Paul Tetlock finds that more liquidity at Tradesports leads to greater inefficiencies

The abstract:

I show that the most liquid securities markets exhibit significant pricing anomalies, such as overpricing low probability events and underpricing high probability events, whereas less liquid markets do not exhibit these anomalies. I also find that the prices of illiquid securities converge more quickly toward their terminal cash flows. These results are consistent with the idea that liquidity is a proxy for non-informational or noise trading, which can impede market inefficiency; but they are inconsistent with models in which increases in liquidity have no impact or a favorable impact on efficiency.

His paper is here (via Justin Wolfers). This finding makes sense to me, given my differentiation between gamblers and traders, very early on in the life of this blog:

But there are more than gamblers and speculators at TS. There are traders. And what separates the traders from the gamblers is the driver to trade--namely, information (over impulse). Now there certainly is noise--in any market--as well as information, so that leads to negative expected profits. But there is also informational advantage. And data modeling.

It's not just meat heads betting sports non-informationally that tend to overshoot. I seem to have observed in equity analysts and debt rating agenices, too!

With increased awareness and acceptance of prediction markets, coupled with decreased transaction costs and regulatory barriers, more informational traders should come (and arb out the gamblers and speculators).