The March 2004 issue of "Stocks, Futures & Options" magazine (http://www.sfomag.com) will include an article from Moore Research Center, Inc., discussing the seasonal tendencies of gasoline and natural gas during March-May.
Few industries are more basic than energy, and few changes are more dependable than those that follow from one season to the next. So, if change creates opportunity, has the vernal equinox generated opportunities in the energy sector? If so, of what consequence and with what degree of reliability?
Numerous factors affect markets, but some conditions and events recur annually. One such, of course, is the annual cycle in weather from hot to cold and back to hot. Thus, nature sustains --- despite weather, economics, even wars --- an annual cycle in grain supply. In some years it may be more and in some less, but there is always a seasonal bulge in supply. Similarly, cold weather creates demand for heating fuel. In some years it may be colder and in some warmer, but there is always a seasonal bulge in demand.
And so it is that annual cycles in nature generate and sustain annual patterns of supply and demand. The thrust of this article is to discuss the seasonal approach to markets: what it is, how and why it works, what it can and cannot do, how traders might benefit from it, and how it applies during spring to two primary energy markets, gasoline and natural gas.
Natural Market Rhythms
Traders and merchants have for centuries used a seasonal approach to advantage in markets directly affected by such cycles. Those who prospered most anticipated best. Already realizing that price responds to supply and demand, they reasoned a step further: If cycles in nature create patterns in supply and demand, then perhaps those patterns generate price patterns. If so, the market's own natural rhythm can help one anticipate recurring price movements in the future.
Thus, the seasonal approach to markets is designed to help traders anticipate the timing and direction of future price movement. If one can find out how price tends to move at a given time of year, then one can prepare well in advance to take advantage of it doing so again in the future. Armed with knowledge of history, a trader can also better understand current market activity --- and act upon it rather just react to it.
Consider gasoline. Does the annual cycle of hot weather to cold and back to hot again affect supply and demand for it? When is consumption least and when is it greatest? Road conditions are worst in the depths of winter, when ice and snow make driving most hazardous. With school in session, families stay close to home. Thus, gasoline consumption tends to be low during winter.
But with the vernal equinox comes improving weather, which improves road conditions. Improving road conditions increase driving, which increases consumption of gasoline. Weather and road conditions continue to improve into the Memorial Day weekend at the end of May, the traditional opening of US driving season. With school out for the summer shortly thereafter, families begin to vacation and driving increases. The seasonal peak for gasoline consumption comes during July and August.
Thus, an annual cycle in weather creates an annual cycle in consumption of gasoline. How does that translate into a pattern of supply and demand? Petroleum is refined into two primary products, heating oil and gasoline, patterns of consumption for which are opposite. Refiners in several of the largest producing states, such as Texas and Oklahoma, are subject to tax on year-end inventories. Thus, they have financial incentive to pump as much product into the proverbial pipeline as possible by then.
That flush in supply is not of great concern during the season of least consumption. But during February, as the heating season ends and before the driving season begins, many refiners take some downtime to retool, to reformulate, and to perform maintenance. Those shutdowns, temporary thought they may be, interrupt the stream of supply. Further, the most populous regions of the country require gasoline to meet higher emission standards beginning no later than May. Thus, starting with inventories low, daily consumption begins to rise into summer and the industry needs inventory prior to peak driving season.
Demand accelerates because daily consumption rises at the same time as the industry accumulates inventory, forcing prices higher to encourage production until supply reaches equilibrium with demand. The more prices rise, the closer to capacity refiners will produce. If all goes well, inventories will be sufficient and producers refining near capacity in time for the opening of driving season in May.
Thus, the annual cycle in weather creates a pattern in supply and demand for gasoline which then gives rise to seasonal price phenomena --- tendencies for price to move in the same direction, with greater or lesser intensity and in a more or less timely manner, with a certain degree of reliability each year. In other words, an annual pattern of changing conditions --- cause --- creates an annual pattern of price response --- effect. In a market influenced by annual cycles, seasonal price movement can become almost self-reinforcing as if the market had a memory of its own. Why? Consumers and producers can fall into their own patterns of behavior. Once they depend on seasonality, vested interests will maintain it.
