Lakewood Ranch-based StatWeather helps companies predict future weather...and profits. By David Ball; photo by Lori Sax.
Every company tries to prepare for that rainy day when a freak event or market change can send it reeling. But for businesses like hedge funds, energy traders, agribusiness and utilities, that potential disaster can be a literal rainy day, heat wave or hurricane. Unpredictable weather—made more extreme by climate change—can spike energy costs, increase fuel demand and decrease production.
Enter Ria Persad, 39, a former child math prodigy, whose Lakewood Ranch-based company StatWeather is at the bleeding edge of an industry that predicts the weather.
By analyzing 120 years of weather data with proprietary computer models and statistical theory, StatWeather can predict with impressive accuracy the probability of a daily weather change or extreme weather event in a specific location or region up to three months in advance. Weekly changes can be predicted up to a year.
Compared with the traditional 14-day outlooks from the National Weather Service and other forecasters, the advantages are obvious. If an energy company can predict a hurricane strike, it can stock up on commodities like oil or natural gas while prices are cheap or sell excess reserves when the price goes up. If a utility knows when a cold snap is coming, it can prepare for increased energy usage from heated buildings. If a farmer sees a heat wave approaching, he or she can harvest early or plant late to maximize crop yield.
This weather hedging, as it’s known to risk managers and the financial industry, is nothing new, but climate change has made the industry even more competitive; Persad’s company is among a young group of start-ups providing long-range weather prediction using advanced, computer-based statistical models. StatWeather is earning a strong reputation with mentions by Forbes magazine and the Wall Street Journal. The business, launched in 2012, was named “Newcomer of the Year” by London-based Energy Risk magazine this summer.
“Energy Risk considers the 2013 Newcomer of the Year Award to be a significant validation of StatWeather’s strategy, accuracy and product execution,” Energy Risk editor Mark Pengelly wrote in a release accompanying the award. “During an 18-month period, StatWeather had a 74 percent accuracy rate, in contrast to most of the company’s reputable competitors, which had a 30 to 40 percent accuracy rate as determined by Energy Risk.”
That accuracy begins with the woman in charge. Persad was born in Trinidad and Tobago and immigrated with her parents to the U.S. as a small child. She was schooled in Boston, where her aptitudes in math and music became apparent early on (she still performs on piano under her married name of Ria Carlo). As she entered high school, the math department chairman enrolled her in an accelerated curriculum, though he died just a week after Persad began her freshman year.
“I was struggling hard that first year, but because this man saw something in me, I just did everything I could to do what he wanted me to do,” Persad says. “It was one of his last dying wishes to see me succeed. If it wasn’t for him I’d probably be a lawyer.”
Persad breezed through her school’s entire advanced math curriculum, and, at 14, she enrolled in classes at Harvard University. When she was15 she entered a contest at the prestigious Massachusetts Institute of Technology and won first place by developing new calculus theorems that solved previously unsolvable problems. She entered more theorems the next year and won another first-place prize.
By 17 she landed an internship at the Lawrence Livermore National Laboratory in California developing climate models with supercomputers—a still-novel concept in the early 1990s. “That was the beginning of me and weather and supercomputers,” she says. “I was hooked.”
Study in math and astrophysics at Princeton University was followed by graduate work at the University of Cambridge and Rice University. Various research stints around the country, including one at NASA, eventually led her back to forecasting, first for investment bank Lehman Brothers, followed by energy giant Enron and then Duke Energy in Houston.
“It’s where very advanced science met Wall Street,” she says. “You have some really excellent scientists doing groundbreaking work, all in the financial industry.”
Persad left her job a year before Enron’s collapse from widespread accounting fraud in 2001, a scandal that spilled over into interdependent companies like Duke Energy. She started contracting with companies, including Goldman Sachs, as a forecaster, crunching data with her self-developed methods and formulas. In 2008 her fiancé, Mark Carlo, a chemical engineer and University of South Florida graduate, suggested she go into business for herself and develop forecasting products she could sell to customers around the world.
“My husband took my prototypes and spreadsheets and, with the help of another programmer, turned them into really nice software,” Persad says. “Someone can just type in the name of a city and get tons of information about what’s going to happen on any given day along with the confidence level and margins of error.”
