Handicapping Reality

April 28, 2011 9:00 PM

How do I learn and practice handicapping skills?

I just got through Buffett's biography, "The Snowball" and the main narrative is that Buffett is a master handicapper.

I'm aware that this area is within the realm of actuarial sciences which includes statistics and probabilities. I've never ventured beyond high school Algebra, but have basic understanding of how probabilities work.

So my question is twofold:

1) What branch of mathematics do I have to learn and how do I work my way up to it from H.S. Algebra. Any good free, guided online resources?

2) How the hell do you handicap reality? How do you assign values to all the possibile outcome in probability space. Like for instance, when people say that there is 1 in 100 chance of a nuclear war, how do they come up that figure? Especially when there is limited historical data.

posted by pakoothefakoo to education (7 answers total)

One skill to practice is that of making educated guesses about things where the answers seem impossible on first look. Fermi problems are a type of problems used to teach physics students this skill. "Estimate how many hairs are on your head" - seems impossible, but suppose we give it a try, how would you begin to tackle that problem? Maybe by estimating how many hairs in a smaller region, and then how many of those smaller regions make up your head. (etc) This can help you get in the problem-solving habit of breaking things down into more manageable units.

Kahn Academy offers free video tutorials on math topics including probability and statistics. He also has introductory business videos, which might be useful to you as well.

Also, as I understand it, Buffet follows a style of investing that's called value investing, which involves finding out a LOT of information about the companies he invests in. So he is not handicapping in the dark or going by his gut.
posted by LobsterMitten at 9:33 PM on April 28, 2011

Read this book: Predicting Presidential Elections and Other Things, by the Yale economist Ray Fair who has made a very successful election model. It's ALL about using historical data. I don't know who is predicting a 1% chance of nuclear war but I highly doubt they're using real statistics.
posted by acidic at 9:41 PM on April 28, 2011

Excellent point - elections and baseball statistics (especially the modern development of "sabermetrics", where you could read the book Moneyball by Michael Lewis for a fun introduction) are two great areas for detailed practice examples. A lively interesting writer on both of these topics is Nate Silver, who writes a blog called 538 (for the number of presidential electors, originally covered the 2008 Presidential race) which is now housed by the New York Times.
posted by LobsterMitten at 9:50 PM on April 28, 2011

And there are so many equally valid ways to make predictions. Ray Fair and Nate Silver both make amazingly accurate predictions, but use totally different methods. Fair uses old voting data and economic info such as GDP; Silver synthesizes various polls.

So aside from learning statistics, you should also try to develop a really good understanding of data. The best way to do this is to read papers and identify the mistakes in them.
posted by acidic at 10:21 PM on April 28, 2011

There is nothing Warren Buffett is doing that is more than simple arithmetic with maybe punching some numbers into a financial calculator to do present value calculations.

Value investing "handicapping" isn't really like handicapping a horse race or an election. Its handicapping in the sense of determining how large a margin of safety you require for a given investment, and knowing when you don't know something.
posted by JPD at 5:03 AM on April 29, 2011

Thanks for all the responses. Very useful.

There's a lot of reference Buffett and Munger's talks regarding investing being like a pari mutuel system (which I don't understand 100%). Figuring out your odds and placing bets if it pays 3 to 1 with 50-50 odds (or something) I just wonder how they figure out these "odds," and how to acquire those skills.
posted by pakoothefakoo at 7:08 AM on April 29, 2011

yeah - but that's not really statistics - its closer to scenario analysis. At least the way it is practiced.

find a copy of Seth Klarmann's book as well.
posted by JPD at 7:54 AM on April 29, 2011