Both the chart and COT patterns of this year are strikingly similar to 1996. 1) In both periods, the S&P hadn't seen a major drawdown in 4-6 years. 2) In both periods, commercials became more bullish around March and continued to become more bullish throughout the summer, even during the drop. 3) In both cases, prices broke out of a downward sloping channel to a new high, only to fail and drop below the 200 DMA, giving the S&P its largest decline in years. If you look at the first test of the 200 DMA (red line) in the 1996 chart, you'd think you were looking at 2007 YTD. That doesn't mean the rest of the chart will turn out the same, but it shows that technically, we're not seeing anything "new" or "unusual." Oh, and COT data is available here: http://www.cftc.gov/cftc/cftccotreports.htm
If you notice the commericals had a different position in the large SP contract vs the Emini. I noticed that they were short the whole way up too and decided that it was not enough information to figure out where they were going. I would guess that there are hedges behind the positions so who knows what they are really thinking. I don't look at the report any more except for the position of the small speculators on the emini's. When they get super bullish or super bearish the market tends to turn somewhat. John
Agree completely, COT makes for some good market folklore until you test it with rigorous statistics. Commercials almost always fade moves; it doesn't mean they're accumulating a position of course, they're going to take the other side of paper, then arb and/or hedge off the risk with spot. But when you look at the COT chart, it's very sexy when you see them fade a big move which reverts to the mean. I think it just plays into people's need to believe there's some omniscient party out there who always makes money and can be shadowed profitably.
Not telling anyone what to do. However, if this site were around in July of 1996, it would be filled with all kinds of bearishness and 10,000 reasons why the economy and equities are going to hell in a handbasket. Everyone would say the correcton was far from over...it would have to drop another 10 to 20%. Of course, they'd be wrong. I'm just saying that it doesn't look any different this time, technically speaking. I'm not "predicting" anything. One more thing--the volume usually tapers off near the end of a selloff. We definitely saw that on Friday...much lower volume than at the beginning of the pullback and much less than the reversal on Wednesday.
Scott Barrie of CFEA did some good work refuting COT as a reliable indicator; I think he has a TASC article floating around which you can google for. I have no interest in picking a bone with you, if you have a quantifiable edge using COT in the S&P, keep it to yourself and trade up a storm, because it's very valuable. All I'm saying is use common sense. A completely random long-only mean reverting strategy in the S&P will always appear to be an edge, until you correct for upward drift. If Barclays or Susquehanna decides to hedge a massive cash MOC trade with futures and it's reported as a 100,000 car commercial S&P long, you can wipe your ass with that information - it's just another random variable.
Fair enough. I certainly don't rely on COT alone, just as I wouldn't rely on seaonality or anything else as a stand-alone system. FWIW, as I said earlier, TASC has also published articles validating COT's reliability more recently. Here's a completely different indicator that's detecting a bottom from Jay Kaeppel (disregard the site--I'm not into the Optionetics stuff, but like some of Kaeppel's work): http://www.optionetics.com/market/articles/article.asp?id=17776 It's been very reliable recently, though the results are far from statistically significant. However, I've seen and tested similar systems that have a good longer-term track record. Sell signals should be based on sentiment reversals, short-term overbought conditions or a modest profit target. These systems don't necessarily tell you a major rally is about to occur, just that a short-term bottom is near.