That's because the market generally prices volatility at a premium relative to the actual level of volatility in market history. In other words, volatility is generally priced too high compared with the level that's likely to prevail. Selling volatility, as a result, allows traders to capture the premium. Presumably, buyers of volatility are willing to pay a premium for the same reason that people buy home insurance and understand that there's a price for protecting assets.

For example, the VIX's implied estimate of S&P 500 volatility has averaged roughly 19% from 1990 to 2007, according to a study by Bjorn Eraker, a finance professor at the Wisconsin School of Business ("The Volatility Premium.") Yet the historical annualized standard deviation of the S&P 500 has been much lower during that period: roughly 15.7%. The 3.3 percentage point difference suggests that the expected premium for writing options (selling volatility) on the S&P 500 is "substantial," Eraker notes.

As for the short volatility positions in Emerald and Astro, the more immediate factor is the historical bias for mean reversion in stock returns in the short term. To the extent that this reversion endures, daily volatility is likely to exceed weekly volatility. If so, the long daily/short weekly volatility spread will be positive, which might mean a modest positive return.
Since mean reversion is crucial for the strategy, it's reasonable to wonder if it's likely to persist. A number of studies tell us that the evidence is encouraging. An early example is Werner DeBondt and Richard Thaler's 1985 study, "Does the Stock Market Overreact?" (in the Journal of Finance, July 1985). Yes, they answer. Scores of subsequent studies report similar results.

But let's be clear: Emerald and Astro are counting on mean reversion during relatively short periods. There's plenty of supporting evidence here as well, including a 1990 paper that shows that stock portfolios with positive returns in one week "typically" suffer losses the next, and vice versa ("Fads, Martingales, and Market Efficiency," by Bruce Lehmann, in the Quarterly Journal of Economics, February 1990).

Faith In The Past
The track records for these strategies are limited, but recent history suggests that mean reversion works. The Emerald index, for instance, compares favorably with the stock market from the period just ahead of the late-2008 crisis (Figure  2). Emerald has outperformed the S&P 500 by a wide margin for the four years through the close of this past May.
It also acted as a hedge when stock prices sank and volatility surged in the crisis. And Emerald did not give up its gains once the market stabilized (Figure 2). Such a wide divergence won't prevail over time, but recent history at least offers some perspective after a period of extreme market volatility followed by relative stability.

But this is no free lunch. A key test of Emerald was an extraordinary eight-day run of nonstop losses a few years ago for the S&P 500, which appears only as a brief drop in Chart 2. This was arguably the worst point of the crisis, cutting the S&P by nearly 23% in the eight trading days through October 10, 2008. Emerald also took a dive, giving up nearly 15% in that stretch.

Mean reversion, in sum, was MIA for eight days. Keeping the faith in this hedge required steely discipline and a strong conviction in mean reversion's powers of perseverance. As it turned out, mean reversion revived on the ninth day with October 13's powerful single-day 6.8% rise in the stock market.

This past summer witnessed another stress test, when the S&P 500 swooned during a white-knuckle ride that left stocks lower in 10 out of 11 trading days in late July and early August. Once again volatility's mean reversion disappeared for an unusually long stretch, pushing the strategy's viability to the limits along the way. But the selling wave eventually subsided, and volatility's "normal" daily/weekly relationship revived on the 12th day.

How does one muster the confidence that mean reversion will continue to triumph? A fair amount of data crunching of the historical record helps. "We got the formula [for Emerald] and programmed it ourselves," says Miccolis. "We looked at how it behaved. If we can't reproduce it in our own spreadsheets, we don't trust it."

Scarpo's firm also put the numbers through extensive testing, reviewing how Emerald's subsequent three-year results looked from each and every date from 1998 onward. Of the several thousand start dates, only one-half of 1% of the three-year results for Emerald turned in a lower return compared with owning the S&P 500 on its own, says Kevin Comstock, a research analyst at Private Wealth Advisors.