Playing the Odds in Gas and Power Markets

on October 03, 2013 at 2:30 PM

World Poker Tour - Doyle Brunson North American Poker Championship - Day One

When it comes to energy trading, there’s very little that’s 100% certain. But you don’t need to know everything to get a leg up in natural gas or power markets. You just need to know more than the other guy.

Software firm EarthRisk Technologies has unveiled a mid-range temperature forecasting product designed to give its clients that advantage. The product, TempRisk Apollo, seeks to enable traders in gas and power markets to quantify weather risk up to five weeks out. It does not rely on the kind of traditional, or “discrete” forecast you might see on a weather report, but presents large-scale weather events, such as cold fronts or heat waves, in terms of probabilities, letting customers know the chances that a any of several discrete forecasts will prove accurate.

“Energy trading is based on weather forecasts, not actual weather,” EarthRisk chief executive John Plavan told Breaking Energy. “On any given day, there may be a full spread of potential probability outcomes.”

Having probability data for a range of possibilities – rather than just trusting, or not trusting, a meteorologist’s single forecast – can help a trader to determine whether market expectations reflect the most likely outcome, or reveal mispriced risk.

“Other forecasts may have either very low confidence or very high confidence that a particular weather event is going to happen, and if you know it’s highly variable you can trade against that variability,” Plavan said. “It doesn’t matter what the weather actually is, it matters what the market forecast is.”

EarthRisk’s probabilistic assessments draw on historical data. “We look at it purely statistically,” said Plavan. For example, if specific conditions have historically preceded the same sort of weather event 30 days out 60% of the time, they can shed light on what may happen 30 days from now.

But this also leaves the model open to disruption from shifts in global weather patterns that may be associated with climate change. “We’ve put a lot of effort into de-trending the data and adjusting the models to account for that,” Plavan said. “Things aren’t changing dramatically overnight – they take many years and cycles. We look for large global patterns that lead to large-scale events covering large regions, and that scale of pattern is easy to account for.”

That focus on large-scale weather events also means that TempRisk Apollo is, at least for now, of limited utility for forecasting extreme weather events, such as hurricanes that force companies to shut down oil and gas platforms in the Gulf of Mexico. “We’ve done the preliminary science to do tropical cyclone genesis and severe storms. The methodologies that we use for temperature are directly applicable, but the atmospheric science is totally different,” he said.