Momentum – an unusual type of factor exposure

The phrase “smart beta” may be fairly new, but the idea of factor exposures has been around for decades. Not all factors are the same, however. Take momentum, for example.

Can momentum really be a good thing?

On the surface, buying a stock because the price has gone up is an odd thing to do (it would obviously have been a better buy at a lower price.) And momentum is a technical factor, not directly linked to the underlying business in which it is investing. What’s more, momentum creates a positive feedback loop in the stock price: a rising price attracts more buyers, which can push the price up further. So it tends to add to market volatility, and may play a part in the development of bubbles and crashes.

With all that against it, why is momentum a popular factor to invest in? The answer, of course, is that it’s worked. The historical pattern is strong enough to lead to a formal Russell Investments position that “the premium to medium–term high momentum stocks is robust and exists across all regions.”1 Even allowing for the higher trading costs typically associated with momentum investing, it’s been a value-adding approach.

One of a range of possible explanations for the factor’s success is that if enough investors believe in a momentum effect, then the belief could become self–fulfilling as a result of the positive feedback loop created. But does that logic really hold up?

An artificial stock market

To answer that question, we can look to a neat model created by Brian Arthur and some colleagues at the Santa Fe Institute. This model is described in a 1996 paper Asset Pricing under Endogenous Expectations in an Artificial Stock Market.1 The artificial stock market to which the title refers consists of individual agents who create models or theories of the market, test those theories against actual behavior, and adopt the ones which perform the best. As the market evolves, each participant continues to test new theories or, as Arthur puts it: “agents’ expectations co–evolve in a world they co–create.”

The researchers found that if they made participants in this artificial stock market slow to adapt their forecasts to the market’s behavior, then the market converges to classical economics’ rational market: participants develop similar models to one another, trading volumes are low, there are no bubbles or crashes, and technical (momentum) trading does not emerge. If, however, “we allow the traders to adapt to new market observations at a more realistic rate, heterogeneous beliefs persist, and the market self–organizes into a complex regime. A rich ‘market psychology’—a rich set of expectations—becomes observable. Technical trading emerges as a profitable activity, and temporary bubbles and crashes occur from time to time.”

Other features of real-life markets – such as fat tails, volatility regimes and persistence in trading volume—also emerge in the model in this case.

The artificial stock market used in the paper is a fairly simple representation of investor behavior, but the results do seem to confirm that the logic behind a momentum effect may hold, as well as providing support for the wider “complex system” view of markets.

1Russell Investments Beliefs: Equity momentum as a return source. Internal research document, April 2014.
2Republished as Chapter 3 of Complexity and the Economy (2014). Oxford University Press.