Brugler Marketing & Management LLC

Special Research Report

November Soybean Price & Probability Study

July 1-November 1 Time Period

Released July 10, 2003

November soybeans have a wide average annual trading range. Their 15 year composite pattern shows a drop of 60% of their average annual trading range between the July crop report and their typical late October bottom. Even in the relatively tame period from 1997-2002, the 4 month average trading range is $1.13 per bushel. The contract is volatile because it encompasses the key yield-determining period for soybeans in the US, which is late July to late August. So what can we expect to see in terms of price ranges for November beans?

Here is the track record for the contract since 1970, showing the highest and lowest prices seen during the 4 month period, along with the closing price on the first trading day of July and the closing price of the first trading day of November.

November Soybean Futures (July 1-November 1 Period)

As you can see, the full data set includes 33 years. The average price performance for the entire period is shown as '70-02 Average'. We've also looked at the data from 1980-2002, on the belief that soybean production was much smaller in the 1970's, and there was no South American competition in the export market. The 70's might not therefore be relevant. Average prices have been higher since 1980, but the distribution is similar. We also looked at data only since Freedom to Farm was enacted and Brazilian production expanded more sharply. That period runs from 1997-2002. It was popular a year and a half ago in some circles to insist that "the world is different now" and that we would never see $6 beans again.  While 2003 disproved the $6 argument, we're clearly in a two season world marketing year, the US season and South American season.

We have two harvest lows rather than one, and there is some impact on price patterns. That's pretty clear if you compare the post-1996 averages with the post-1979 averages! Our July 1 starting point is only 2 months after the Brazilian harvest ends, and the world market doesn't have to place big weather bets on the US crop if it is still chewing through large competing supplies. That average July 1 price dropped from $5.93 in the 1980's and 90's to only $5.19 in the last six years.

Let's accept the premise that certain calamities befall a percentage of the US crop every year, dragging bean yields down from optimal to USDA national average, and that futures prices oscillate around the projected cash average price as weather, export and other info is acquired. What could we reasonably expect for a price range over the next 4 months, knowing that we began the period at $5.59 and had a pretty good handle on acreage and yields as of that date?

First of all, note that new crop futures are still trading 40 cents per bushel above where they were a year ago at this time. The market IS aware of the tight old crop ending stocks situation, and apparently insatiable Chinese demand. We are carrying a price premium, and the US average cash forward contract price (source DTN NSI) had been above loan rate until the post July 4 sell off.

On average, the summer or fall high in soybeans has been 70 cents ABOVE where ever we were standing on July 1. The average low is 43 cents below July 1 if you use the nearby data set, and 67 cents under when you use the 23 year version. The higher average low is likely a function of the lower average starting price in the nearby data.  We haven't dropped as much in recent years because we were already down!

With average price movement, we'll see a trading range of $1.37, if we use the longer term set, or $1.13 using the shorter set. The only difference is the average low. Based on our July 1 price, that low is likely to be between $4.92 and $5.16.  Thus far, 9 days into the month, we've seen $5.345.

On a calendar basis, prices typically are 45 cents to 75 cents away from their July 1 starting price by November 1. That's using the absolute value column to try to measure the average swing. The last column in the main table shows the mean, with losing years canceling out gaining years. It shows an interesting shift. In the full data set, we typically lost 10-15 cents from July to November. Since 1997, we've averaged a 15 cent GAIN. However, note that only two years caused the positive result, 1997 and 2002.

We also ran a regression analysis on the data, to see if there is a correlation between July 1 prices and the subsequent highs and lows. There is. It is fairly strong for the lows, with an r-squared of .82. It's not bad for the highs, either, at .71. These allow us to set some probability bands.

The best fit high and low correspond well with those you'd see under the average move model discussed above. The most likely high for the period is $6.29, and the most likely low is $5.11. However, to allow for 2/3 of all possible trading years, the high could fall anywhere between $5.53 and $7.05. I know, you're thinking that with average water that deep you can drown! However, it makes a relevant point. We have traded as high as $5.615 since July 1. If this is one of the ugly price years where yield estimates start big and get bigger, we've already traded into the price band where the high could be. Present oversold technical conditions not withstanding, it is statistically possible that rallies won't get back above that $5.61 area. It doesn't take much imagination to see how we could get to $7.05, either. All we need is hot and dry weather in late July that continues throughout August. Throw in an early frost if you really want to juice it, although you usually can't mix the two in the same year. With tight old crop stocks, any threat that would take 150-200 million bushels out of the production estimate would get us to $6.29.

What's the downside risk? The best fit low is $5.11, but it could fall anywhere between the $5.345 already seen and $4.74.  Of course, there are also "outlier years" beyond the 2/3-probability curve, and that's where the analysts calling for $4.40 or $4.60 harvest lows are living. The best argument for getting that low is a combination of a 3 billion bushel US crop, spec funds initiating a net short position instead of their current net longs, and maybe a market trying to signal South America not to plant so many beans in October.

Seasonal studies by MRCI show a strong correlation between 2003 price action and the 1987 and 1973 patterns. We were following 1988 as well, but lost that correlation recently. In 1988, there was a drought, and it was over by early July. That year we peaked November beans on June 23. Thus far in 2003, we have to call the peak June 13. In 1987, the high was on June 16, and prices moved steadily lower until putting in an early harvest low on August 12. The 1973 pattern was different, again topping in mid-June, and sliding into July 9. However, that year we rallied more than $2.00 into the August crop report.

Conclusions

The price strength in May and June was contra-seasonal.  It didn't fit either the 5 or 15 year average patterns.  The sell off in early July did fit the seasonal, however. A rally from some sort of low in July into the August crop report appears in most of the charts we looked at. The 4 month low for November beans is most often in mid to late October, and that is true for both the correlated years and the entire seasonal group.  We have to assume that any low seen here will be exceeded in the middle of harvest, unless we have a major fundamental shift. So, from a seasonal/cyclical standpoint we're looking for a rally to sell over the next 3-4 weeks. We don't have to be aggressive, since we are already short both futures and via put options. However, we might want to cash those out before the end of July, and look to book more cash sales if such a rally unfolds.

In price terms, we have strong odds of a low between $4.92 and $5.11 over the next few months, with some risk of $4.74 or lower. Selling Nov 460 or lower puts for income is probably feasible at a July swing low or a 1987 style early harvest low. Upside potential is pretty fancy, with both the average moves and regression suggesting $6.29 is still possible. Selling Nov 640 and 660 calls on rallies should be safe, and those that we've already sold have an excellent chance of expiring worthless. With oversold technicals and prices within 10 cents of good support, this is probably not the time to sell those calls, or add to hedges, or sell more cash beans. 

THERE IS A RISK OF LOSS IN FUTURES AND OPTIONS TRADING. FUTURES TRADING IS NOT APPROPRIATE FOR ALL INDIVIDUALS. ANYONE ACTING ON THIS INFORMATION IS DOING SO AT HIS/HER OWN RISK. CONSULT THE  FUTURES AND OPTIONS RISK DISCLOSURE STATEMENTS YOU SIGNED WITH YOUR BROKER BEFORE TRADING. PAST RESULTS ARE  NOT NECESSARILY INDICATIVE OF FUTURE RESULTS. 

© 2011 Brugler Marketing & Management LLC                                          (402)-697-3623