inTrade Manipulation

Wolfers points to some evidence of manipulation at inTrade. It appears that inTrade is a low liquidity betting market, and during some of the periods of calm, it's more vulnerable to manipulation since large trades can go unnoticed and may take some time to readjust, which will only happen as others notice the arbitrage opportunities and take advantage of it. That probably explains why a month ago, I couldn't reconcile what inTrade was showing in the main contract versus the state contracts. The state contracts showed Obama needing only to win one of the five swing states in order to win. Or to put it differently, McCain needing to win all five of them to hold on. And these states were, at that time, Virginia, New Hampshire, Colorado, Nevada and Ohio. He couldn't lose a single one of them. So to win, it was going to hinge on the joint probabilities of winning all five, which given that they were so close to begin with, should've translated into a very clear advantage of Obama winning, that should've translated into a much higher contract price at the national level.

Which means, I should've just bought the national contract on Obama at that point, since it was so clearly undervalued based on the state contracts. Wolfers says basically that was probably right - the state contracts never showed the same kind of movements and volatility as those national contracts did, nor did other betting markets either in the US or in the UK. So, what does it mean? I think one thing it means is that because inTrade is so illiquid, there are more arbitrage opportunities there exist in more liquid markets. The more liquid the market, the more efficient it is, the more the current stock price reflects all available information and thus the less likely it is that you actually possess information than isn't already factored into the price. This is called the efficient market hypothesis for those not familiar with it. But, it sounds to me like inTrade is not efficient, which means why it may be less reliable for accurate predictions all the time (except Wolfers says it probably is still accurate in the longterm, but not always at every moment. Those weird anomalies do eventually get ironed out after all) but could be a source of easy money for the better.