Recent news stories about insider trading and the conviction of a hedge fund manager reminded me of a simulation we used to run that is just as relevant today. This simulation is different from the ethics simulation I wrote about elsewhere in the blog. The ethics simulation focuses on the individual decision: when faced with the situation, will you make the ethical decision? The insider trading simulation focuses on groupssharing insider information for mutual profit.
In the simulation, traders receive insider information about the prospects of a company. Prior to trading, they are allowed to exchange messages; they can choose whom to communicate with, and if they receive a communication, they can choose to see it or not. They know who is sending them the communication. After the communication, the stocks are traded in the FTS Interactive Markets; in these markets, the trader buy and sell the stocks from each other, and in the particular simulation, the trading mechanism was a double auction where everyone is a dealer, so they trade for their own account. There were cash prizes for the best performers, and all the information they received from the system was correct (and so unambiguously useful).
We wanted to explore two issues. The first was whether insider trading mattered: did the market behave significantly differently from one without insider trading? Second, how hard is it to detect such trading? To focus on the latter, we decided to give the tick-by-tick trading to "regulators" who did not participate in the simulation, and asked them to analyze the data. If they could "convict" a trader, the regulators would get the cash prize of the trader. We decided not to give the regulators the content of any message (so no “wiretaps”) but did tell them whether any two parties had communicated (so, for example, phone records).
Ironically, what actually happened illustrates the difference between theory and practice. In terms of our original focus, we found that the value of information was short lived, that information was reflected in prices much more quickly with insider trading, and that trying to detect patterns of insider trading purely from the data was difficult.
But there was a lot more. First, we found that there were various rings that formed; people could be part of multiple rings, and they were careful not to pass information from one ring to another. Second, false information was passed. Third, there was a lot more communication of information than we expected; we even found that some groups had created their own codes. Fourth, we found evidence of coordination in trading, for example, gradually bidding up a stock, even taking some initial losses. Finally, we were surprised by the amount of communication that took place; apparently the traders were pretty confident that the likelihood of getting caught was low!
Even more fascinating was the behavior of the regulators. They quickly discovered that discovering insider trading from the trading data only was difficult at best. So they adopted a different strategy. First, they went after the big winners. Second, the way they obtained convictions was by getting others to testify and using the data to provide documentary evidence of both the coordinated patterns of trade and the gain realized. What helped them was that even within a ring, some traders made a lot more money than others; matching the pattern of communication to the pattern of winnings turned out to be the key.
The simulation also laid the basis for putting in context a wealth of discussion about insider trading: the need for laws, market surveillance issues, relationship to market efficiency, and so on. If you want more information, please contact us; we will be happy to show you how to run a similar exercise.