Todd's Take

Buy low, sell high has a logic that is easy to accept, but the catch has always been coming up with a definition of what's low and what's high that can also stand the test of time. A little like defining jazz, we might have a sense of what we think low and high are now. But can we come up with a method that holds up for more than just a few years?

The value charts that I write about at the start of each year are good efforts in this regard. That work should continue to stand the test of time because production cost is such a key component in understanding market behavior. Also, DTN's five-year-range market factor is another valuable concept for understanding the highs and lows of grain prices, which I wrote about last year.

Aside from those two, I wondered if a statistical tool like the Bollinger Bands indicator could help us define when prices are high or low, so I started experimenting with the help of DTN's ProphetX software.

Old-timers may remember John Bollinger as an analyst on the Financial News Network back in the 1980s. The Bollinger Bands indicator he developed was an interesting attempt to bring the tools of statistical analysis to price charts, and he applied it to both stocks and commodities.

The indicator is made up of three lines. The default setting shows a simple 20-day moving average in the middle, surrounded by two other lines set 2.0 standard deviations above and below the moving average.

The basic idea is that the boundaries of two standard deviations provide an expectation of how volatile prices are expected to trade around the moving average while also providing a flexible framework that would change with conditions over time. There's more Mr. Bollinger could teach us, but for our purposes, I like the idea of seeing if standard deviations based on past history could help us anticipate the market's highs and lows.

Of course, a 20-day average was much too short for what I had in mind, so I brought up a monthly chart of spot corn prices and set the Bollinger Band to different lengths, looking at time periods from 12 to 120 months.

It did not take long to see a major flaw in the method. In unusually strong bull markets, like what happened in corn from 2006 to 2008, the lower Bollinger band would actually drop lower, in this case, from $1.68 to 56 cents a bushel. The flaw was that corn's higher prices increased the expected volatility of the entire model, both up and down. In the real world, however, there is no reason to believe that dramatically rising corn prices somehow also make dramatically lower corn prices more likely.

While Bollinger's statistical method did not hold up well for long-term markets, I still liked the notion of prices trading in relation to a long-term moving average. Prices are complex. Yes, they reflect the supply and demand of the commodity they represent, but they also reflect a lot of other things, including the emotions of market participants with short attention spans.

It may be one of the unexplainable paradoxes of life, but while individual prices can be flighty and capricious, there is a certain wisdom in large numbers. Gather all those prices together over a significant length of time and you get a sense of what's reasonable when short-term distractions are stripped away. As it turned out, the real star of Bollinger's bands was not the effort to measure price volatility, but the unflappable moving average in the middle.

I went back to ProphetX and looked at the idea of putting price envelopes around a five-year average. It soon became apparent that a 30% price band above and below the five-year average defined spot corn prices very well from 1975 to 2005.

The big rally in corn that began in late-2006 took prices far above their upper 30% boundary and is a good example of how it is more difficult to define the high end of prices than it is the low end, as perceived shortages are famous for stoking price volatility. Trading near the lower end of the range, however, has a far less volatile track record, as prices tend to become inelastic in the face of large surpluses.

Not only did a 30% price envelope do a good job of identifying potential lows in spot corn, they also worked well for all three wheat contracts and even lean hogs. The lower boundary of soybeans seemed to do better with a 25% envelope and cattle with 20%.

The lower band of minus 30% is currently at $3.35 for spot corn and $4.14 for spot Chicago wheat. As usual, there is no guarantee that prices cannot trade lower -- after all, they are capricious.

Buy low, sell high is bad advice for short-term traders and anyone averse to risk. But for those willing to look at markets from a long-term perspective, here is another example of corn and wheat prices trading at levels traditionally defined as low where support has appeared in the past.

Todd Hultman can be reached at todd.hultman@dtn.com

Follow Todd Hultman on Twitter @ToddHultman1

(AG/BAS)

Copyright 2016 DTN/The Progressive Farmer. All rights reserved.