Stock Manipulation — AEO
Some friends of mine are convinced that loss of the up-tick rule, the increase of hedge funds, and a robust market of short sellers make the stock market ripe for equities manipulation. Their complaints tend to focus on their own holdings — convinced that any release of good news should prompt an advance in the stock price, rather than the declines that frequently follow.
My theory, when it comes to American Eagle, is different. Specifically, more than 100% of the outstanding float is held by insiders and institutions (23.28% and 79.30%, respectively). If these are considered the more stable holdings (i.e., less frequently traded), then the relatively small number of remaining shares seem likely to have a disproportionate influence on the stock price and its movement from one day to the next, as the stock is thinly traded by short-term longs and shorts.
To test this theory, I downloaded the daily stock prices and volumes for American Eagle since January 1 of 2000 — providing more than eight years worth of data — and crunched the numbers.
First, let’s consider the issue of whether large variances between the opening and closing prices are strongly correlated with the volume of shares traded on a given day. We may assume that on days when significant news arrives in the market, significantly higher levels of activity (volume) would occur, as the market responds to the news. This, however, does not appear to be the case, as this correlation table indicates:
Indeed, when grafted as a scatterplot, we see the frequency with which significant price moves occur on low volume:
Of course, the cynic may argue that a “holiday” effect is present, where news is released on days when the market is taking summer vacation, celebrating the December holidays, etc. Since significant holidays tend to occur over multiple days, we should consider changes in volatility for, both, stock price and volume from one day to the next. The next chart does precisely that:
With both of the scatter plot charts, I have circled those events where significant price moves occurred on exceedingly small volume days.
And this informs our consideration of stock market manipulation, in general, and AEO in particular. Here are my conclusions, as posted on a different website yesterday:
At this point, it is worth noting two things. First, while “means, motive, and opportunity” must be present to commit a crime, these three attributes are not sufficient for a crime to have been committed. Who among us has not had the means, motive, and opportunity to commit nearly any crime you can name (murder, rape, jaywalking, littering, etc.) but elected not to commit the act? Guilt, on the one hand, and suspicion of guilt, on the other hand, are not synonymous.
Second, the first item should not indicate that I believe no manipulation of the stock occurs. Personally, I believe it occurs with greater frequency than most of us realize. In fact, in the early 1960s, my father-in-law was charged (but never convicted) of stock manipulation — at a time when the up-tick rule was in place and shorting was less prevalent. Manipulations, however, are short-duration events (i.e., lasting less than two years), because it is simply too difficult to sustain the farce for longer, in the face of sustained and audited financial performance by the underlying company. As the expression goes, eventually, “the truth will win out.”
Given these two realizations, the folly of short-term trading (long or short) becomes evident. It is a folly because, for the average investor, there will always be someone with deeper pockets and more sinister motives, intent on divesting you of your invested capital.
Shortly after posting this, I received a response from Bill (a good friend who often serves as devil’s advocate), questioning whether the stock is, indeed, thinly traded — since average daily volumes exceed 500,000 by a large amount. My answer:
Your comment indicates that average share volumes in excess of 500,000 shares per day represents a large volume (compared to American Eagle’s daily volume of 4.16 million shares traded — just over eight times that 500,000 share threshold). The 500,000 share standard, however, serves a different purpose than identifying whether a disproportionate change in price follows from the number of shares traded. For most investors, the 500,000 figure indicates sufficient volume-liquidity exists to exit their position when the time comes.
So, let’s look at the question quantitatively, to determine whether my contention that 4.16 million shares lends itself to disproportionate price changes.
First, something in excess of 4 million shares traded daily sounds abundantly large. There are, however, 296 stocks traded on the New York Stock Exchange with larger daily share averages (out of the 2417 stocks listed on the NYSE). By volume, American Eagle is just outside of the top 12% for the exchange.
