Intrade’s betting odds on the 2012 presidential election have differed significantly from those available elsewhere. For the 48 hours preceding the election, the difference in the implied probability of Obama winning on Intrade relative to other betting agencies like Betfair, was 8 to 15 percentage points. This persisted until a large share of Ohio votes had been counted and Colorado and New Mexico were starting to count, at which point the difference quickly evaporated. Over the previous 3 weeks or so, the difference had moved in the range of 5 to 10 percentage points. The
Just consider how much money is spent every year driving to voting booths just to align yourself with a certain attitude/philosophy. A number that is dwarfed by inefficient charitable contributions chosen to support causes that capture our imagination and let us look good rather than for their success.
Given these facts does it really seem so implausible that it was demographics and the lure of showing off your confidence that caused this result?
The lack of arbitrage in response could easily be explained in the same fashion. It wouldn't feel like stating your confidence your candidate couldn't lose but rushing in to take advantage of someone's bad gamble. -------Another likely possible explanation is that the market simply isn't liquid enough to provide an accurate price signal in the face of even one large corporate or individual investor wishing to buy insurance against a political outcome.
Most individuals don't have the time, psychological distance, research and hedging opportunities to make it worth their while to dump in the fairly sizable quantities of money that would be needed to move the market back and the overhead/uncertainty (esp the gambling angle) discourage professional well funded attempts at arbitrarge. However, this doesn't mean that some individuals/organizations might not have an incentive to hedge their economic bets on the election.
----
As far as Romney vs. Obama they do stand in considerably different situations with respect to influencing the market.
Romney has a much larger personal fortune, the disposition of which he has kept far more private than Obama. Also, as he was losing the risk was much less to him. Had Obama tried to influence the market it would have caused a scandle during his next term. Finally, republicans may be more likely to be moved by the indications of a market.
I'm surprised that the issue of comparing prediction markets to the best statistical analyses (based on poll results) has drawn no other real attention here. After all, isn't this really a variant of the question of the efficiency of the market versus central planning (which is, in essence, statistical prediction). Robin's ingenious procapitalist concept could prove the undoing of capitalist ideology if statisticians prove superior to the market.
It should be embarrassing to be outdone by statisticians because the statisticians' analyses were available to the traders. The markets are supposed to excel at combining information, but in, fact, they seem to have done worse due to whatever traders believed that they added to the available statistical knowledge.
There are biases known to affect the prediction markets--unlike the best statisticians. What happens to the "wisdom of the crowds" when Silver and Wang, "loners," did better?
The capability of science and technology grows faster than the complexity of markets, which they are destined to replace.
[For a communist far future, see Iain M. Banks Culture series.]
I don't think this is quite right. Romney's interest, it is quite true, was in not appearing weak. But people like to vote for winners, and being the overwhelming favorite would have helped Obama. (That's why the Democrats liked Silver and the Republicans hated him.) What gives rise to the impression that Obama wanted the race to appear close is that Obama couldn't be seen as believing he was a shoe-in, for that would appear to "take the electorate for granted" (as in the first debate).
Both candidates have an incentive for the race to be perceived as a dead heat. One doesn't want their base to take anything for granted, the other doesn't want to be perceived as weak. The media likewise has an incentive for the race to be perceived as close, to keep people watching.
Since for Obama and Romney it's more or less a zero-sum game, their incentives couldn't have possibly been aligned. Although I'm not actually sure who benefits from a particular margin...
But anyway, Wang was #1 at 0.012, Silver and Intrade were very similar at 0.044 & 0.048 respectively.
So, the statisticians tied or did better than the markets. (But what about BetFair?)
Since if we adopted your minimalist 'the only thing that matters is whether the real outcome was assigned >50%', Intrade did as 'perfectly' as many - why, it predicted Obama with >50% odds! What is this manipulation everyone is getting all hot and bothered about?!