Bayesian probability furthers drive the accuracy of her forecasts. The Bayesian model, which looks at an entire history of a data set but gives more weight to recent trends, has become one of the hottest topics in data thanks to Nate Silver, The New York Times journalist and statistician who used the method to correctly predict the 2012 presidential election. Silver was recently recruited by ESPN to perform the same analysis with sports statistics.
“In simple terms, that would mean a city that sees a storm during the same week in nine out of every 10 years would have a nine in 10 chance of suffering another during that same week in future years,” Pengelly writes in Energy Risk. “As well as being accurate, this method delivers results that are presented in ways useful to traders—including probabilities, confidence intervals and a comparison with consensus forecasts.”
Persad incorporated the business in 2009, but it took until 2012 to develop her first true forecasting products, which she brought to a risk management conference and gave to attendees to use free for a month. She challenged them to check her forecasts with the actual measurements to see if they held up. At the end of the month, Persad had her first clients.
“We were making our bottom line dependent on whether these forecasts would come to pass,” Persad says. “We predicted things like the fall storms [among them, Superstorm Sandy] which were rare, and it validated our methodology. The industry was sitting and watching this and people were making money off of these forecasts.”
How much money can her clients make or save? Well, no client wanted to speak with Biz(941) on the record, as even the knowledge of a company as a StatWeather client could threaten its market advantage. However, Persad says a large chemical company was able to use StatWeather’s forecast to correctly predict a temperature spike in late June that saved the company what was likely hundreds of thousands of dollars.
“If you’re off by just one or two degrees or you get the timing wrong, they could lose that money,” Persad says. “One client said our weather forecasts actually pay for [themselves] in a week.”
StatWeather’s forecasts and products range from $300 to $4,500 a month based on the size of the client business. Packaging multiple forecasts can cost up to $10,500 a month. Similar subscription services are used by companies across the country. In San Diego, EarthRisk Technologies helps Spanish energy giant Iberdrola plan its natural gas acquisitions. EarthRisk only gives forecasts up to 40 days out, however, in comparison to StatWeather’s 90 days.
That difference has led to a doubling of StatWeather’s clients this summer, and Persad expects to double again in just a few months. She wouldn’t say exactly how many clients she has, although she says the company added 80 farmers in one day during a recent agriculture trade show. Persad says sales agents in Mexico are working to sign 10,000 farmers to use StatWeather services.
This growth has all occurred since fall 2012, when Persad moved her company from Tampa to the state-of-the-art Energy Court Center in Lakewood Ranch. She says she was attracted by the building’s innovative and eco-friendly amenities, and also the quality of life in the Sarasota-Manatee area and its ability to attract talent.
“People don’t realize that you have some great software developers and programmers here,” Persad says. “People come for the lifestyle, but it’s not just retirees; it’s also young professionals who want to start family here.”
StatWeather’s Lakewood Ranch-based staff of six includes highly skilled programmers, computer modelers and meteorologists, many hired locally. Four other employees forecast from remote locations around the country. Persad says she expects to hire another person this year and to continue to grow as her customer base expands, with a target on international clients.
“Our company is bootstrapped,” Persad says. “We are not based on big loans or third-party investments. My husband and I provided the seed money, and we grow organically as our revenues grow.”
StatWeather Florida Forecast
Good chance for a hurricane
StatWeather expects at least one major hurricane to make landfall this year with a split probability of hitting either the North Carolina coast or the Miami/South Florida coast—making a Florida strike a statistically high 50-percent chance. The highest risk time looks to be early September.
Wet and cold this fall
In July, Florida was two-to-three degrees cooler and up to 50 percent wetter than in 30 years. StatWeather sees cooler and wetter weather continuing through November. The wettest parts of the state are most likely to be northern Florida, the Alligator Alley corridor and near Miami.
Cooling trend for Florida
Over the last 100 years, much of the U.S. has been warming, though the Southeastern U.S., particularly Florida, has had a substantial cooling trend. One hundred years ago, temperatures in Florida ran a few degrees warmer on average. Scientists attribute Florida’s cooling to changes in global circulation patterns.