David Korn makes the interesting observation that the sentiment data from the AAII are highly skewed toward the bears. Indeed, with roughly 58% bears and only 24% bulls, this is one of the most bearish readings since 1989 (N = 906 weeks). Of course, I wanted to see if such highly bearish sentiment truly is associated with intermediate-term market returns.
First off, we've only had 25 weekly occasions in which bears have been over 50%. Of these readings, 16 occurred during the late 1990-early 1991 period, which was a major market bottom. Other times we've seen over 50% bears have been February-March, 2003 (another major bottom); individual weekly readings from July - October, 2002; and individual readings in March and December, 2000.
On 20 of these 25 occasions, the Dow Jones Industrial Average (DJI) has been down for the 10 weeks *prior* to the bearish sentiment reading. In other words, bearish sentiment has tended to occur after prolonged market weakness. This is true of the current bearish sentiment reading as well.
When bears have exceeded bulls by more than 20% in the AAII survey (N = 46), the next ten weeks in the Dow have averaged a gain of 4.48 (37 up, 9 down). That is a whopping advantage over the average 10-week gain in the Dow of 1.93% (596 up, 310 down) for the entire sample.
Conversely, when bulls have exceeded bears by 30% or more (N = 142), the next 10 weeks in the Dow have been up by an average of only .83% (78 up, 64 down)--less than half the average return for the sample.
Interestingly, when bears have exceeded bulls by more than 20% and there has been a bullish advantage going forward, the *previous* ten weeks in the Dow have been down on average by over 4%. When the bulls have exceeded the bears by more than 30% and there have been subnormal returns going forward, the previous ten weeks in the Dow have been *up* on average by over 4%.
So there you have it. Down markets followed by extreme bearish sentiment have tended to be bullish in the intermediate-term; up markets followed by strongly bullish sentiment have tended to yield subnormal returns in the next two months. Occasions in which we've had more than 50% bears have tended to be either major cyclical bottoms or intermediate-term low points during bear markets that have preceded sharp rallies. Please note that these are small samples, particularly given the clustering of bearish readings at certain points in time.
Per the Trading Psychology Weblog's Monday entry, note also that we have been seeing a recent reduction in stocks registering new intermediate-term lows--a surprising finding as bearishness has expanded.