THE BOOK--Playing The Percentages In Baseball

Only issue I see is that some people just don’t want to believe that Ortiz, or whoever, isn’t spectacularly clutch.  For most people, it’s more fun to believe that their lucky unwashed Varitek shirt is contributing to the Red Sox success, and that David Ortiz legitimately becomes a .700 hitter in the bottom of the 9th while JD Drew becomes a .050 hitter.  So those folks - the ones you’d hope would be convinced by this - would be unlikely to take the bet because ignorance is just too bliss.  I like the generosity though, that’s great.

They can believe the Clark Kent to Superman heroics all they want.  Just keep it to themselves.

They may believe the Pats will blow the Giants out of the water, but they certainly won’t accept a 24 point line, will they?  They have *some* doubt the Pats won’t be that good, right?  Given a 24 point line, they’d bet *against* the Pats, right?

This is all a question of probabilities of true talent (the mean) and the random variation around that mean (luck).

Even if the Pats have a, I dunno, 80% chance of winning the game (1:4 odds), you are not going to bet if you are given 1:9 odds (90% chance of winning).  Even if you believe that the Pats are the far better team, there is always random variation to contend with.

And that’s the clutch argument.  Even if Ortiz is a clutch hitter (and he wasn’t in 2007, but was by far in 2005/06), random variation can still sink you.

What’s clear when I read the posts here:
http://www.baseballthinkfactory.org/files/newsstand/discussion/the_book_blog_mgl_clutch_and_other_projects_redux/

Is either:
a) people didn’t read the chapter in The Book that dealt with streaks

and/or

b) state their assertions but aren’t willing to back it up

Tango, have you done any follow-up on your clutch project?  Are you going to have a web site as with your fan fielding projects?

Can anyone set up a web site for the hot/cold project?  I’d really like to see that happen.

Any other suggestions/ideas?  Phil (B.)?

David Pinto is offering his services for constructing a hot and cold web site which can be linked to from his site.  I am trying to figure out the details of the methodology (e.g., log in and choose a player and whether he is hot or cold.  At some point take player off hot or cold “list.").  Any suggestions would be welcome.

I’ll be running the Clutch Project during Spring Training.  It’s an easy one to setup, because of the way I’m setting it up (they select one Clutcher, I’ll select one great hitter, per team).

For the hot/cold, that’s alot more work.  The matchups one is also a bit of work.  If someone wants to do the coding for you, we can host it here.  I’m thinking if you can partner with Forman at b-r.com, that might make more sense, since he’ll have the real-time stats that the hot/cold would necessitate.

No matter how simple you make the design, and how many times you explain, many baseball ‘analysts’ in the mainstream media simply won’t understand the methodology, and will dismiss it. ‘Baseball is played by people, not by numbers’. No matter how clearly it is demonstrated, many of them simply won’t participate/listen. That’s the only flaw in this project, I don’t think you’ll get the people you’re trying to reach (Gammons, Joe Morgan, Phillips, etc.) to participate, or even listen to the results after the fact.

Bobby, I pretty much agree.  But at least people will be able to cite a “real time” rather than a retroactive study.  The biggest problem I see is sample size.  I have not sat down (I don’t know if Tango or anyone else has either) and figured out what kind of sample sizes are likely to be generated, and thus the likely sample errors.  That might be a problem.

Also, while I am pretty sure that there will be little or no differences in clutch/non-clutch, I am curious about the batter/pitcher matchups and hot/cold streaks.  For one thing, baseball people are CERTAIN that they can tell you which batter/pitcher matchups are better than others.  The results will be pretty simple to understand.  We can use, BA, OBP, OPS or whatever, adjusted for platoon advantage.  I am even willing to forgo adjusting for the G/F platoon effect.

Tango, why would you think the batter/pitcher matchups would be hard?  We can ask each person to name 5 (or one, or whatever) good and one bad batter/pitcher matchup (from the standpoint of, say, the batter) and then tally up the results at the end of the season, just like with the clutch project.  It would be nice to do it on b-r, or in conjunction with them, as I suppose their engine could give you the batter/pitcher results all through the season (would that need a “pi” subscription?).

The hot/cold thing is tricky but it would be fascinating!  Again, is there virtually ANY baseball person who does not think that they can identify when SOME players are “seeing the ball well” or “pressing”.  Heck, I’m even willing to let them choose injured players as “cold” (maybe).  Their biggest criticisms of retrospective hot and cold studies, like that in the book (by the way, they do show a very slight predictive value, as with clutch, for batters, and quite a large one for pitchers, if you haven’t read The Book or have forgotten), is that, “You don’t know in retrospect when the streak has started or ended (a silly argument, but let’s leave that aside), and you can’t necessarily identify a hot or cold player from his stats.  You have to SEE him to do that.”

Well, here is the opportunity for everyone to SEE the players and to identify when the streak has started and/or ended!  And, like the batter/pitcher matchups, the results should be easy for anyone to understand.

What are the proponents of hot and cold players going to say if their hot and cold players hit the same as they always hit (as measured by some long-term numbers) whether they are on the hot or cold list?  I guess they will say that everyone else screwed up THEIR numbers since all we can do is combine everyone’s “picks” to get a large enough sample size to have any meaning.

Another (big) problem along those lines (blaming everyone else) is going to be that lots of individual people will of course have GREAT results, be it clutch, batter/pitcher, or hot and cold.  All of them are going to claim, “Look I knew how to identify the (clutch, hot, etc.) players, everyone else didn’t!” How are you going to convince them that their results were likely the result of “selective sampling?” They won’t even know what you are talking about.