A few years ago, I read a blog post claiming that the Eurodollar Commitment of Traders report could be used as a leading indicator of the US stock market. I was skeptical, but intrigued, but I never really dug deeper, until now. I wrote a script to load the CoT reports from the CFTC into InfluxDB (fed into Quantopian using my CSV proxy app).
For each CoT report, the value I send to Influx is the ratio of net-long to the total open interest, calculated thusly:
oi = float(r.get('Open_Interest_All'))
long = float(r.get('Commercial_Positions-Long (All)'))
short = float(r.get('Commercial Positions-Short (All)'))
val = ((long - short) / oi) * 100
Then, I shift the CoT data points forward 52 weeks, and watch a short-term and long-term SMA on that ratio. When the trend if going up, I buy the S&P, when it is going down, I short.
As you can see from the results, it doesn't seem to actually work very well (except in Oct. 2008, when it did really well). Does this just prove that the theory behind the Eurodollar CoT being a leading indicator is bogus? Am I misinterpreting the theory? Can I improve on this idea?
Thanks for any ideas!