You're distorting my comment. I never advocated "minimalism." I said it made sense to look at predicting the electoral vote because that's what Silver focused on perfecting and because he did it perfectly, whereas no one else did. To test the prediction markets, you should look at the best the statisticians can provide, and Silver focused his efforts and succeeded in predicting the electoral vote better than anyone else. Other measures, some in some ways more sensitive, can be devised pertaining to electoral votes, but whether they are "better" depends on what your formulas and intellectual effort is concentrated on.
Did Silver do better than the prediction markets generally or better than Wang generally? That's your question, not mine. (Problem is, you insist on imagining it's mine.) Silver performed perfectly on what he tried to optimize, and it's instructive to test the prediction markets against the best performance by a statistician (with respect to the one prediction where he performed best).
> Well, you trivially can't do better with respect to predicting electoral votes, which was Silver's primary objective.
Both Linzer & Wang had better Brier scores on electoral votes.
> Since what's of practical interest is the electoral votes
Since you're complaining about harping, I'll harp some more: electoral votes aren't of 'practical interest', winning the Presidency is. There too Silver assigned a lower probability than others (Wang, Jackman, & Linzer).
> So how did the prediction markets stack up against Silver )or against Wang on Senate races)?
If you had read the link... But anyway, Wang was #1 at 0.012, Silver and Intrade were very similar at 0.044 & 0.048 respectively.
> The point is the comparison between statisticians and the markets.
The whole point about looking at the full set of predictions is to make that comparison you want possible! Since if we adopted your minimalist 'the only thing that matters is whether the real outcome was assigned >50%', Intrade did as 'perfectly' as many - why, it predicted Obama with >50% odds! What is this manipulation everyone is getting all hot and bothered about?!
Well, you trivially can't do better with respect to predicting electoral votes, which was Silver's primary objective. If you want to throw other predictions into the pot, what's "better" depends on your choice of criterion of overall success.
Since what's of practical interest is the electoral votes, the question of how the prediction markets predicted the distribution (whether directly or by implication) seems the most important, at least in a direct sense. So how did the prediction markets stack up against Silver )or against Wang on Senate races)?
Your response is mostly irrelevant to my point, harping, LW style, on a single, isolated phrase ("do better"). We all know other things were predicted. The point is the comparison between statisticians and the markets.
> the statisticians seem to have carried the day--if only because it is impossible to do better than the perfect prediction of electoral votes.
Of course you can do better. First, the electoral votes is not the only prediction to make: you can predict total number, you can predict popular vote, you can assign higher odds to Obama winning than many did, you can assign higher odds to all 50 states' outcomes, you can more precisely predict vote-shares for all 50 states, you can predict all ~30 Senate races *and* again their vote-shares, and so on. Compiling predictions, I think I wound up with something like 163 unique 2012 predictions from Nate Silver alone.
When you take these full prediction sets and run the commonly accepted measures among these statisticians (Brier & RMSE), you find that far from all doing equally well, there are striking rankings, and for example, Nate Silver does not always come out on top. Indeed, Wang trounced him soundly on the Senate races because he predicted them all correctly while Silver lost *2*.
I believe that US gambling laws naturally drive Americans to InTrade (which IIRC has some special legal status which allows it not to count as 'gambling') over UK betting sites like BetFair.
If by "important" you mean important, and by "signal their intellectual or moral seriousness" you mean "not have their interests screwed over by markets gone haywire, again" then I agree.
Well, there are markets to abolish that are far more important than prediction markets to keep people's interests from getting screwed over . Start with the stock market!
I really lack a good sense of why there's resistance to allowing prediction markets, but I think the only rational basis probably applies equally in general: to Internet gambling public impulsiveness and addiction.
Then why didn't people put larger amounts in to bid away this arbitrage? The market was too shallow and illiquid for it to be worth much effort,and trouble? The arbitrage spread was very illiquid and would have collapsed quickly before you could gain that much anyway?
