In a conversation one usually doesn't have time to look up stats. beane may be wrong here and in fact you are probably right, but damn. I think you are being a little harsh here.

Well, I don't think I'd call Beane an idiot, but it's really shocking to me that he would think that the chances are never better than 55%. I mean, with all the analysis his staff does, and all the times they've lost in the first round of the playoffs, you'd think they'd devote some time to analyzing playoff series and stuff like that.

Again no offense as you are smarter than I, but does anyone here see mgl as this dude with a really big head you looks around at everyone and gets frustrated by their lack of genius? Kinda like that guy in Princess bride?

I see MGL more as someone who looks around and gets frustrated by others' lack of pure objectivity.

I think it's great that we have someone like him around here--really, what would Primer be without one--and I usually find his uncompromising tone hilarious, even when I think he's being a little harsh.

Notation:
P(n) - probability that the yankees win game n.
P(!m) - probability that the yankees lose game m.
P(n!m) - probability that the yankees win game n and lose game m.
P(n|!m) - probability that the yankees win game n given that the yankees lost game m.

Observe that 1 = P(n) + P(!n) = P(n|m) + P(!n|m)

I'll do a three game series to illustrate, but the priciples hold for longer series as well, just more terms.

What is the probability that the Yankees win a three game series? They win the series if they win the first two games, or if they win one of the first two, and win the third.
= P(12) + P(!123) + P(1!23)
= P(1)P(2|1) + P(!1)P(2|!1)P(3|2!1) + P(1)P(!2|1)P(3|!21)

[Even this much is a simplification of the real problem, since the odds of winning the second game may depend on how the first game played - different odds after a pitchers duel than after a blowout.]

It is common to make the simplification that the games are independent events - that the events of the first game have a small enough effect on the outcome of the second that they may safely be ignored. In this case, all of the conditional terms reduce [ P(n|m) becomes P(n) ], and the expression can be written

= P(1)P(2) + P(!1)P(2)P(3) + P(1)P(!2)P(3)

Another common assumption is that the probabilities are similar enough from one game to the next that we need not fret over the difference - that P(n) = P(m). Rewriting again, using P(W) = P(1) = P(2) = P(3), and P(L) = P(!1) = P(!2) = P(!3)

= P(W)P(W) + P(L)P(W)P(W) + P(W)P(L)P(W)

With this set of assumptions, you are treating each game as an independent Bernoulli Trial. The order that we do the multiplcations doesn't matter, so we can simplify this a bit

= P(W)^2 + 2 P(W)^2 P(L)

There are a couple different ways to think about what this means; fortunately they all give the same answer (and are in fact mathematically equivalent - though it's a ##### to demonstrate).

For me, the most intuitive is to consider that the yankees win if they win one game and the last one.

= [ P(W) + 2P(W)P(L) ] * P(W)

I introduce the symbol (NcM) to denote the number of distinct ways you can choose m objects from a collection n. It happens that
(NcM) = N! / ( M! (N-M)! )
so we can rewrite the expression above as

= [ (1c1)P(W) + (2c1)P(W)P(L) ] P(W)
= [ (1c1)P(W)^1 + (2c1)P(W)^1P(L)^(2-1) ] P(W)
= Sum [N: (Nc1)P(W)^1P(L)^(N-1) ] P(W)
= Sum [N: (Nc1)P(L)^(N-1)] P(W)^2

For best of series, the rule is that you must win N out of 2N-1 (in this example N=2 ). So the general expression is

= SUM [M: ((M+N)c(N-1))P(L)^M] P(W)^N
where you sum the terms for M=0 through M=N-1.

Another common assumption is that the probabilities are similar enough from one game to the next that we need not fret over the difference - that P(n) = P(m). Rewriting again, using P(W) = P(1) = P(2) = P(3), and P(L) = P(!1) = P(!2) = P(!3)

You really really need two assumptions for P(n)--one for home games and one for road games. One of the primary reasons that the probabilities in a 7-game series are not much different from those in a 5-game series is that the home field advantage gets diluted from 3/5 games to 4/7 games in the former.

Since a seven game series isn't really that much better than a 5 game series in terms of having the best team win, what is the difference betweena nine game series and a five game series? In other words what do people think of the playoffs being 7/9/9 instead of 5/7/7. I guess it isn't a big difference, but more games is always fun, it does make ahe games in. Plus it may make the 5th starter pitch a game or two. One advantage some teams have is that they don't have to use that last starter.


And more playoff revenue!!!!

David Justice. He was probably the best veteran we’ve ever had in our clubhouse. Unbelievable. He was worth every dime we paid him. ........... He was everything you would want. ....... There’s been other guys that people thought were team leaders that were not at all.

damn--he wouldn't name names

Rhymes with Shason Shiambi.

On the site it mentions that the interview was in person and transcribed from a recording, so the comment was extemporaneous. MGL's point still stands that even off the top of his head Beane should know better.

The article will not tell any Primate anything they don't already know about baseball. It is, however, a very in-depth piece with Beane and a real treat for A's fans like me.

Voice of Unreason - Does discounting the varying starting pitchers and parks, the asumption that p(1) = p(2) = p(W), make this method meaningless for use in predicting the likihood of winning a baseball playoff series? Thanks in advance for you thoughts.

Rhymes with Shason Shiambi.

Or possibly Shiguel Shejada? Mr. Swing at Everything himself?

I have no idea, just speculating.

MGL's point still stands that even off the top of his head Beane should know better.


Why??

So dorks can like him more?






(I'm a dork too)