When working full time at a small hedge fund I did a full sentiment report once a week. A sentiment extreme is not a trading signal in and of itself but we did have notable sentiment extremes on the January and February market lows, as well as the mid August top. I wrote at the time that such extremes often contribute to a key high area - and they did.
There are other methods out there - The Fat Pitch blog also does a good job of writing up the BAML asset managers survey, which I think is worth considering as well especially from a longer term perspective.
But still the four I track are working pretty well (along with pivots, other technicals, valuation & fundamentals and timing) in getting turn areas. Since August the bullish extremes have been worked off, and we understandably saw the two specifically option related meters (put-call and ISEE) make bearish extremes into the election. But now things are moving the other way.
Keep in mind that November through January is seasonally bullish, and we are likely to see higher than usual sentiment numbers that may not drop the market in the same way as the rest of the year.
Daily put-call is dropping back into normal, but not at bullish extreme yet; weekly put-call nowhere near extreme.
ISEE has yet to make daily spike highs, which we are likely to see near tops.
NAAIM as of last week is nowhere near extreme, but probably made a big jump this week.
But AAII bull % has soared to 46%, by far the highest of the year.
Data from 2005: 14% percentile (meaning only 14% were higher than this one)
Data from 2010: 12% percentile
Data from 2012: 10% percentile
Data from 2014: 7% percentile
Data from 2015: 3% percentile
Data from 2016: highest value
So what matters here? This is either somewhat high, extreme or notably extreme depending on where you start your data series. Thus conclusions would be approaching a top but not yet, upside limited, or cut longs and perhaps short depending. This is exactly why I am against pure quantitative analysis without a qualitative component (everyone seems to be a big fan of the quant side these days, while forgetting that how they select the data impacts the results). Let's consider the market environment. We are in a post QE world, but it was tapering through most of 2014. So I think the numbers that matter here are 2015 and beyond.
Sticking with that year for the remainder:
Bear %s only 35% percentile, and the bull-bear spread 7% percentile.
As of now this and the other meters are things to watch. The more the others join in a sentiment extremes, the more likely we are to see a shakeout of weak hands and possible top area. This means I am now watching for lower put-call, daily ISEE spike highs (125+), NAAIM mangers 90+, and a bit fewer AAII bears to warn of a major top.
But still, as long as SPX and INDU hold above YR1s, bulls have the ball.