Why Noise Traders Do Not Move Markets

Analysis of “Investor Sentiment, Beta, and the Cost of Equity Capital” In the study by Antoniou, Doukas and Subrahmanyam (2015), the researchers look at CAPM, beta, noise traders, periods of optimism and periods of pessimism to test a number of hypotheses: first, that optimistic times lure noise traders into buying high beta stocks, which causes high beta stocks to be overpriced; Noise traders are defined as “young, single males, who lose the most from investing, and are the most susceptible to overconfidence” (Antoniou et al., 2015, p. 352).

This definition is not explicitly given but is rather stated in terms of most unsophisticated traders being characterized by these qualities. Considering that about 90% of the market consists of insititutional investors—not retail buyers and certainly not noise investors—the ability of a small percentage of unsophisticated investors to drive price action in high beta stocks for a sustained period of time should be viewed as dubious at best.

Volume alone should be an indication of how much interest is occurring in high beta stocks—and if high beta stocks are being pushed upwards in price during optimistic periods on low volume, there is likely to be an opposite reaction among sophisticated short sellers, which will inevitably drive the price down if support levels are not bolstered by buyers other than noise traders.

If high beta stocks maintain high prices for a sustainable period (even if that period is optimistic—a problematic term in and of itself since the entire market has been “optimistic” since the first round of QE and only now with quantitative tightening seeming like a sure thing does a hint of pessimism begin to be surfacing) there is very likely to be a good portion of institutional investors also buying.

So the hypothesis that noise investors are responsible for inflating high beta stocks over a sustained period of optimism is likely inaccurate at best. The second hypothesis is that during pessimistic periods, noise traders flee high beta stocks because sell side analysts are less euphoric (“optimistic”) in their analysis, which causes high beta stocks to be positively priced.

The problem with this hypothesis, as with the first one, is that perceptions of overpricing and underpricing are completely subject and depend entirely on one’s perspective, which is only proven correct or incorrect over time as the stock prices rises or falls in response to market conditions, news catalysts, and myriad other factors.

To suggest that noise investors along are responsible for price action in high beta stocks is to ignore the realities of market conditions and wider economic and financial conditions in the developed world, where institutional investors and fund investors have been forced into adopting a risk-on strategy to obtain the desired yield, as interest rates have been kept near zero since the crisis of 2007-2009.

Now that the Fed Funds Rate is rising, one can see the anxiety of the market, with volatility rising and high beta stocks seeing investors get weak hands.

A better factor to study would have been to measure high beta stock price action by movement in the Fed Funds Rate—but the researchers’ interest is in noise investors and whether they are the reason high beta stocks have lower returns in optimistic periods than in pessimistic periods (the noise investors cause the stocks to be overbought in optimistic periods).

The conclusion should be foregone—but it is not, so it is helpful to observe other forces in the market that could account for the overbought status of high beta stocks in optimistic times. First there is the rise of algorithmic trading that dominates the market today. It is estimated that 90% of volume in equities market is due to algorithmic trading. That is an astonishing figure but it corresponds with the percentage of trading done by institutional investors.

So these figures alone indicate that if any stock is overbought one has to look at who or what is doing the buying and noise investors are simply not that powerful or influential of a force in today’s markets to account for that status. If anything, they are momentum traders—but then so too are many quant fund managers. Algorithmic trading has ushered in the era of buying the dip (but that is in large part thanks to a market supported by QE).

It remains to be seen whether the buy-the-dip approach will be sustained with QT.

Nonetheless, with so many trades being done by algorithms that could easily be characterized as noise investors too since they are programmed to look at data (headlines, non-sequiturs, and just about anything that can be imagined) before making a decision on whether to buy or sell: their only difference is that they can do so much faster and with much more dry powder at their disposal than so-called unsophisticated traders who do not typically have nearly as much capital at their disposal (which is why they flee during downturns).

Moreover, one could easily show that the market has been overbought since the end of QE—but that has not stopped major corporations from conducting share buybacks to keep the price of stocks jacked up so that corporate leaders can benefit from selling to bag holders. As smart money is shown to be exiting the equities markets, retail investors may indeed be the bag holders if the market correction does turn into a bear market.

However, they are also likely to be the ones buying the dip as noise continues to be pumped out that stocks will bounce back—and anyone paying attention over the last decade is likely to think as much. Antoniou et al. (2015) make a number of unsupported assumptions that do not serve the study well—for instance, they assume that “analyst bias is likely to be stronger in situations when uncertainty is higher” (p. 356).

But why is this likely to be the case? They assume it would be so because doing so assists them in their hypothesis—that sell side analysis is what drives noise investors to buy overpriced high beta stocks and that pessimistic times chase them away. But not all sell side analysts are going to have pessimistic tones during downturns and not all noise investors are going to listen to sell side analysis anyway.

There are far too many factors that go into the decision making of investors, sophisticated or unsophisticated to control for them—and Antoniou et al. (2015) simply do not control for them at all.

In fact, that is the main problem with the study: they conduct a test that allows them to interpret findings in support of their hypothesis, but their test should not be considered valid because it cannot be shown to measure what it intends to measure: variables are not controlled for, so the study is extremely limited in terms of what it can actually show.

The researchers do admit that “although we cannot rule out the possibility that our sentiment measure captures variations in a macroeconomic state variable, and that beta and its pricing covary with this variable, such an explanation should also accord with negative beta pricing during optimistic periods, which is challenging” (Antoniou et al., 2015, p. 366). Indeed, challenging is an apt term and indicates that the more is happening with this model that meets the eye.

Overall, the various tests performed to build out the robustness of the study do not help to solidify the study’s validity. For example, for Table 7, the researchers note that “the table presents order imbalance (SOIB) for small investors calculated from TAQ data for the low and high beta portfolios. For this test we use NYSE and AMEX stocks for the period 1980-2010” (p. 38). Note the years—essentially three decades prior to the great market intervention of the Fed known as unconventional monetary policy.

Also, the data comes before the era of algorithmic trading really comes into.