I am struggling to understand the following statement I found in a book about algorithmic trading:

Unfortunately, we cannot trade on the mean reversion of returns

If it is a proven fact, then what are the reasons? If it's only based on the authors opinion, why might he have come to that conclusion?

  • 1
    It's simply the author's opinion, nothing more, nothing less.
    – misantroop
    May 7 '17 at 6:30
  • This statement appears in chapter 2 of Algorithmic Trading by Ernest P. Chan.
    – Flux
    May 1 at 0:45

More information about the context would be helpful. I can guess a few things the author might be meaning (these are common arguments, not necessarily my views):

  1. Apparent mean reversion is an artifact of the microstructure of the markets. For example, trades executing alternately on the bid and then ask. It looks like it goes back and forth, but if you try and buy you will execute at the ask. If you try and sell, you will execute at the bid.

  2. Returns do not mean-revert predictably enough for the gains to doing so to outweigh the trading costs

  3. A strategy designed to exploit mean-reversion has a great deal of risk (because sometimes mean reversion doesn't happen...in a big way, for example). Therefore the gains you make on such a strategy are not good in a risk-adjusted sense.

  4. At least one side of the transaction that would make money from a mean-reversion strategy is too illiquid to trade in a cost-effective manner. Many pricing anomalies are only present in illiquid stocks. Such a strategy might involve short selling of securities that can't be easily short sold, for example.

  5. Mean reversion doesn't happen over the right time horizon for the trader. Most proposed mean reversion strategies I have seen involve holding the asset for much longer than an algorithmic trader will normally do.

The question of whether mean reversion can be traded on is neither proven nor disproven. There is good empirical evidence on both sides of the question. certainly there are funds and individuals who do trade on mean reversion. The question is whether they happen to have been lucky and whether the trends the use will continue (or more generally, whether someone else attempting a similar strategy has a better chance of making money than a monkey throwing darts).

Note that a book on algorithmic trading is not your best bet for a reliable analysis of whether there is mean reversion in the market. The algorithmic trading world is always rapidly changing. Strategies make money for a few months and then they don't. The features algorithmic traders use to make money are not long-lived nor statistically reliable. This draws participants most comfortable in that environment, rather than those who only believe in patterns that are statistically and economically meaningful over long horizons.

  • True: More context would be helpful. Thats why I needed to ask you folks :-) May 6 '17 at 17:56
  • 2
    3 is the big one. You can "reliably" make hundreds of dollars using mean reversion for months and then suddenly, one day, you have a huge loss because of Trump getting elected, or Brexit, or Theresa May announcing unexpected elections, or a million other things. It's easy to think you're an investing genius for a few months if you construct the equivalent of a reverse lottery where you take very large risks to get fairly reliable small gains. May 8 '17 at 8:32

This seems to explain it: we can't do it because it won't be profitable reliably.

Screenshot of text

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.