Just consider how much money is spent every year driving to voting booths just to align yourself with a certain attitude/philosophy. A number that is dwarfed by inefficient charitable contributions chosen to support causes that capture our imagination and let us look good rather than for their success.
Given these facts does it really seem so implausible that it was demographics and the lure of showing off your confidence that caused this result?
The lack of arbitrage in response could easily be explained in the same fashion. It wouldn't feel like stating your confidence your candidate couldn't lose but rushing in to take advantage of someone's bad gamble. -------Another likely possible explanation is that the market simply isn't liquid enough to provide an accurate price signal in the face of even one large corporate or individual investor wishing to buy insurance against a political outcome.
Most individuals don't have the time, psychological distance, research and hedging opportunities to make it worth their while to dump in the fairly sizable quantities of money that would be needed to move the market back and the overhead/uncertainty (esp the gambling angle) discourage professional well funded attempts at arbitrarge. However, this doesn't mean that some individuals/organizations might not have an incentive to hedge their economic bets on the election.
----
As far as Romney vs. Obama they do stand in considerably different situations with respect to influencing the market.
Romney has a much larger personal fortune, the disposition of which he has kept far more private than Obama. Also, as he was losing the risk was much less to him. Had Obama tried to influence the market it would have caused a scandle during his next term. Finally, republicans may be more likely to be moved by the indications of a market.
I'm surprised that the issue of comparing prediction markets to the best statistical analyses (based on poll results) has drawn no other real attention here. After all, isn't this really a variant of the question of the efficiency of the market versus central planning (which is, in essence, statistical prediction). Robin's ingenious procapitalist concept could prove the undoing of capitalist ideology if statisticians prove superior to the market.
It should be embarrassing to be outdone by statisticians because the statisticians' analyses were available to the traders. The markets are supposed to excel at combining information, but in, fact, they seem to have done worse due to whatever traders believed that they added to the available statistical knowledge.
There are biases known to affect the prediction markets--unlike the best statisticians. What happens to the "wisdom of the crowds" when Silver and Wang, "loners," did better?
The capability of science and technology grows faster than the complexity of markets, which they are destined to replace.
[For a communist far future, see Iain M. Banks Culture series.]
I don't think this is quite right. Romney's interest, it is quite true, was in not appearing weak. But people like to vote for winners, and being the overwhelming favorite would have helped Obama. (That's why the Democrats liked Silver and the Republicans hated him.) What gives rise to the impression that Obama wanted the race to appear close is that Obama couldn't be seen as believing he was a shoe-in, for that would appear to "take the electorate for granted" (as in the first debate).
Both candidates have an incentive for the race to be perceived as a dead heat. One doesn't want their base to take anything for granted, the other doesn't want to be perceived as weak. The media likewise has an incentive for the race to be perceived as close, to keep people watching.
Since for Obama and Romney it's more or less a zero-sum game, their incentives couldn't have possibly been aligned. Although I'm not actually sure who benefits from a particular margin...
But anyway, Wang was #1 at 0.012, Silver and Intrade were very similar at 0.044 & 0.048 respectively.
So, the statisticians tied or did better than the markets. (But what about BetFair?)
Since if we adopted your minimalist 'the only thing that matters is whether the real outcome was assigned >50%', Intrade did as 'perfectly' as many - why, it predicted Obama with >50% odds! What is this manipulation everyone is getting all hot and bothered about?!
You're distorting my comment. I never advocated "minimalism." I said it made sense to look at predicting the electoral vote because that's what Silver focused on perfecting and because he did it perfectly, whereas no one else did. To test the prediction markets, you should look at the best the statisticians can provide, and Silver focused his efforts and succeeded in predicting the electoral vote better than anyone else. Other measures, some in some ways more sensitive, can be devised pertaining to electoral votes, but whether they are "better" depends on what your formulas and intellectual effort is concentrated on.
