The Greek philosopher Thales of Miletus is said to have become the first options trader based on an accurate weather prediction. Expecting a good harvest, he cornered the olive press market and ended up making a fortune. Weather science has come a long way since ancient Greece and so too has its intersection with markets.
Weather forecasting may be both art and science, but the artistry of it can be quantified, says Stephen Bennett, founding partner and chief science and products officer at EarthRisk Technologies.
Bennett's experience as a meteorologist includes a stint with Enron Corporation, the collapsed Houston-based energy company, which then led him to take on a role at hedge fund Citadel. Citadel, he says, applied a highly quantitative, analytical approach to weather and EarthRisk now uses that mind-set in its weather forecasting approach.
Throughout his career, he has seen an increasing level of sophistication for predicting weather in the one- to two-week window, though the numerical weather prediction models begin to break down at the longer end.
"Models put the earth in a big computer programme and it simulates the movement of every air molecule on the globe. That works pretty well to about a week, but the errors grow over time," he explains.
The high degree of accuracy in the shorter end of the time scale is a major factor in the successful use of weather data streams to calibrate predictions in energy prices. Most energy trading shops have deep-level access to many different weather data sources and use the information as often as six or seven times a day to revise their expectations, he says.
Daily temperature fluctuations drive energy demand, so if it is a little bit warmer today than yesterday, the air conditioners are going to be running longer for example.
Though weather prediction has a long history with using probabilistic methods, the application of those outcomes in financial markets is relatively new. Bennett expects that it is going to be one of the biggest changes for how people work with weather data over the next five years.
"(A weather forecast) needs to be treated as a distribution, not as a single point, and that is a major delineating factor right now in its successful use," he says. In other words, expect the weather forecast is going to be wrong, opt for a distribution of possible outcomes and understand what portion of those possibilities link to specific events in the commodity being traded.
"That is the way of the future as more shops understand the nature of distribution and apply that to their decision-making as opposed to a deterministic forecast," he says.
He adds, however, that any automated strategy would need to take into account not just forecasting, but how much to weight any weather strategy day-to-day. "There are long periods of time that go by where factors such as geopolitical events, fundamental events such as power plant outages, are far more important. I don't think you could build an algorithmic, automated strategy based only on weather, let it run for a year and be profitable. You have to know when to turn it on and off."
Energy v ags
Trading of weather effects in energy markets contrasts significantly with the agricultural markets, where exposure risks often have more to do with weather at certain times. Corn crops, for example, hit a critical two-to-three week window in July and if it is a particularly hot and humid summer, plant growth is stunted.
"The agricultural markets are very sensitive to weather but only at specific periods of time. Sophisticated agricultural players certainly access weather data and information to assess their risk when it exists. It is just they don't need seven-times-a-day access the way energy markets do," Bennett says.
Some of the best analysts and commentators looking at agricultural markets are only right about half the time, says Shawn Hackett, chief executive of the eponymous firm Hackett Financial Advisors.
Despite the increasing precision of short-term forecasts, weather predictions don't easily translate to trading in the agricultural markets, he says.
"We still cannot predict weather with any degree of accuracy that would allow for one to take an actionable position without taking a tremendous amount of risk. There is always some variable that comes into play that nobody was expecting," he says.
Profitable speculation would need to get booked well in advance of any event that affects prices, and by definition, if a precise weather model was developed there would be little time to act as everyone would jump to buy at the same time. "It would only provide a benefit if it was a proprietary model," he says.
Traders are not just interested in the direct impacts from weather events. They also are keenly interested in predicting the ripple effects across related products.
A case in point is the severe and extensive US drought that nobody saw coming. It destroyed or damaged portions of major field crops in the US Midwest, particularly corn and soybeans, and the situation led to increases in farm prices that in turn affected animal feed and the livestock sector.
EarthRisk heat cold risk: An example of probabilistic temperature analysis by EarthRisk
In the spring of 2010, livestock producers could "sit on their hands and make money", Hackett says. But subsequent bad weather followed by the drought meant feed prices went from $3 to $8 per bushel. Cattle prices went from 80 cents per pound to $1.30 per pound in 2012.
