The Master of Money: "The Snowball: Warren Buffett and the Business of Life" review
tnr.com
tnr.com
Oh, wow. I thought Lewis was in finance, but apparently he doesn't realize that the peak-to-trough performance of one member of an average is very likely to be higher than the peak-to-trough performance of that average. This is true, even with a two-stock index:
Year Stock A Stock B Avg.
0 10 12 11
1 12 10 11
2 5 5 5
Note that A and B both had a larger peak to trough loss than the average. And yet, the average consists of A and B.e railed against derivatives as weapons of mass destruction, and now turns out to have been sitting on a $68 billion pile of credit default swaps and exotic put options on various stock market indexes.
He complains specifically about counterparty risk. Someone who read what he said would know that; someone who heard thirdhand quotes might not. The derivatives he holds don't have any counterparty risk for him, because he's getting all the cash up front and paying out based on future events.
In other words, perhaps part of his success belongs to a bygone era that won't come around again.
This is drastically different from "quant analysis", which tries to determine a stock price, or engage in arbitrage, from previous movements, relations to other stocks, and a host of other data largely irrelevant to computing the implicit value of a company. This information comes from real-time data streams from god-knows-where.
To the extent that anybody was a quant back then, Graham probably counted: he did use lots of statistical rules, and he didn't use very much else.
another good book on buffett which has some sections dedicated to buffett vs quants and buffett vs 'its all luck' ppl is "The making of an American Capitalist" (http://www.amazon.com/Buffett-American-Capitalist-Roger-Lowe...) -- first book I got on him, perhaps the best (got snowball, haven't read it yet).
What would be really great would be a wikified list of his transactions, linked to the financial statements and news stories that were current at the time.
By looking at certain bits of data, they attempted to find these 'cigar butt' companies - Graham more than Buffett. The way they did that was to comb through a lot of data in a way that can now be done by a computer extremely quickly. So perhaps 'quant' is wrong... shall we say 'data driven'? Of course that's not all Buffet did/does, but it just struck me as an edge that he had then that he couldn't have now.