Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
Citigroup’s Panic/Euphoria model, featured in the July 8 commentary “Don't Miss Your Chance to Catch a Bull Market,” fell into panic territory at the end of June 2011. According to the model’s originator, Citigroup strategist Tobias Levkovitch, this indicates a roughly 90% probability that equity prices will be higher in six months and a 97% chance of gains in 12 months.
In contrast, my model, as explained in “Improving on Buy and Hold: Asset Allocation using Economic Indicators” generated an initial sell signal early in May 2011 and forecasts a declining market from the middle of August 2011 onwards.
Citigroup’s model should not be relied upon to predict market direction. For example, the historic chart of the model for the 2007 and 2008 period incorrectly signaled positive forward returns for the market early in 2008. Now, the current chart has been modified so the model looks as if it had correctly indicated negative forward returns in 2007 and 2008.
The Panic/Euphoria model
The model uses the Market Sentiment Composite, a measure of investor sentiment. This metric tracks the mood of investors, which is then translated into a probability of the market's advancing or declining over the near term and over the next 12 months.
It is supposed to be a contrarian indicator that produces a bullish signal when market sentiment is overwhelmingly negative, and vice versa. The model’s methodology is undisclosed, and no statistical data is available to support the claim that the Composite value is an indicator for the 12-month forward return of the stock market. For example, Citigroup does not disclose the correlation coefficient used to measure the strength of the linear dependence between the Market Sentiment Composite and the 12-month forward return.
Figure 1 shows the Panic/Euphoria model. There are two horizontal lines: the upper green line (the euphoria line) and the lower red line (the panic line). When the Market Sentiment Composite (the blue line) is below the red line, this is considered bullish, and it signifies a positive 12-month forward return for the market. And conversely, when the Composite is over the green line, it signals a negative 12-month forward return.
In order to test whether the Panic/Euphoria model can provide useful trading signals for the stock market, I have superimposed buy and sell signals for the S&P 500 on the Panic/Euphoria chart. Buy signals are assumed to occur when the Composite penetrates and falls below the red panic line (in expectation of positive forward returns), and sell signals should occur when the Composite rises above the green euphoria line (in expectation of negative forward returns). This is shown in figure 1.
I then transferred the signals onto the S&P 500, which is shown in Figure 2. The value graph in this figure indicates the value of an investment made in August 1988 in the S&P 500. Following the buy and sell signals to July 2011, one would have had an average annual return of 11.74%, while a permanent investment in the market, excluding dividends, would have produced an average annual reurn of only 7.11%.
The projected return based on the signals from the model seems impressive, but there is a problem when one investigates the positions of the euphoria and panic threshold lines and the historic values of the Market Sentiment Composite, which were instrumental in generating the buy and sell signals.
Reading the model more closely
The sell signal at the beginning of 1999 occurred about 18 months before the market peaked. As can be seen from Figures 1 and 2, the market was considerably higher 12 months after the signal date than it was on the signal date, even though the model required it to be lower. One would have to shift the green euphoria line up to a Composite value of about 0.90 for the model to have predicted a negative 12-month forward return. However, moving this line upward would not have produced the September 2007 sell signal, because the euphoria line would then not have intersected the graph of the Composite.
It would appear that the green euphoria line was retrospectively placed in a position to make both signals possible and to make the model accurately reflect past market conditions, thanks to the benefit of 20/20 hindsight.
Figure 3 shows the chart of the model as published in October 2008. The chart is for the period October 2007 to October 2008, which was taken from the November 6, 2008, article Citigroup's Panic/Euphoria Model: A Useless Sentiment Indicator.
In this chart, the euphoria line is at +0.6, and the panic line is at -0.3. The model predicts a positive 12-month forward return at the end of January 2008 and also in the middle of March 2008, when the graph of the Composite is well below the panic line. The May peak of the Composite is below the euphoria line, thus never signaling a negative forward return.
On January 20, 2008, the New York Times, in Burned but Bullish at Citigroup, reported that Mr.Levkovitch, the model’s originator, predicted a 26% gain for the S&P 500 by the end of 2008. This would be consistent with the model’s forecast at the time. We know now what really happened – 12 months later, the market was about 40% lower.
In the current chart (Figure 1), the data from October 2007 to October 2008 do not coincide with the historic data (Figure 3) for the same period. Figure 1 displays the following changes:
- The euphoria line has been lowered from 0.60 to +0.45.
- The panic line has been lifted from -0.30 to -0.18.
- The May 2008 peak value increased from 0.60 to 0.70.
- A new January 2008 peak with a value of 0.60 has been added.
- The low points of the Component for January and March 2008 changed in value from -0.60 and -0.50 to about +0.20 and +0.10, respectively.
The effect of these changes is that the current model, as depicted in Figure 1, wrongly claims that it predicted a negative 12-month forward return in 2007 and 2008, when in fact the unmodified 2007-2008 chart of the model (Figure 3) forecasted a market gain. Surprisingly, according to several websites (for example, Why There’s a 97% Chance of a Big Market Rally), Mr. Levkovitch now writes, “Note that this proprietary metric … surged into euphoria in both 2007 and 2008, implying forward equity market weakness as well.”
One can only conclude that this model was recalibrated after 2008, with changes made to the historic parameters and data, in an endeavor to show that it correctly forecasted the equity market weakness that occurred in the second half of 2008 and during 2009.
Also, if the panic line had been maintained at the 2008 level of -0.3, then the current position of the Market Sentiment Composite would not be in panic territory and therefore would not predict a positive 12-month forward return for equities.
The shape of the Panic/Euphoria lines
Instead of the threshold lines being straight, they are actually curved. In Figure 4 the red dashed line above the Composite (the blue line) indicates the approximate time when the market produced a 12-month forward return of 5% or more. The green dashed line below the Composite shows when the market lost 5% or more 12 months later.
These red and green curved dashed lines are the “correct” panic and euphoria lines for the current model. They do not follow any particular trend and cannot be projected forward in a mathematically meaningful way. Thus, no prediction can be made about the market’s 12-month forward return, because one does not know where to locate the threshold lines for the August 2010 to July 2011 period. One will only find this out later when the market’s value from August 2011 to July 2012 is known.
No information about future market direction can be obtained from this model. As others have observed before, it is a useless sentiment indicator.
Georg Vrba is a professional engineer who has been a consulting engineer for many years. In his opinion, mathematical models provide better guidance to market direction than financial “experts.” He has developed financial models for the stock market, the bond market and the yield curve, all published in Advisor Perspectives.
Read more articles by Georg Vrba, P.E.
