When pitching investors, you normally don’t want “hedge fund" and “earthquake” in the same sentence. But Bill Miller, already known for quirky investing methods, is starting a hedge fund that will make bets based in part on a computer model designed to predict natural disasters.

Called Seismic Value Partners 1, the hedge fund marks Miller’s first foray in that business after decades managing mutual funds at Legg Mason. He won approval last month from the U.S. Securities and Exchange Commission to open Miller Value Partners, a money management firm that will oversee his hedge funds.

Miller has licensed the model from OpenHazards Group, a Davis, California, company run by engineers, mathematicians, scientists and business people. While many experts are skeptical, the idea is to apply the mathematics of forecasting the probability of seismic activity to the chances of a stock market crash.

“We have used our experience derived from doing physics to take what we hope is a unique approach to modeling financial markets,” said OpenHazards Chief Executive Officer Bill Graves, a physicist who served as the first head of Cisco Systems Inc.

If it works, the model could help Miller dodge the type of cataclysmic loss that he suffered during the recession when he was caught holding too many financial stocks. His Legg Mason Value Trust plunged 55 percent in 2008, tarnishing a reputation earned by beating the Standard & Poor’s 500 Index for 15 consecutive years.


‘Extreme Events’


Like stock market crashes, “earthquakes are extreme events,” said Lisa Rapuano, who helped Miller run the Legg Mason Special Investment Trust during the 1990s. “This has the potential to help Bill avoid big mistakes, which is one of the downsides of his investment style.”

Miller, 66, declined to comment.

Some scholars dismiss the idea of applying an earthquake model to stock-picking.

“I don’t think it has diddly to do with financial markets,” said Joseph McCauley, a physics professor at the University of Houston who has written several related books, including “Dynamics of Markets: Econophysics and Finance,” published by Cambridge University Press in 2004. “I’m surprised somebody is still messing with that stuff.”


Trading Algorithm


Open Hazards Chairman John Rundle, a well-known seismologist who teaches physics and geology at the University of California of Davis, has been adapting his theories on natural disasters and financial markets with help from members of the Santa Fe Institute, a non-profit research organization that Miller has been involved with for decades.

Miller’s fund will use signals from the model to make daily determinations on whether to purchase or sell short securities that mirror the performance of the U.S. stock market, such as the SPDR S&P 500 ETF Trust, according to regulatory documents. The goal is to outperform the S&P 500 over periods of a year or more with less volatility.

Miller will likely employ the trading algorithm as an “overlay” to individual stock picks, Rapuano said. That could protect his hedge funds when markets are sinking and add to their returns when markets soar.


Mass Extinctions


The concept that seemingly unrelated catastrophes, ranging from mass extinctions to avalanches and stock market crashes, might follow similar underlying patterns dates back to the 1980s, when Danish physicist Per Bak began writing on the subject. Bak’s concept helped spawn an industry known as econophysics and triggered thousands of academic papers on the subject.

“It’s something everyone would like to do, but it is inherently close to impossible,” Gene Stanley, a physics professor at Boston University, said about trading stocks on the model.

Miller has long been a value investor, finding stocks that trade at a discount to their intrinsic value. But he’s not above relying at times on scientific theories rather than Wall Street’s traditional yardsticks to ferret out good investments. Some of his most profitable ideas came in part from the Santa Fe Institute where he is chairman emeritus and the largest donor.


Amazon.com Inc.


He was an investor in Amazon.com Inc. at a time when most value managers thought the company was overpriced, using standard industrial-growth metrics. Miller and Rapuano developed a model for how Internet businesses might grow based on network theory that they had picked up through the Santa Fe Institute, Rapuano said. They also received help from Doyne Farmer, an external scientist at the organization who explored chaos theory to make bets on Wall Street.

Miller’s contrarian approach worked for years as his Legg Mason Value Trust fund annually outperformed the S&P 500 from 1991 through 2005. But it plunged in 2008 and trailed the index in two of the three ensuing years, leading to investor redemptions. Miller stepped down as Value Trust manager in April 2012. 


Larger Losses


Miller rebounded running the $1.3 billion Legg Mason Opportunity Trust, generating average annual returns of 26 percent from 2012 through 2015. This year, he has once again suffered larger losses than the broader market, with the fund plunging some 28 percent so far in 2016, compared with a 9.2 percent decline for the S&P 500 counting dividends.

Miller no longer works at Legg Mason, and runs the Opportunity Trust through LMM, a Baltimore-based investment adviser that he owns along with his son John and his former employer. LMM filed to set up several mutual funds under the Millers’ name in 2013 and on Jan. 28 received clearance from the SEC to open Miller Value Partners, according to the agency’s website.

Madelyn Dillabough, a Legg Mason spokeswoman, said in an e- mail that Miller’s arrangement with LMM remains unchanged.

John Seo, co-founder of Fermat Capital Management, a Westport, Connecticut, firm that invests in catastrophe bonds, said he could envision a broader use of the earthquake modeling. While Wall Street traders and companies often have extensive data on how their portfolios would perform in a market crash, they seldom know how likely such a downdraft would be.

“That’s where the earthquake guys come in,” Seo said. “They are good at assigning probabilities.”