WEATHER FORECASTING AND
ENERGY TRADING
NOTES FROM THE TRENCHES
April 15, 2010
Vincent Kaminski
Rice University
Jesse H. Jones Graduate
School of Business
1999 VK-9060359-1
Outline
Energy Impact on the Commodity Markets
Notes from the Trenches: My Interactions with Weather
Forecasters
The Future Trends
Weather Derivatives
Algorithmic Trading
In-house Weather Modeling?
1999 VK-9060359-2
WEATHER IMPACT ON
THE COMMODITY
MARKETS
1999 VK-9060359-3
The Channels of Transmission
The impact of weather on the agricultural markets
The dominant transmission channel: the supply side
The impact unfolds relatively slowly over time, except for some
extreme event
Extremely cold weather in Florida affecting orange juice market
Freeze in the coffee producing regions
The potential for systemic impact due to long-term weather trends
(Pinot Noir from Scotland?)
The impact on the demand side: local and short-term
The impact of weather on the agricultural markets was
recognized in the relatively early stages of development of the
futures exchanges
1999 VK-9060359-4
Words of Wisdom From the Past
So much for the financial conditions.
The spring had been backward, cold, bitter, inhospitable, and Jadwin began to suspect that the
wheat crop of his native country, that for so long had been generous, and of excellent quality,
was now to proveit seemed quite possible--scant and of poor condition. He began to watch
the weather, and to keep an eye upon the reports from the little county seats and "centres" in
the winter wheat States. These, in part, seemed to confirm his suspicions.
From Keokuk, in Iowa, came the news that winter wheat was suffering from want of moisture.
Benedict, Yates' Centre, and Douglass, in southeastern Kansas, sent in reports of dry, windy
weather that was killing the young grain in every direction, and the same conditions seemed to
prevail in the central counties. In Illinois, from Quincy and Waterloo in the west, and from
Ridgway in the south, reports came steadily to hand of freezing weather and bitter winds. All
through the lower portions of the State the snowfall during the winter had not been heavy
enough to protect the seeded grain. But the Ohio crop, it would appear, was promising enough,
as was also that of Missouri. In Indiana, however, Jadwin could guess that the hopes of even a
moderate yield were fated to be disappointed; persistent cold weather, winter
continuing almost up to the first of April, seemed to have definitely settled the question.
Frank Norris (1870-1902), The Pit, New York, 1903
1999 VK-9060359-5
The Channels of Transmission (2)
The channels of transmission in the case of energy markets:
primarily through the demand side
The impact on the supply side: primarily through specific events of
varying magnitude, both in terms of scope and duration
Hurricanes
Precipitation impact on hydro power
Special local events
The events require good understanding of local conditions
The impact varies from long-term to short-term and fairly local
Fog in the Houston Ship Channel
High waters of the coast of California
Low water level in the Rhine river
Wells freeze-offs
1999 VK-9060359-6
The Channels of Transmission (3)
The challenge: weather forecasting for an energy trading operation
Practically everybody has the access to the same source data and
analytical skills (for a price)
The competitive advantage can be acquired through:
Better interpretation of the conflicting forecasts (when the models diverge)
Improved skills in the process of selection of the weather forecasting firms
Moving up the supply chain to beat the competition
Faster acquisition, processing, visualization, and delivery of the
weather information
Elimination of some links in the supply chain
Direct acquisition of the model results and atomic weather data
Most important: better understanding of the impact of weather on the
energy markets, contingent on other conditions (trading positions,
conditions of the energy infrastructure)
1999 VK-9060359-7
THE EARLY DAYS
1999 VK-9060359-8
Weather Function in Energy Trading
Financial firms trading agricultural futures had weather
teams supporting traders
Many weather analysts made contributions to energy trading
in free time or when market conditions required their
involvement
The first energy weather group was started by Enron
The Enrons move was quickly replicated by the rest of the
industry
The weather group was considered to be critical to the
success of Enron trading
1999 VK-9060359-9
UNFOLDING TRENDS
1999 VK-9060359-10
WEATHER DERIVATIVES
1999 VK-9060359-11
Weather Derivatives: Definition
A special version of weather insurance
Structured as derivative transactions
Regulatory considerations were the most compelling reason
Designed to hedge the risk of low severity, high probability events
By definition, actively traded on multiple platforms
A very successful take-off
The merchant energy industry used conditional probability distributions
(unlike some counterparties in the insurance industry)
Some companies exploited successfully the human island phenomenon
and data errors
Very limited repeat business
1999 VK-9060359-12
Weather Derivatives: The Future
My opinion: its not a viable business, as opposed to highly
structured, customized weather insurance
The reasons:
Weather derivatives put on head the very principle of weather
derivatives (the few will pay for the losses of the many instead
of the many paying for the losses of the few)
The high probability events represent permanent and recurring
conditions of doing business and are most effectively
addressed by incorporating the cost in the price of a product
The cash flow smoothing aspect of weather derivatives: the
premiums are effectively capped by interest rates
1999 VK-9060359-13
In-house Modeling
Some energy trading organizations move to direct acquisition of
certain raw or intermediate data
Model outputs
Certain critical data sets
For example, sea surface temperatures
It can not be excluded that at some point some companies
will engage in direct modeling of weather and/or
computational activities
Will the market size justify this effort?
1999 VK-9060359-14
Algorithmic Trading
Several energy trading organizations are experimenting with trading
strategies based on data mining
Search for the patterns of markets response to the weather data
Automatic execution of transactions
The challenges
The difficulty of identifying and capturing all the relevant data
The reaction to the weather data depends on many factors, which may
not be measurable
Example: positions held by other traders
The volume may not be sufficient to bother
1999 VK-9060359-15
Contact information:
vkaminski@[Link]
1999 VK-9060359-16
