Volatility represents risk for most investors. For Intech Investment Management, it’s a source of return. Adrian Banner, Intech’s CEO, says his firm doesn’t pick stocks, read brokerage research, look at fundamental data or forecast individual stock returns. Instead, Intech’s team starts with a universe of stocks, estimates their volatilities, and uses those numbers to come up with an optimized set of portfolio weights. Each week, they rebalance the weights.

Banner says the firm’s approach to investing is based on research started more than 30 years ago by its founder, mathematician Robert Fernholz. “In 1982, he first had the realization that one could build portfolios that should be more efficient than a cap-weighted equity portfolio—without picking securities,” Banner says. “It’s possible to do that using only the volatilities and the correlations between stocks.”

While that idea may still sound novel today, there’s a lot of solid math behind it, Banner says. “The challenge was turning that math into real portfolios,” he says.

Intech started doing that in July 1987, when it launched its flagship Intech U.S. Enhanced Plus portfolio. The $13.4 billion strategy, which invests in Standard & Poor’s 500 Index stocks, returned 10.35% a year on average net of fees from July 1, 1987, through August 31. That was 1.05 percentage points better than the 9.3% annual return of the benchmark over the 28-year period. Gross of fees, the strategy outperformed by 1.45 points. Intech oversees about $51 billion in assets, mostly for institutions. The 86-person firm is a subsidiary of Denver-based Janus Capital Group.

When Banner, 40, was growing up in Sydney, Australia, he was equally interested in music and math. He earned a Ph.D. in math at Princeton University and joined Intech as director of research in 2002. A jazz pianist, he sometimes plays before meetings at the firm’s West Palm Beach, Fla., headquarters and has performed at the Montreal International Jazz Festival.

Banner says Intech’s basic insight is simple. Consider a stock. Over a couple of years, its trend may be up. Yet along that general course, the stock will tend to make a variety of moves—zigzags above and below the trend. “So that’s volatility,” Banner says. “And it can be harnessed—if one is willing to trade.”

The swings offer opportunities to buy low and sell high. The way to do that, Banner says, is to first establish a policy for weighting the stocks in a portfolio. Then, as prices move, rebalance back to the weights. Rebalancing in essence buys low and sells high, Banner says. “You’re selling previous winners, which by definition are high, and you’re using that capital to buy previous losers,” he says.

How much juice can rebalancing add? Intech has worked that out in its math. The key to it is compound return, Banner says. Let’s say you buy a stock for $100 and hold it for two periods. In the first, it gains 25%. In the second, it drops 20%. The average of those returns is 2.5%—pretty good.

The problem is that because you’d hung on to the stock over the two periods, you actually ended up right where you started, with a return of zero. The stock rose to $125 at the end of period one and dropped back to $100—(20% of $125 is $25)—at the end of period two.

Compound return “is what you actually experience with buy and hold,” Banner says. In effect, volatility is a drag on compound return when compared with average return. Mathematically, compound return is approximately equal to the average return minus one-half of the variance. (Variance, a measure of how widely a set of returns is spread around their mean, is the square of volatility.)

You can break what Banner calls the “curse of compounding” if you have more than one stock to work with. “What if you’re allowed to dynamically allocate capital between two or more assets?” Banner asks. The compound return of a portfolio, like that of a single asset, is the average of its returns minus one-half of its variance. Thanks to modern portfolio theory—which Nobel laureate Harry Markowitz set out in 1952—you can also describe the portfolio return as equal to the weighted average return of its constituent stocks. “Now, the plot thickens,” Banner says. The individual stocks tend to have higher volatilities than the portfolio as a whole: So the average stock variance tends to be larger than the portfolio variance.

If you rearrange the terms of a couple of equations, Banner says, you end up with an expression that shows exactly how much excess return you can get from rebalancing: It’s equal to one-half times the weighted average stock variance minus the portfolio variance. “That’s essentially the math formula that shows how the variances and covariances contribute to the compound return,” Banner says.

Rendered with a gamma and a couple of other Greek letters, it’s also the master formula of what Fernholz called stochastic portfolio theory. Fernholz originally got there by combining stochastic calculus—“that branch of math that studies changing quantities like calculus but with a probabilistic component: hence the fancy word stochastic,” Banner says—with modern portfolio theory.

That may seem daunting. For Banner, though, it’s the alternative that’s difficult: picking stocks that outperform. “We know that’s very hard to do,” he says.