Tuesday, September 14, 2010

Using Robot Traders to Stabilize Markets

Larry MacDonald reported on interesting research into using “robot traders” to prevent market bubbles. Unfortunately, this research (paper available here) by Suhadolnik, Galimberti, and Da Silva fails to address some serious real-world constraints.

The idea is that based on the researchers’ model of trading patterns, adding some contrarian robot traders that trade against the herd tends to stabilize stock markets and prevent bubbles. This sounds plausible, but it brings up serious questions about how this idea could be implemented.

To their credit, the researchers mentioned one of the practical problems (although it is buried on the third to last page):

Where will the money for robot trading come from?

“Financial resources involved in stabilizing the stock market is not addressed.” I don’t know if contrarian robots would make or lose money over the long run, but contrarian strategies can definitely lose money for very long periods of time. I definitely don’t trust the government to use tax money to trade stocks with robots.

The researchers did their analysis assuming that the robots would represent “20 percent of the total market turnover”. This would require a massive amount of money. The researchers describe this as a “limitation” and go on to say the following.

“We believe the current model could still be slightly changed so as to remove this limitation by considering an extra constraint on robot trades, which will make them dynamically sustainable; but this should be clearly addressed by future research.”

This vague promise seems little more than a plea for more funding. There is another problem that the researchers didn't mention:

The Law of Unintended Consequences

If government-controlled robot traders ever become a reality, the stock-trading game would change completely. Traders would no longer concern themselves with taking money from each other; they would focus on taking money from the government robots. How comfortable would you feel with the government robots playing the stock market with billions in tax dollars with traders all over the world trying to beat the robots?

This is an interesting line of research, but I have little hope that it will ever become practice.


  1. Could a robot trader still where a suit and tie? Sorry, i know it is immature, but i have a hilarious image in my head of robots working on the trading floor.

  2. @Jan: Good one :-) Now I'm thinking of a robot in a tie scribbling trade details onto scraps of paper.

    I agree that "robot" sounds funny in this context. I would have used "computer trading", but I decided to go with the terminology from the paper.

  3. Good point, that was the first thing that popped into my mind as well!

    When we're talking about governments stepping in to prevent bubbles, they have options available other than robot traders: taxes, regulations (e.g., setting/adjusting limits to the amount of leverage that can be used for things as they get into bubble territory), etc.

    Of course, when we're talking about governments stepping in to prevent bubbles, they have a fairly terrible track record of it (US reducing capital gains taxes just as the tech bubble was bubbling up; mortgage standards and the housing bubble).

  4. @Potato: On the one hand, I'm leery of the government's competence in meddling in the affairs of the stock market. On the other, unrestrained activity would lead to disaster. This is not an easy problem to solve, but I'm pretty sure the answer is some sort of clever regulation, not risking billions or even trillions of dollars.