Hackett is quick to note though that the domino effect of price spikes higher along that chain is not as straightforward as it may seem. The initial response to high grain prices is a significant increase in supply of livestock - hogs, chickens and cows - coming onto the market, which drives the price of meat down, not up.
"Cattle ranchers can't make any money, so they sell of their herd as fast as they can because feeding any more high priced grain is uneconomical. Then, after that liquidation is over and they've cleared out the decks, the price goes higher because rebuilding the herd can take up to two years in the case of cattle," he says.
Should the US see a good crop this year and feed prices drop, by some 50%, then lower cattle prices would result, but not until 2014/15.
About two-thirds of the firm's clients are commercial operators that are either users or producers of a commodity. In general, risk mitigation is the key concern for this group and Hackett advises on how to use futures and options to make more timely and profitable decisions. The rest of his clients are speculators.
For his speculative clientele, Hackett advises to look at relative prices among products that are correlated - the big three soft commodities (cocoa, coffee, sugar) and the big three grains (corn, soy beans, wheat).
"These markets are reverse-correlated, so softs go up, grains don't. You want to be short the grains, long the softs in a spread trade during times when the softs are historically cheap. When the softs are outperforming, you want to be long grains and short softs. Those are the kind of cross correlations across relative prices and spread trades that I show my speculative customers," he says. "It is one of the least risky ways to consistently perform well."
At time of writing, Hackett points out that the undervaluation of softs versus grains is at historical extremes. "Given the non-correlated nature of these markets, being long the softs complex and short the grains looks like a propitious investment opportunity."
In terms of using weather as a triggering event, however, location is crucial, Hackett says.
Stephen Bennett, EarthRisk Technologies
"I don't think you could build an algorithmic, automated strategy based only on weather, let it run for a year and be profitable. You have to know when to turn it on and off."
"Every market has its key country, or countries, that really control a price and in terms of modelling weather situations, you need to know which ones you need to be paying attention to. For Arabica coffee, the country to watch is Brazil, whereas for Robusta, it's Vietnam. For cocoa, Ghana and Côte d'Ivoire are important."
Still, numerous other factors can affect commodity prices just as much as weather, or sometimes more. For instance, new production techniques, such as those that are allowing the use of shale for oil production, can quickly change the entire nature of a market. "We are looking at a whole new era of global supply that was inconceivable four or five years ago when we were supposed to hit peak oil," Hackett says. "Agricultural producers are always looking to get the most (yield); corn farmers today produce twice as much corn per acre than they did 25 years ago. So even at stagnant prices income per year has doubled."
At the same time, the use of corn in ethanol is purely political and pricing has much more to do with government policy. "In about five or 10 years, there will be no corn used for ethanol. Ethanol will be produced but it won't be from corn." He says, adding that Brazil's use of sugar in ethanol is much the same situation. In other words, if traders want to create models, they need to fact in policy issues and the high degree to which these can lead to market regime changes.
Another complication is the currency market, since major commodities are priced in dollars. The trend of the dollar can often be the main reason for a bull or bear market.
Iain J. Clark, a financial mathematician and author of a book on commodity option pricing, also sees issues with building purely weather-based models for commodities trading. "It makes sense that weather and weather predictions should drive energy markets for gasoline and futures in gas and electricity, but the liquidity just isn't there to do this on a systematic basis," he says.
Clark has been a quantitative analyst since 1998, working across the industry at J.P. Morgan, BNP Paribas, Lehman Brothers, Dresdner Kleinwort, Standard Bank and UniCredit.
"From a network perspective, you could come up with models that attempt to capture both the driving factors and correlations between them, which indeed could vary and in some cases break down. Since 2011, for example WTI (West Texas Intermediate) and Brent are far less correlated," he says, referring to the two dominant crude oil products.
Softs grains correlation
A chart from Hackett Financial Advisors showing
the ratio of a softs index (coffee, cocoa and sugar) at a
historically low 0.80 the value of a grains index (corn,
soybeans and wheat)
Beyond commodities and energy markets, weather or seismic events can have a strong impact on various equity sectors.
