The worst thing an indicator can do is not to be consistently wrong but to have no edge whatsoever. This is because if it is consistently wrong or almost always wrong, then it can be used as a contrarian indicator. Jim Cramer would be a good example. If on the other hand the indicator is more or less right on the money, then it is a leading indicator. A good example would be the OEX options market.
The TickerSense blogger sentiment poll is useless as a sentiment indicator because it has no edge, it is just random data. Noise. To this useless sentiment indicator, we can add another one: Citigroup’s (NYSE:C) Panic/Euphoria Model.
This is an proprietary aggregate measure of sentiment from the quants at Citigroup and is published weekly by Barron’s in their sentiment section. If anyone had any doubts as to the value of this indicator from Citigroup
As we suffered one of the sharpest drops in stock market history, as volatility spiked to record highs, as sentiment measured by traditional investor surveys reached epic pessimism, as breadth redlined, the Citigroup’s Panic/Euphoria Model yawned and continued napping:
I’m not sure what goes into the calculation of this number and frankly, I don’t care. Whatever recipe Citigroup Investment Research is using, it is garbage. What I can’t understand is how anyone can continue to come in in the mornings and crunch numbers for this indicator when it is completely separate from reality. Why is a reputable publication like Barron’s legitimizing this drivel by disseminating it every week?
Although the make up of the indicator is secret, there are some hints that Citigroup has tweaked it in. You can see that in the chart below (from April 28th 2008), the low in March 2008 does not coincide exactly with the same trough in the most recent iteration of the graph above. The low in March was deeper (more “panicky”) and it registered closer to the end of the month while in the most recent data, it is shallower and occurs mid-month.
The problem with tweaking garbage is that even if Citigroup corrects it, the data won’t be historically comparable. Just like TickerSense, the best thing to do is to acknowledge reality and discontinue it.