Did Silver do better than the prediction markets generally or better than Wang generally? That's your question, not mine. (Problem is, you insist on imagining it's mine.) Silver performed perfectly on what he tried to optimize, and it's instructive to test the prediction markets against the best performance by a statistician (with respect to the one prediction where he performed best).
> Well, you trivially can't do better with respect to predicting electoral votes, which was Silver's primary objective.
Both Linzer & Wang had better Brier scores on electoral votes.
> Since what's of practical interest is the electoral votes
Since you're complaining about harping, I'll harp some more: electoral votes aren't of 'practical interest', winning the Presidency is. There too Silver assigned a lower probability than others (Wang, Jackman, & Linzer).
> So how did the prediction markets stack up against Silver )or against Wang on Senate races)?
If you had read the link... But anyway, Wang was #1 at 0.012, Silver and Intrade were very similar at 0.044 & 0.048 respectively.
> The point is the comparison between statisticians and the markets.
The whole point about looking at the full set of predictions is to make that comparison you want possible! Since if we adopted your minimalist 'the only thing that matters is whether the real outcome was assigned >50%', Intrade did as 'perfectly' as many - why, it predicted Obama with >50% odds! What is this manipulation everyone is getting all hot and bothered about?!
Of course you can do better.
Well, you trivially can't do better with respect to predicting electoral votes, which was Silver's primary objective. If you want to throw other predictions into the pot, what's "better" depends on your choice of criterion of overall success.
Since what's of practical interest is the electoral votes, the question of how the prediction markets predicted the distribution (whether directly or by implication) seems the most important, at least in a direct sense. So how did the prediction markets stack up against Silver )or against Wang on Senate races)?
Your response is mostly irrelevant to my point, harping, LW style, on a single, isolated phrase ("do better"). We all know other things were predicted. The point is the comparison between statisticians and the markets.
> the statisticians seem to have carried the day--if only because it is impossible to do better than the perfect prediction of electoral votes.
Of course you can do better. First, the electoral votes is not the only prediction to make: you can predict total number, you can predict popular vote, you can assign higher odds to Obama winning than many did, you can assign higher odds to all 50 states' outcomes, you can more precisely predict vote-shares for all 50 states, you can predict all ~30 Senate races *and* again their vote-shares, and so on. Compiling predictions, I think I wound up with something like 163 unique 2012 predictions from Nate Silver alone.
When you take these full prediction sets and run the commonly accepted measures among these statisticians (Brier & RMSE), you find that far from all doing equally well, there are striking rankings, and for example, Nate Silver does not always come out on top. Indeed, Wang trounced him soundly on the Senate races because he predicted them all correctly while Silver lost *2*.
See https://docs.google.com/doc... and http://appliedrationality.o...
"If I were a Democrat supporter"
How is that relevant? Isn't making money improving Intrade's perfomance good regardless of your political position?
Or maybe Democrats are more subject to overconfidence.
The question isn't why the price was so low. It's why it differed from other makets.
I believe that US gambling laws naturally drive Americans to InTrade (which IIRC has some special legal status which allows it not to count as 'gambling') over UK betting sites like BetFair.
"What might discourage prediction-markets' proponents is the possibility that statisticians perform better than the market"
Unless, of course, that proponent happens to be a statistician :)
If by "important" you mean important, and by "signal their intellectual or moral seriousness" you mean "not have their interests screwed over by markets gone haywire, again" then I agree.
Well, there are markets to abolish that are far more important than prediction markets to keep people's interests from getting screwed over . Start with the stock market!
I really lack a good sense of why there's resistance to allowing prediction markets, but I think the only rational basis probably applies equally in general: to Internet gambling public impulsiveness and addiction.
Then why didn't people put larger amounts in to bid away this arbitrage? The market was too shallow and illiquid for it to be worth much effort,and trouble? The arbitrage spread was very illiquid and would have collapsed quickly before you could gain that much anyway?
You can put any amount in by wire transfer, regardless of where you live, but it takes a few days.