The Big Picture

« New Column: The Street Gets Inflation Threat Backwards | Main | How Much is My Blog Worth? »

A Year of Panic Redux

Saturday, February 11, 2006 | 07:13 AM

That didn't take very long, did it?

Last week, we discussed Citibank's Panic/Euphoria chart in A Full Year of Panic ? (I don't think so). In this week's Barron's, Mike Santoli gives Citigroup's Tobias Lefkovich an opportunity to defend the indicator from a barrage of criticism.

(Incidentally, Barron's is free this week, so you can read the entire discussion there).

I have several beefs with the Citigroup's proprietary Panic/Euphoria Market Sentiment indicator:

As we mentioned last week, Panic is a momentary phenomenon -- it is transitory. But as the chart at right shows, we have been at Panic levels for about a year. That makes little sense.

Second, the basic concept reflects a flawed understanding of sentiment and contrarianism. Markets go up when the crowd is Bullish; Markets go down when the crowd is Bearish. Merely leaning against the crowd is a sure fire path to the poor house. That concept is the basis of cliches such as "Don't fight the tape" and The Trend is your friend."

If you want to use any tool as a contrary indicator, then you are looking for specific extremes -- not a year's worth of sentiment reading. We detailed dozens of examples in the Contrary Indicators piece here.

Next, I must take issue with this statement: "negative sentiment is the condition that tends to precede higher markets by six to 12 months." That's simply incorrect, and for several reasons:

- For the most part, markets tend to go up over time. Saying markets will be higher in a year because of XYZ is disengenuous; They statistically are likely to have been higher anyway;

- How much higher? Most U.S. markets are flat to up moderately over the past 12 months; The prior panic indicators don't seemt o be working (Dow up 1% is hardly a victory)

- The data mining issue is ever present in these backtested models. This tool needs to be tested in the real world -- prospectively not form fitted to a past that will not repeat precisely.

There is an easy solution to turning this marketing tool into a potentially useful technical indicator

The two peaks -- under the 0.6 level I referenced last week -- were decent lows: Early May 05, and mid October 05. Tobias needs to adjust his scale, and move the Panic zone down to 0.6.

Here, I did a quick mock up of what that would look like:

May 2005 was a good low, as was mid October 2005. Lower the panic thresh hold, and maybe you got something. (Must I do everything around here? Sheesh)

I understand the difference between marketing and analysis, and this is as fine a piece of marketing as any. With a few tweaks, maybe it could becoem a decent analytical tool . . .

Source:

The Bull Peters Out
Kopin Tan and Michael Santoli, THE TRADER
BARRON'S MONDAY, FEBRUARY 13, 2006
http://online.barrons.com/article/SB113961724924371315.html