Blogs / Analysis Blog

Characteristics of “Junk” Rallies

A lot of people have been focused on what they called the “junk” characteristics of the stock market rally that began in March. The implication is that there was a massive short squeeze [?], which is another way of saying that too many people were positioned short the same unpopular stocks and then were forced out of their positions, forcing others out of these shorts, etc., and ultimately driving up the market. The evidence for this short squeeze is that stocks with high momentum and low short interest vastly underperformed in the rally.

Below are returns of the first two weeks of April for Price momentum and Short Interest bucketed into quintiles, using Instrument Explorer:


Short Interest:

That the stocks with the highest short interest and lowest price momentum outperformed their peers on the opposite end of the spectrum isn’t the only thing that’s interesting. It is important to note that the relationship between factors and returns is monotonic—this is how we know that the outperformance isn’t just random.

So this leads us to the question of what has happened in similar situations—if the rally is a “junk” rally, what does that mean? There are a lot of ways to go about asking this question. In order to get an approximate measure of the monotonicity over time, we created an index for each of the short interest quintiles. The cross section of the weekly returns of the quintiles are then correlated to the monotonic series “1, 2, 3, 4, 5″. A correlation of 1 will mean that there might be a short squeeze while a correlation of negative 1 implies that the shorts are getting things right.

The correlation (in blue) is very noisy; apparently short squeezes are common, and the relationship reverses often. We will try to cut out the noise by looking at the 8 week moving average of our Index (green).

As expected, there has been a recent spike in positive correlation. This has happened in 2005, late 2004, early 2003, late 2002 and the most extreme was the late 1999.

The end of the last bear market is particularly interesting. There was a rally in 2002 which was accompanied by a short squeeze. The market fell after this rally but did not reach new lows, and the next short squeeze led the way into the new bull market. Interesting, but so far there is no clear cut answer as to what short squeezes might signal.

Another way to measure the short squeeze is to look at the return spread of the highly shorted quintile minus the return of the lower shorted quintile. When this return is higher than the return of the S&P 500 then it can be said that the short squeeze is helping drive the market up. One way to calculate this is to look at the high minus low return minus the absolute return of the S&P 500. When this number is above 0, a short squeeze is helping push the market up. When the 4 week moving average is above 0, this could be a sign that a rally has been driven by a short squeeze, and perhaps should be faded.

One possible quick and dirty analysis looks at the day after the short squeeze indicator passes the threshold and measures how the S&P performs over the next two weeks, as shown in the table below.

The annualized return here is a slightly negative 0.23%, or zero for our purposes given the SPX return over the period, meaning the indicator doesn’t give any direction. Here we allowed re-entry if the line passed back above; limiting each segment to only two weeks gives a slightly positive return, which is probably just noise. If we set a higher bar—only look at even more significant short squeeze rallies according to this indicator, and look at the SPX performance for two weeks after—the market return is more positive, and it’s up to the judgment of the analyst whether or not this is significant.

A system Indicator Twiddle in Strategy shows that when short interest gets to be more important, the market has tended to rally. In our Twiddle we vary the threshold level for our indicator (with 0 matching the Chart above). The results here denote the profit factor, which is defined as the ratio of (number of winning trades * average winning trade P&L) / (number of losing trades * -average losing trade P&L). The numbers would be even more positive—significantly so—if not for the huge sell-off after September 15, 2008, when short interest importance was near an all-time high. Like many indicators, this one broke badly in the market crash of late 2008.

There are several follow-up studies that could be interesting here; sector-specific analysis yields interesting results. And although the market direction is not clearly predicted by the importance of short interest, it seems to have significance for upcoming volatility. Also, testing whether or not the short-interest factor is meaningful in general, and how meaningful it is (e.g., by looking at the absolute value of the index we built) also yields indicators that are carefully watched by various hedge funds. Similarly, the importance of other common factors that describe the behavior of market participants, such as momentum, earnings momentum, growth/value, etc., is also a significant area for research.

From this we’ve learned that there is evidence that a short squeeze has helped lead the way out of sell-offs to sustainable rallies in the past, and that at least in the simple way we’ve defined it, a short squeeze is probably not something to trade against for its own sake.

Other Blogs