Friday, September 4, 2009

Falkenstein, Finding Alpha

I have only flipped through Eric Falkenstein’s Finding Alpha: The Search for Alpha When Risk and Return Break Down (Wiley, 2009), so this will not be a review but rather one of those “takeaway” pieces. The central thesis of the book is that risk in general is not related to return. If true (and my instinct says it’s probably an oversimplification), it turns a lot of conventional wisdom on its head. On the other hand, and a more intuitive hypothesis, assets with either extremely low betas or extremely high betas have below average returns. The quintessential low beta example is money under the mattress. As an example of high beta assets he cites far out-of-the-money call options. The beta of calls, of course, increases the farther out of the money you go—that is, as call options essentially become lottery tickets. And beta moves inversely to the delta of the call. He reproduces a graphic from a 2001 study that shows the monthly returns for call options grouped into quintiles based on their deltas.

Another study, looking at data from 1996 through 2005, found that the highest out-of-the-money calls with one month to expiration lose on average 37%. “If there’s a risk premium in equities,” he concludes, “it certainly is not amplified in options in any way, because you lose money, on average, buying leverage market positions by way of call options.” (p. 76) Falkenstein’s argument is flawed because the investor buying a call option is not simply leveraging a position the way he would with a margin account at the applicable margin loan rate; he’s paying a premium to buy a leveraged position. This cuts into potential profits and, indeed on average, tilts toward losses. But this isn’t evidence against a risk premium in equities. Rather, it’s a cautionary tale for those wanting to buy call options.

As people search for alpha they try to find anomalies in the markets. Falkenstein claims that most so-called anomalies don’t really exist, that they are the product of flawed research. For instance, a 1985 study showed that for each year since 1933 going long a portfolio of the worst performing stocks over the previous three years and short the best performing stocks over the same three years would generate an annualized return of 8% over the next three years. This study, like so many that purport to uncover anomalies, produced positive results only because of the tendency of low-priced stocks to move from their bid to ask price. The purported calendar anomalies (the weekend effect and the January effect) apparently suffer from the same fault—the bid-ask bounce in low-priced stocks.

In looking for an alpha strategy Falkenstein says that you should start with the basic metrics of the assets in question—their average returns, their average annualized volatilities, and their cyclicality. The Sharpe ratio of any acceptable alpha strategy should not be radically different from the average Sharpe ratio. For instance, in their glory years convertible bond arbitrage generated Sharpes between 1 and 2. Thus, he argues, an investment strategy with a prospective Sharpe above 2 is undoubtedly the result of overoptimized backtesting and will fail.

No comments:

Post a Comment