By Todd Hultman
DTN Analyst
I don't want to sound like the old guy yelling at kids to get off my lawn, but I have to admit I get grumpy when I hear people spit out market advice based on some thin bit of folklore. I'm not normally bothered by astrology columns, psychic hotlines, or the Farmer's Almanac, but when it comes to lazy market advice, I have a short fuse.
And the problem isn't just with the advice givers. When it comes to understanding markets, we all have blind spots that need to be challenged. Last spring, the common wisdom was that soybeans couldn't go higher because there were too many of them. Hmmm.
What's the best antidote to lazy thinking? Research, research, research.
As much as possible, I try to quantify and test everything. When done correctly, the numbers don't lie and, often, I have to adjust my thinking because of what the numbers prove. It's a good way to separate the wheat of what works from the plentiful chaff of cheap advice.
A recent example of this happened when I noticed how corn turned higher in September and October, up from its multi-year lows. Fundamentally, this did not make a lot of sense while USDA was anticipating a record harvest of 15 billion bushels, but technically, it was impressive that the monthly stochastic for corn was turning higher from an oversold position.
It then hit me that as many times as I have seen the monthly stochastic identify long-term changes in trends, I had never tested this popular indicator. Thanks to DTN's ProphetX software, I was able to fill a spreadsheet with the monthly closes of spot corn going back to 1994 and their corresponding stochastic values.*
From there, I plotted a scatter chart of monthly stochastic values on one axis and percentage changes in the price of spot corn one year later on the other. As fans of the indicator would expect, the data showed a clear tendency for lower stochastic values to result in higher corn prices one year later. Likewise, high stochastic values tended to result in lower corn prices the following year.
Since 1994, monthly stochastic values below 20 resulted in higher corn prices one year later 62% of the time. Actual results varied widely, but those within one standard deviation of the average ranged between a 23% loss and a 48% gain. Because the percentage gains were larger and more numerous than percentage losses, the line of best fit put the expected gain between 14% and 18%.
On the other end of the scale, stochastic values above 80 resulted in lower corn prices one year later 72% of the time with expected losses of 0% to 4%. Again, actual results were fairly wide and those within one standard deviation of the average ranged between a 40% loss and a 5% gain.
For some, it may be frustrating that low stochastic values don't necessarily guarantee higher corn prices or vice versa, but the data clearly shows a bullish bias for price outcomes when the stochastic K line is below 20 and a bearish bias when the stochastic is above 80. This is no small finding because fundamentally based market opinions are often bearish when the stochastic is low and bullish when it is high -- just the opposite of how prices actually move.
Technicians will point out that George Lane's instructions for his stochastics indicator were more complicated than my simple test covered and that is true. But even they may be surprised to find that historical data confirms the use of the monthly stochastic K line as an effective gauge of overbought and oversold conditions. Combined with other market clues, the stochastic indicator is a valuable piece for understanding the market's ongoing puzzle.
The practical lesson as we head into November is that we have another valid reason to ignore corn's bearish market sentiment as prices are rising from a low stochastic value of 13 in August.
While others focus almost exclusively on the size of harvest and try to pretend they can estimate next year's corn demand, I suggest we pay attention to the market's more substantial clues. The monthly stochastics indicator for one, is currently bullish for corn and deserves more respect.
* This study only paid attention to the stochastic's K line. %K was set to 3 and %K Slowing (or the time period) was set to 14. The staff at Investopedia.com offers a broader description of the stochastics indicator at: http://bit.ly/…
or
http://www.investopedia.com/…
Todd Hultman can be reached at todd.hultman@dtn.com
Follow him on Twitter @ToddHultman1
(CZ/)
© Copyright 2016 DTN/The Progressive Farmer. All rights reserved.