From RISKS 26.19, Financial market automated amplification of trades:
"The two men worked out how the computerized system would react to certain trading patterns—allowing them to influence the price of low-volume stocks." Although the article gives no indication of how they did it, the day traders “gave false and misleading signals about supply, demand and prices'', which caused the robots to take action—which the day traders then took advantage of.
Yeah, I did that once.
When I was in grad school, I often took part in economics experiments. The behavioral econ folks would put out a call for volunteers, and you’d go to their computer lab, play a game with (against) other volunteers for an hour or so, and leave with cash in hand, based on your performance in the experiment. Usually at least $5 for showing up, to $50 or more for a good performance in the game.
Most of the time the games went something like this: You would play several rounds. At the beginning of each round, you would be given some amount of certain commodities and/or “cash” on the computer. You would be told that at the end of the round, each of the various commodities would have a certain “cash” value to you. During the round, you could trade commodities with other participants by submitting a bid or ask price to the market’s book, and the market would execute transactions by matching the offers. At the end of the round, your commodities were converted to “cash” at the rate previously disclosed to you, and that “cash” would go into your pot of real cash to be paid out at the end of the session.
In one game, we started each round with a certain amount of commodity A and a different amount of commodity B, which changed from round to round. We were told the value of A and B, which also changed from round to round. After a few rounds, a pattern became clear: We would each start with some inventory of both A and B, but for each person, only one of the two was valuable. You might start with 10 A and 20 B, but at the end of the round, your A’s would convert to 15 cents each, but the B’s would be worth nothing to you. During the round, you would try to trade your B’s to someone else, for whom B’s were valuable but A’s were valueless. It also seemed that everyone who valued A’s valued them identically, and everyone who valued B’s valued them identically at some different value.
The tricky part was that if you valued A’s, you had no idea what B’s were worth to people who valued B’s, and it changed from round to round. Your only insight came from watching the order book and the executed trades. If people were paying a lot of cash or a lot of A’s for each B, you could conclude that they were getting a lot of real money for B’s in this round. Each round, the orders started out pretty randomly, as participants tried to feel each other out. Toward the end of the round, the A/B exchange rate would converge, presumably to something close to the fair market value.
I realized that in the initial confusion of each round, the first orders placed in the book could be a strong influence on the participants, as everyone was trying to figure out the valuations. So I decided to manipulate the market. Let’s say my initial configuration for the round was something like this:
|Commodity||Initial quantity||Value at end of round|
My goal, clearly, was to sell off my B for the best price I could get. So what did I do? As soon as the round started, I entered the following orders as quickly as I could:
- Sell 1 A for $0.05
- Buy 1 B for $1.00
These offers were the first, and briefly only, information that anyone had about the value of A and B. Some people would bite, and make compatible offers that would clear the market, and everyone would see that those were “the” current market values for A and B. As soon as the prices gathered some traction, I sold off all my B at high prices, and bought as much cheap A as I could.
I was able to do this for a couple of rounds before the experiment ended. I think I walked off with $80 or $100 for an hour of (fun!) work.
I have a feeling that the RISKS item involves something similar. In a thinly-traded market, you can send a false price signal pretty cheaply.