The seasonal approach originates from the premise that each market has fundamental forces peculiar unto itself that act upon it every year. If one can find empirical evidence of a pattern in market reaction to those forces, then one can more broadly define seasonality as a consistent market tendency to repeat similar price movement annually. So defined, the principle is subject to being observed and quantified in any market.
The word "pattern" implies a degree of reliability. Because a primary function of futures markets is to anticipate, prices tend to move when anticipating change, such as improving weather, and to adjust once that change is realized. When that change is annual, a recurring cycle of anticipation/realization evolves. That dynamic is intrinsic to the seasonal approach to trading --- designed to anticipate, enter, and capture recurrent trends as they emerge and exit before they are realized.
The focus of seasonal research is to find recurrent trends within a seasonal price pattern. Computers can now construct a daily seasonal pattern of price behavior derived from a composite of actual daily price activity through several years. Properly constructed, that pattern reflects where, on any given day of the year, a market has tended to be trading within that market's annual price cycle.
Referring to the seasonal pattern for July Unleaded Gasoline, for example, the numerical index on the right-hand vertical scale reflects the observed historical tendency for its price most consistently to be high when at 100 --- the seasonal high --- and for price most consistently to be low when at 0 --- the seasonal low. Thus, the pattern portrays that specific contract's tendencies to reach its annual peak and its annual nadir and also to trend in between. From this visual reference, a trader can better judge current price activity and anticipate future price movement.
Now consider the seasonal pattern that has evolved over the 14 years of trading in natural gas for delivery in July. The market for natural gas rests on twin pillars of seasonal demand. Northern regions depend on it for heat during the cold winter whereas warmer southern regions depend on it to generate electricity to run air-conditioners during the hot summer. The two regions are relatively separate and as are the distributors who serve them. Thus, during the depth of winter when consumption is highest in northern regions, it is least in southern regions. But then southern distributors must begin to anticipate summer retail demand. By March, they tend aggressively to accumulate inventory.
How aggressively and how regularly? The seasonal trend from late February into mid-/late April appears irresistible. Does a trader just jump in? Or can seasonal analysis go yet another step?
Comparing actual daily closing prices for this specific contract during those 14 years, computers can simulate all possible combinations of daily entry and exit. With standards set for such considerations as statistical reliability, duration of trade, and historical profitability, computers can discover entry and exit days which have tended to generate optimal price movements.
|1||Buy Unleaded Reg.(NYM)-July||3/01||5/09||93||14||1||15||2059|
|2||Buy Unleaded Reg.(NYM)-June||3/20||5/14||93||14||1||15||1885|
|3||Buy Unleaded Reg.(NYM)-June||4/08||5/09||93||14||1||15||1456|
|4||Buy Unleaded Reg.(NYM)-June||4/12||4/29||93||14||1||15||1095|
|5||Buy Natural Gas(NYM)-June||2/25||4/15||100||13||0||13||2862|
|6||Buy Natural Gas(NYM)-July||2/25||4/21||92||12||1||13||3108|
|7||Buy Natural Gas(NYM)-May||3/14||4/21||92||12||1||13||2638|
For example, refer to the table of seasonal energy strategies. Statistics in the sixth row simply state the following: "July Natural Gas has closed higher on about April 21 than on about February 25 in 12 of the last 13 years, generating an average move equivalent to $3,108 per contract per year." (MRCI research shortens the trade window when an optimized entry or exit date falls on a weekend or holiday; hence, the word "about." Because natural gas futures began trading on April 3, 1990, there are only 13 years of data for these strategies but enough to construct a 14-year seasonal pattern for July futures.)
