I wanted to find out for myself. So I gathered all the historical price and COT data for the 5-years treasury market. I compared each week’s report to the next week’s rolling average (in order to best capture fluctuations and changes throughout the week). I then built four different kinds of classification models to predict whether prices would go up or down based off the changes of position of the various kinds of traders, and I found that the data led to correct predictions about 62% of the time across all four models. This strongly suggests that the COT can be effectively used as a guide for speculators as the legend has long claimed, but that it should not be fully relied upon. I encourage readers of this blog not to take my word for it, but to conduct their own analysis and see for themselves.
A false set of books! (Clue here… yet seems scant interest) Posted all this many years ago and yet the scam still goes on. Imagine analyzing Enron back then when your data points were based upon a phony set of books.