|
Fontanelle, IA | I vaguely remember that in Stats - high P values and corresponding R squares show correlation.
Pretty sure an R squared of 7's would not show correlation so you either need to 1. find a better fit line or 2. increase your data or 3. reject your hypothesis? In this case, stocks to use ratios are often correlated to prices - commodity analytic firms use them all the time - so you need to find a better line. I don't remember how to get your confidence interval from your R-square.....
KST1 gave you a suggestion as to better fit regression line.
Maybe your graph is labeled incorrectly but you use current stocks to use vs. current price in order to "predict" what future stocks to use % correlate to a price. Then, this would be the "average/mid point" of predicted price values. then you would give a range of prices based upon some calculation of standard deviation that correlates back to your data set.
If not, price would only change 1 X per month - the USDA report date that reports "updated" stocks to use.
Lastly, I would think that under excessively/crushingly high stocks to use percents, the regression curve would level off somewhere around the variable cost of production around soybeans. Just enough to cover the expenses but not enough to cover fixed costs so that we all "attempt" to plant them. Just a GUESS - been a long time since I've been in the Holiday Inn
| |
|
Bloomberg's reporting.... Why are soybean futures even $8.30?
