Death comes to us all. Life's but a walking shadow | As you know I've been trying to understand how the different categories of traders respond to price by analyzing the Commitment of Traders data set. It seemed reasonable to me that the various categories of traders would have certain common objectives and would respond to changes in price in some generally consistent ways. For example, producers & merchants would generally respond in a similar manner to a price event. Likewise, those traders in the category of Managed Money would respond in a different but generally consistent way. As would the traders in the Other Reportable and Non-Reportable categories.
To test this idea I constructed a data set consisting of the weekly change in the price of C1 corn (from Quandl) to carefully correspond to the weekly intervals reported by the Commitment of Traders Report for the corn futures (only). I went through the whole delta price table and corrected for holidays, etc. I even spot checked the COT results with some calculations of my own and they seemed consistent.
So I have a table with the weekly change in price and the change in futures positions (long & short) for the various trader categories. You would think that these positions should change in response to changes in price. For example, if the price drops, long positions should decline and short positions might increase, at least among the speculators. In particular, you would think that the Non-Reportables aka the small speculators would be quite sensitive to prices changes. Likewise the Managed Money traders would take advantage of price moves to make profit.
So I plotted scatter plots of changes in positions vs the changes in price. In most cases there didn't seem to be much if any consistent responses. The correlation coefficents were vanishingly small.
Faced with this result I thought that the very largest price changes at least should show corresponding larger changes in position I filtered the data set to just large positive or negative price changes. Again, very little correlation. The only category that show anything like a consistent response was the producer/merchant short category. There at least the direction of the changes in position were consistent. A decline in price was accompanied by a decline in short positions as they lifted hedges, an increase in price accompanied by an increase in positions as they made sales and put on hedges. There didn't however seemed to be a linear or even curvilinear relationship just a qualitative relationship.
I would have thought that the small speculators (aka non-reportables) would have responded quickly to changes in price but that's not what I saw. A large price increase could be accompanied by a large drop in both long & short positions. Maybe the time intervals obscure the relationships but change in price is "change in price".
Well, supposedly "negative" results are just as valuable as positive results but I'm still puzzled.
|
LongKC, What do you make of this?