According to the United Nations Office for Disaster Risk Reduction, the 2011 floods in Thailand reduced the world's industrial production by 2.5%. Japanese car manufacturers were hit hard in the aftermath, while computer manufacturers found themselves facing a shortage of hard disks.
The risk of global contagion from natural hazards is a key theme for global risk analytics company Maplecroft. Its head of natural hazards, Emily White, says those companies with thousands of suppliers spread across the world are becoming interested in supply chain dynamics.
"The insurance industry and service providers are very good at capturing risk from physical hazards, very good at modelling direct physical damage to assets. Where I think there is perhaps more work to be done is in capturing the interdependencies along the supply chain," she says.
Understanding those issues requires identifying where the asset exposures are. White notes the recent Pacific Catastrophe Risk Assessment and Financing Initiative, which resulted in a massive data set identifying exposed assets and populations across a number of Pacific islands. The data is being used for disaster response and has been used to structure a bundle of catastrophe swaps. White thinks the use of technology to understand where assets are exposed will play a vital role in the future.
There is a role for using weather data in predicting a company's performance, though to what extent remains in question.
Kyle Beatty, Atmospheric & Environmental Research
"In a case where an extreme event like a hurricane is threatening the US coastline, there could be active trading taking place around how significant of an impact that hurricane might have."
EarthRisk's Bennett says: "You are not going to find a lot of the major equities trading shops running weather strategies, but I think the few that do tend to use them quite profitably."
Weather also directly impacts closely watched economic indicators such as housing and consumer spending, says Robert Bronson, principal at Bronson Capital Markets Research.
Speaking via email, Bronson says his firm constantly adjusts economic data for direct weather effects and seasonality. This means adjusting data for year-over-year time series comparisons if there is doubt about the government's own seasonal adjustment process.
"They do not adjust for one-time events such as extreme weather: volcanos, hurricanes, tornados, floods, fires, etc and they don't adequately adjust for holiday impacts that are calendar-varying such as Easter," says Bronson, whose clients include hedge funds.
He also says he has identified an indirect relationship related to seasonal weather and stock market cycles.
Maplecroft Exposure index
An index produced by Maplecroft that highlights
economic risk from natural hazards
"When weather-adjusted data makes cyclical turning points, they contribute to a set of timing indicators that collectively yield buy and sell signals," Bronson says. "But rarely do these adjustments create a tradable signal by themselves. What creates tradable signals for leading and coincident indicators is weighted sets of such adjusted data."
Emily White, Maplecroft
"Where I think there is perhaps more work to be done is in capturing the interdependencies along the supply chain."
Meanwhile, data and analytics associated with environmental extremes are developing at a fast rate, which can pose a challenge for markets seeking to leverage these new tools, says Kyle Beatty, senior vice president at Atmospheric and Environmental Research.
Catastrophe bonds are a case in point. The instrument - a type of Insurance Linked Security - essentially securitises risk associated with high severity, low probability events that are not correlated with traditional financial markets. The mechanics of a catastrophe bond are not so different from the use of weather data in energy trading and the instrument can be traded during live weather events.
"In a case where an extreme event like a hurricane is threatening the US coastline, there could be active trading taking place around how significant of an impact that hurricane might have," he says. Currently, there is a nascent market supporting such "Live-Cat" trading as much of the risk from events such as hurricanes is transferred through pre-negotiated, annual or multi-year contracts.
This is another of the many ways in which science and markets are pushing the boundaries of weather data usage. And though it's an impressive evolution since Thales bet on olives, there is yet more work to be done.
Applying methods across a greater variety of markets and longer time horizons means models will need to incorporate many moving parts - local dynamics, currency, among various supply and demand fundamentals - while also weighting weather accurately as either a triggering or supporting factor.
Shawn Hackett says: "I have no doubt we will crack that code one day but at this point, weather continues to be a mystery and continues to confound the best minds in the business…If I could find someone that could (crack the code), it would make my job a whole lot easier, I have an interest in getting it right."