A Suspect Rally

The average stock in the Russell 1000 was up more than 6% from Tuesday through Friday, Aug. 25 to Aug. 28. Bespoke Investment Group ran its famous decile analysis on these results to see which factors have led the way higher, and which have lagged.

To run this analysis, the analysts break the Russell 1000 into deciles (10 groups of 100 stocks each) based on the various stock characteristics such as valuation, overseas revenue exposure, institutional ownership, percent gain/loss in prior month, and amount shorted. They then calculate the average performance of the stocks in each decile over the last three days.

Bespoke said Friday that the main takeaway from the analysis is that last week’s bounce was driven by investors and traders buying up the stocks that were beaten down the most during the huge 10%-plus correction witnessed from Aug.18 to Aug. 25.

The decile of stocks that fell the most during the correction gained an average of 11.53% since the low, which is far more than any other decile. The short interest category is also a standout, as the most heavily shorted stocks have bounced significantly more than the least heavily shorted names.

This is important because it suggests most of the surge higher last week was short-covering, not real bargain hunting. There’s a big difference. Major rallies always start with short-covering, but later have to be followed up by investors buying the best stocks at lower prices. Until the latter shows up, the rally will be suspect.

Also I would note that rallies of 3% or more are characteristic of bear markets, not bull markets, for this very reason. So when bulls see a 3%-plus gain in an index, it should make them worry, not celebrate.

The Bespoke analysts’ other findings:

— When broken up by market cap, the decile of the smallest stocks in the Russell 1000 significantly outperformed all other deciles. Small caps got hit hardest during the correction, and they have bounced the most since then.

— Stocks with low or no dividend yields slightly outperformed, but high yielders (like energy producers) have also done well during the rally.

— Performance is almost a straight line higher when going from the least to the most heavily shorted deciles. That proves there was a whole lot of short covering going on.

Breadth Thrusts

Jason Goepfert’s work on sentiment for the past few years has been sensational. The Sundial Capital founder and head analyst is one of the best around at studying and learning from the key technical, psychological and emotional issues facing investors in times of panic and their aftermath.

He pointed out Thursday that buying interest has been unusually aggressive on the NYSE in the rally that followed the Monday crash. More than 85% of total volume on the exchange went into issues that were positive on the day, coming on the heels of what had been excessively oversold conditions. That combination has almost always led to further gains in the months ahead, though there were often serious stumbles first in the short term.

Goepfert recalls that decades ago, pioneering market analyst and money manager Martin Zweig popularized the idea of watching for such “breadth thrusts,” or stretches when a) stocks were very oversold, then b) buying interest came in so aggressively that it greatly skewed breadth data higher. This situation shows that buyers were so desperate that they were buying almost every stock and almost all volume was going into those positive stocks.

Goepfert notes there have been multiple occasions to discuss these over the past 15 years and almost all of them have worked out, preceding a further rise in stocks over the intermediate-term.

We saw only two days of buying pressure off of the panic last week but they showed the kind of buying pressure that has preceded other breadth thrusts, he observes. Up volume on the NYSE has exceeded 85% each of the past two days. This comes on the heels of what had been a very oversold 10-day average of Up Volume, and a historically oversold 5-day average. It has been rare to see a more depressed 5-day average than seen Tuesday, he reports.

To put this view into actionable form, Goepfert went back to look for other times that the 10-day average of Up Volume was oversold (below 45%), then there were back-to-back days of at least 80% Up Volume. He also separated out those instances when the 10-day average was the most oversold heading into the up thrust.

Goepfert’s data shows that the dates that most closely mirror our current situation were 30-Oct-1962, 30-Oct-1987, 29-Nov-1971 and 18-Dec-2014. Those were the dates when the 10-day and 5-day averages were the most oversold, and the back-to-back days of Up Volume were the most impressive. Stocks struggled in the month after the 1987 and 2014 occurrences but did very well in the months after that.

In summary, the type of trading seen Aug. 26-27 is what bulls would like to see. It has not been foolproof (big fail in 2008) but when you see a historic level of panic followed by multiple days of aggressive buying pressure, it has almost always led to intermediate-term gains.

Goepfert concludes by warning that this view does not preclude short-term testing of Monday’s panic low or even a modest drop below it, but it would be rare to see a protracted decline under that low, an event only seen during the worst bear markets.

Successful analysts and hedge fund managers use a wide array of data, including lunar cycle observation.

Lunacy, or Legit?

And now for something completely different, let me note that I learned a long time ago that successful hedge fund managers look at a lot wider array of information than you might imagine. There is a lot of interest in arcane Gann analysis, there are some that care about the positions of stars if not astrology, and there is often a lot of attention paid to lunar cycles and solar storm cycles.

I have seen repeatedly that if a phenomenon in the physical world can be tied to security price movement in a way that can be scientifically proven to work, at least in some circumstances, it will be ingested and used in a decision matrix. Every manager is looking for an edge.

With that preamble, here is the lunar cycle observation made by one of my favorite analysts with a broad appetite for data: He reported Friday that when the Dow Jones Industrial Average is up more than 1% two days in a row, and the index is down over the prior six days, and it’s one day before a full moon, then the average return of the DJIA in the next 12 days is -5.36%.

This is a condition that described the Dow on the Thursday close.

The condition of a full moon appearing after a two-day rally amid a weak market previously occurred eight times since 1928. The worst two results were -22% (March 1938) and -11.2% (Feb 2009). Other times the results were more innocuous: 1.5%, -4%, -1%, -3.7%, -2.8% and -1.4%.

We’ll check back in two weeks to determine whether this analysis shot the moon or belonged in the loony bin.

Best wishes,

Jon Markman