Second, those 4.16 million shares produce a 14-day $0.786 average true range for the stock (trading at today’s close of $18.67). This average true range represents a 4.21% average change. While this may seem large, it is down significantly from the $1.3 average true range from late January and early February of this year, and the March high that was in excess of $1.2.
Third, the average daily volume of 4.16 million represents just over 1/10th of 1% of the 383 million shares outstanding. In other words, just over 1/10th of 1% of shares are prompting price movements in the stock that are in excess of 4.21%.
To determine whether this disproportionate effect is unreasonable, I first performed the correlation analysis – comparing intraday and two-day movements in the closing price in comparison to the volume of shares traded, going back to January of 2000. The correlation came in at 0.39, out of a perfect 1.0. In other words, there is almost no correlation between the two.
Next, I created scatter plots for both tests, in order to graphically represent that which the correlation results indicate. In both cases, the circled areas on the graft indicate the abundance of events where significantly disproportionate results were evident.
As mentioned previously, I have no idea whether the stock is being manipulated. All that I can reasonably conclude is that the volatility in the stock’s price follows from a disproportionately small number of shares traded. Because those traded shares are not randomly drawn (reflecting a proportionate mix between those holding short and long positions and those investing in the company versus those trading the stock), it seems evident to me that the small number of shares prompting a disproportionate movement in the price undermines the “efficient market theory.”
After this posting (above), Bill suggested that the efficient market theory may remain intact if viewing daily volumes and price movement as akin to a sampling survey — similar to polls taken by Gallop, The Wallstreet Journal, and other polling organizations. I responded:
When we perform population sampling (with medical research, social sciences polling, market research and analysis, etc.), we tend to sample a smaller population pool than the population as a whole, because a complete sampling of such a large pool would be logistically or financially impractical. As you note, this smaller than total sampling introduces an element of uncertainty – which is commonly measured as a p-value, with, both, parametric and nonparametric statistical analysis … MANOVAs/ANOVAs and chi-squared analysis, to provide examples of both.
In order for the sampling to be legitimate, it needs to be, both, random and consistent with the larger population. The consistency requirement should be thought of as a check on the legitimacy of the sample’s randomness. For example, we may seek the opinions of Democrats favoring Hillary Clinton on their support or opposition to our presence in Iraq — conducting a survey of 4000 declared Democrats. This, by itself, is not sufficient, even if the pool of respondents is randomly drawn — if a disproportionate number of responses come from women versus men, upper income versus lower income, Caucasians versus African-Americans, et cetera. Even if randomly drawn, the sampling would be skewed.
This problem of consistency exists when considering whether stock price movements reflect a randomly drawn sample (especially, when the volume of exchanged shares is small). For such a sampling to be legitimate, it would be necessary to validate whether those trading shares accurately reflect the larger pool of stock owners. While age, gender, and income demographics may represent legitimate considerations, it would also be necessary for such a sampling of traded shares to be consistent with the larger pool of shorts and longs, long-term investors versus traders, insiders versus institutional holders versus the Robert Crawfords (average Joe’s), etc. As the volume become smaller, the disproportionate influence of any group becomes greater.
And this represents an undermining of the efficient market theory – if believing that the collective wisdom of buyers and sellers is a valid reflection of the stock and its intrinsic value. With more AEO shares outstanding than there are citizens in the US and with less than one-tenth of one percent of the total exchanged daily, the problem of inconsistency and non-randomness make the legitimacy of the stock’s price and its movement demonstrably problematic. It is more likely that the stock’s current price reflects the tolerance and expectations of traders (long and short) than it does a legitimate assessment of the company and its intrinsic value. If the average investor / trader were buying and selling based on the fundamental’s, the analyst reports produced by Goldman, Lehman, and Morgan wouldn’t be so popular … and my blog analysis of Ceradyne would not have received 300 visits in the last 48 hours.
For the market to be efficient, it must be intelligent and informed. I may be wrong, but most of those I encounter on other boards are not informed.
Intelligence, however, is not doubt.