This historically factual strategy not only rigorously confirms a segment of the seasonal trend illustrated within the overall seasonal pattern but also highlights just how vigorous and reliable has been that recurring price movement. Other strategies in the table, such as that for July Gasoline, were derived in like manner and reveal similar details of seasonal trends within their respective seasonal patterns.
Even such dynamic trading patterns do not repeat without fail, however. Seasonal research is statistical analysis, factual but performed with the benefit of hindsight. Statistics confirm the past but cannot predict the future.
In other words, seasonal research does not generate a "black-box" trading system to be followed with eyes wide shut. As with any approach, it has inherent limitations. Perhaps the most crucial reason for not trading seasonal research in a vacuum is that it looks backward only. The trader himself must exercise some judgment.
For example, one must avoid being caught in a true contraseasonal move --- one of the more dynamic in markets. Seasonal price movements are a response to normal factors. Thus, unusual and powerful factors are required to override the "norm" and force prices to move in a contrary direction. Because consumers and producers alike often depend on that with which they are familiar, they tend not to recognize and respond to warning signs and eventually to panic when it is too late. As such, a contraseasonal move can also be self-reinforcing until its often dramatic climax.
The gasoline market in 2003 provides a perfect example. Instead of rising moderately into and then "resting" during February, the July Gasoline soared to $1.05/gallon in early March on pre-Iraq war anxiety over possible shortages. Thus, the market was already discounting worst-case scenarios. When none happened, the market collapsed to $0.80 by late March --- and eventually to $0.76 in early May when the seasonal peak normally occurs!
Going forward, however, a seasonal trader not only must try to avoid potential contraseasonal moves but also may wish to attend to more mundane but still practical issues such as shorter-term timing. Fundamentals not only long-term but also short-term inevitably ebb and flow. In some years, for example, summer weather arrives sooner and is hotter and dryer than in others. Traders may visualize each optimized date as resting at the peak of a bell curve of distribution --- that date may not have been the best in any specific year, but that one date more than any other would have generated the best results for all years.
Other philosophical issues concern both the size and validity of the statistical sample. In statistics, bigger is usually better. But in some cases, more recent history may be more useful. For example, Brazil became a global soybean producer in about 1980 --- after which the seasonal pattern for soybeans reversed nearly 180 degrees. Conversely, disinflationary patterns prevalent 1980-2000 may serve traders poorly in an inflationary environment. One may also plausibly ask what discovery is relevant when a computer sifts raw data. Might there be a cause-and-effect relationship --- or is it a statistical aberration? How much can one rely on the isolated fact that a pattern has repeated in even 14 out of 15 years? Are the odds going forward better than 50/50?
Thus, one cannot for granted seasonal trends even of such exceptional consistency. Traders can best employ seasonal research by integrating it with common sense, perhaps a simple technical indicator or timing trigger, and/or some familiarity with current market fundamentals. Doing so can help one first decide if the anticipated seasonal price movement is likely to recur this year and then, because dates are not etched in stone, to refine entry/exit timing.
If fundamentals drive markets, then recurring fundamentals can drive recurring market responses --- a seasonal cause-and-effect relationship. It is impractical if not impossible, however, for a trader to know all relevant fundamentals for even a single market. He can turn to a properly constructed seasonal pattern to identify trends that have recurred in the same direction during the same period of time with a high degree of past reliability --- because that very reliability implies recurring fundamental conditions or events that presumably will exist again in the future and affect the market to one degree or another and in a more or less timely manner. Rather than react, he can anticipate and act boldly.
Seasonal research, as performed by Moore Research Center, Inc., is meant to help traders better make their own trading decisions. A seasonal pattern illustrates the well-worn path a market has itself tended to follow, and specific seasonal strategies can identify when, in which direction, and how frequently that market has moved. In doing so, the seasonal approach can help integrate both technical and fundamental factors. But a market's own consistency provides the foundation for why the seasonal approach to trading works --- in any market.