Manipulating Prediction Markets

A reader (Alex) followed up on a recent post on prediction markets with the following interesting question:

Does this have an impact on accuracy of betting markets?

It’s interesting is that "market" corrected itself relatively quickly, but lower volume market (Iowa) could be more susceptible to someone willing to take large (short-term) loss.

It's true that lower volume markets like the Iowa Electronic Markets are more susceptible to manipulation, because of a concept we finance nerds call "depth". This is a measure of how much a trade of a given amount moves prices. Lower volume markets have less depth, so it takes a smaller trade to move prices.

Manipulators by definition make betting markets less predictive. The reason that these markets are normally good at aggregating information about beliefs is that profit-maximizing investors have incentives to trade on their beliefs. For a contract that pays off $1 if an event occurs (like, for instance the “Bush gets elected” contract), it’s pretty easy to show mathematically that a rational individual would be willing to pay an amount equal to the probability that the event will occur. In other words, if you believed that Bush had a 45% chance of winning, you’d be willing to pay $0.45 for the contract that pays off a dollar if Bush wins.

If market players are motivated by making profits, and the price of the contract is below their assessed “fair” price (in the above example, assume that it was $0.40), they’d be willing to buy the contract. As they buy the contract, the price goes up. It will go up until there is no more perceived mispricing (i.e. until the price is at $0.45).

However, a person might be willing to trade at prices they don’t think are accurate. The article gives an example of a “speculative attack”, where someone puts in an order at an obviously low price in hopes of causing a selling panic. If successful, they will wait until the other sellers drive the price down and then purchase shares on the cheap. Their hope is that they can make more money on the cheaply bought shares than they lost on the unrealistic earlier “triggering” trades.

Another possibility (also mentioned in the article) is that someone wanted to create the perception that Bush was unlikely to win. So, they sold the contract to drive the price down in the hopes that people who watch this market start saying “look at what people are saying with their money – Bush has no chance of winning”. Since prophesies are often self-fulfilling in the world of politics, it could conceivably started a negative buzz that helped torpedo perceptions about Bush’s chances.

In a market like the Iowa Electronic Markets, a relatively small trade can move prices quickly. Likewise, an individual could give money to other operatives (the limit to the amount than can be played on the Iowa Market is $500) and also have them trade to magnify the effect.

Unfortunately, unless someone was willing to keep selling (note: they can do this even if they themselves don’t own the contract by executing what’s called a short sale), the price would quickly rebound, since other investors would perceive the contract as being underpriced. Therefore, the other participants would start buying, until the market is back in equilibrium (i.e. the price is back to “fair levels).

So, speculators or manipulators can make the contract temporarily mispriced, but markets quickly correct. The one case where this might fail is if an individual was willing to keep throwing money away for non-profit making reasons. I’m sure, for instance, that Soros probably has enough spare change in his couch to corner the whole Iowa Electronic Market, so it’s definitely possible. As to how likely, I’ll leave that up to the conspiracy theorists. Besides, I’m running late – I have a black helicopter to catch.