Earlier today I posted this below at Bill Luby's superb Vix and More blog. Excellent sentiment analysis is always rare, as one would hope. It requires art and even perhaps the talents of a forensic psychiatrist. Bill's site is the best. I have seen nearly all sentiment sites available to private traders.
The thread at Bill's shop today concerned sentiment failure, and the learning experience it provides for skilled practitioners. I re-edited this post a bit to correct some minor errors:
I think that sentiment "failure" showed itself at the July top and then was confirmed at the first low. The 2CS, which I define as the five day running total of daily VXO and the CBOE combined P/C, had been giving multiple deliriously bullish readings (<50) in late 1996 and then at the February 2007 highs and once only in April as the indices resumed their climb out of the March 2007 lows. But as the market marched up to July, the 2CS began to rise. It was 71 at the July high. This was a major change of sentiment character. This was a reversion to 2CS range extremes seen at rally highs during the bear market to 2003.
I mentioned my monitor label, "sentiment divergence happens", but that is a short term happening. When divergences are as great as that from February to the July highs, there is definite range shifting going on, and that means that the markets are "in the process of" making a longer term top.
From the July highs to the end of July the 2CS was already nearly 150, not seen since the first half of 2003. This was the confirmation of probable longer term trend change. The delirious bullish high range had shifted but also the rather placid low range had now become somewhat delirious. Bearish delirium reached 198 on July 17.
By then it was evident that a broadening distribution top, called different things by different people--reverse point waves, expanding triangle, etc.--was also forming in many indices. Since distribution in a top takes a long time, it was reasonable that it should last longer that three months from May to August.
On the rally to October new highs 2CS only got as low as 63. No more delirious 50's or 40's as in 2006 and up to February 2007. 63 is almost exactly the number attained on the rally into May 2002 and on other bear market rally tops. The top was in.
Personally I did not run out that day and shout in the streets or liquidate everything and load up on puts. I had been trimming into strength all year.
Sentiment does diverge and it exhibits range-shifting longer term, as it also did in 2002/2003 at the lows. To grasp it requires a mental shift from the notion that the majority is always wrong at the wrong times, as you clearly see. Range shifting is the sentiment equivalent of Dow Theory price waves. I think it's even better.
Chart 1 shows monthly and longer term 2CS sentiment (in red and, notably, inverted to price) at intermediate term highs and lows from 1997-2007. After the 1998 lows it was reasonable on many technical grounds to expect a large distribution top. Despite the internet and junk bubble extending gains, sentiment began to cool decidedly by March 2000 and the April low.
I mentioned my monitor label, "sentiment divergence happens", but that is a short term happening. When divergences are as great as that from February to the July highs, there is definite range shifting going on, and that means that the markets are "in the process of" making a longer term top.
From the July highs to the end of July the 2CS was already nearly 150, not seen since the first half of 2003. This was the confirmation of probable longer term trend change. The delirious bullish high range had shifted but also the rather placid low range had now become somewhat delirious. Bearish delirium reached 198 on July 17.
By then it was evident that a broadening distribution top, called different things by different people--reverse point waves, expanding triangle, etc.--was also forming in many indices. Since distribution in a top takes a long time, it was reasonable that it should last longer that three months from May to August.
On the rally to October new highs 2CS only got as low as 63. No more delirious 50's or 40's as in 2006 and up to February 2007. 63 is almost exactly the number attained on the rally into May 2002 and on other bear market rally tops. The top was in.
Personally I did not run out that day and shout in the streets or liquidate everything and load up on puts. I had been trimming into strength all year.
Sentiment does diverge and it exhibits range-shifting longer term, as it also did in 2002/2003 at the lows. To grasp it requires a mental shift from the notion that the majority is always wrong at the wrong times, as you clearly see. Range shifting is the sentiment equivalent of Dow Theory price waves. I think it's even better.
Chart 1 shows monthly and longer term 2CS sentiment (in red and, notably, inverted to price) at intermediate term highs and lows from 1997-2007. After the 1998 lows it was reasonable on many technical grounds to expect a large distribution top. Despite the internet and junk bubble extending gains, sentiment began to cool decidedly by March 2000 and the April low.
The same was true for the 2001-2003 lows. In both cases it took about 18 months to play out from the first disjunction or divergence point and about six months from the divergence. Although post is primarily about 2007, the early warning signs were already present in 2006 at the May high and June/July lows. Delirium reached it's peak in May 2006 but at the summer lows bearish sentiment greater than any since 2003 was seen. At the time this was seen, correctly, as being quite bullish, but it also was an early warning.
Chart 2 is a daily close-up of the 2007 action, already discussed, with a summary of my interpretation. (Click on chart images for a larger view.)
Recent Comments