I too would like to see prediction markets flourish, for reasons like
reigning in pundits, and to see if there's anything to Robin Hanson's
fascinating proposal.
And I too would like to see something more in the style of Intrade. MPOE-PR, I read
your link above, and clicked through to the link
about mutual betting. And I am open to the idea that I'm wrong - though I'm familiar with basics of game theory, expected value, and the like, I am not an expert.
But so far I still don't see why the mutual approach is supposed to be better. In fact, thinking about it, it still seems to me considerably
worse than the Intrade approach.
You say in your link explaining mutual betting, for example, that the Intrade style will take a house cut. But of course this is also true of mutual betting systems. And you say that weighted mutual betting systems allows for hedging. But of course this is also true of the Intrade approach.
Meanwhile, the time-weighting system seems to have two major disadvantages. The first problem is minor conceptually, but I think a big one PR-wise: it makes probabilities and potential returns relatively opaque, as your stubborn "upseller" demonstrated in your post.
The second problem is much more systematic: information relevant to the bet does not flow linearly - it makes big occasional leaps. But the weighting of time of bet placements does move linearly. So it does not track advantages of information accurately.
For example, I recently re-watched the classic old movie
Trading Places. In it, there's an all-important secret "crop report" to be released on a specified date by the US Department of Agriculture - a report very relevant to the price of orange futures. Let's suppose this is a binary bet ("crops to be above X bushels") and on both a weighted-mutual site and on the intrade-like site this binary bet is trading, on the day before the announcement, at an implied probability of .2 (ie 20% chance of being true, or 4-1 odds).
Suppose the day before the report my own informed estimate puts the truth at more like .4 probability. I have 20 BTC to play with, say. I bet 10 BTC on the weighted-mutual site and 10 BTC on the intrade-like site. And to keep all other factors constant, let's say those purchases are enough to push the implied odds up to .25 on both sites.
Now it is the next day, and the crop report is released. Suppose the crop report makes it very likely for the binary bet to be true. Betting on both sites leap to an implied probability of .8. On the weighted-mutual site, I am now looking at 4-5 payout, plus the very slight advantage of a one-day extra weighting. I am punished for trying to do price discovery before information comes in. On the Intrade site, if the proposition settles at true, I will benefit
much more; even better, I can sell all shares now to lock in a large profit. My risk of betting before more information is released has been rewarded, the way it should be.
- To make it more concrete, say it's the first day the bets open, and there's no house take on either site. On the weighted-mutual site I see the bets at 30-120, so my 10 BTC pushes it to 40-120. And on the intrade-like site, I buy 40 shares (worth 1 BTC each if true, and 0 if false) at .25, for 10 BTC. Suppose after the end of the next day - the crop-report day - no further bets or information come in until the proposition settles as true. Then the weighted-mutual ends at 480-120, so I get my 10 back, plus 2.5 BTC (120/480*10), plus some tiny margin for weight. Meanwhile at the intrade-like site I get my 10 back, plus 30 BTC. (Or I lock in my profit early, and sell my 40 shares at .8, for a 32 BTC gross return.)
Show Posts
