pageok
pageok
pageok
[Michael Abramowicz, guest-blogging, January 30, 2008 at 9:01am] Trackbacks
Manipulation of Prediction Markets:

The most important objection to governmental use of prediction markets is the danger that third parties might manipulate them. If officials deciding whether to expand a highway use a prediction market to forecast traffic in 20 years, road builders might be willing to lose money on the market if in doing so they can change the forecast and influence the public policy debate.

A partial answer is that the stakes of prediction markets (even the heavily subsidized ones I fantasize about) are sufficiently low that transparent attempts at manipulation are unlikely to have much effect. If George Soros announced that he would be willing to risk up to $1 billion to prop up the share of the Democratic candidate on Intrade, hedge funds would gladly take the other side of unjustifiable offers. Maybe arbitrage can't pop a widely recognized stock market or housing bubble, but arbitrage should succeed in individual prediction markets.

The bigger problem is the possibility of hidden attempts at manipulation. If X is bidding up the traffic forecast contract, this may reflect a genuine subjective probability estimate. If so, everyone else should rationally adjust their estimates in the direction of X's trading, especially since X appears confident. Traders will assign some weight to this possibility, and so will not try to move prices all the way back. If X really is manipulating, X will be at least partly successful.

Note, though, that the reason for X's success is that disinterested traders find trades generally to be informative. If I am playing poker, and think that another player has a tell, I might rationally take this into account. Sometimes, the tell is a fake, but I'm better off looking for tells and taking them into account than wearing a blindfold.

Similarly, given a choice of restricting a prediction market to trusted non-manipulators(e.g., government officials), and leaving it open to all, the open market will tend to produce better information, even though manipulation will sometimes be successful. We can improve performance by identifying traders, especially if some earn reputations for accuracy over time.

Nonetheless, if you're unconvinced, or if you think that manipulation might undermine confidence in government, that's no reason to abandon prediction markets altogether. Instead, one can still use them with a small group of trusted players (whether with real or play money). This is still likely to be better than letting just one of these people make a forecast or averaging all of these officials' forecasts.

For a more complete discussion of this issue in Predictocracy, see here. Also, see this article on a model and this article on an experiment showing that manipulators can increase price accuracy by providing extra market liquidity.

anonymous coward:
"Arbitrage" does not mean "a trade with favorable odds," at it seems to be used above.
1.30.2008 9:49am
Gerg:
The pokur analogy is inapt. A better analogy would be that while it's sometimes worth bluffing in pokur in reality most bets are legitimate. That is, most of the time you bet more when you have a strong hand and less or not at all with a weaker hand. Bluffing is something which only works when done in moderation -- and by moderation the actual frequency is quite low, deceptive tv and movie coverage to the contrary notwithstanding.

The analogy is a bit weak since in pokur you only stand to win money through your bets. So the only reason to bluff is if you think your opponent will believe you now and fold or perhaps be deceived by it in future betting.

There is no analogous situation to what you describe with roads where there may be enough profit in building the road that placing losing bets can be profitable even if you make a net loss on your bets.

I think the answer to that is clear though. The scale of the bets needs to be based on the stakes which are the subject of the bet. Just as the price of futures contracts is partly dictated by the value of the underlying securities. A market has to be liquid enough that a bet small enough to be outweighed by profits on the underlying event cannot credibly move the market.
1.30.2008 10:43am
Gerg:
Incidentally it's annoying that you post an entry about pokur but then have a spam filter which blocks comments which use the word when spelled correctly
1.30.2008 10:43am
Michael Abramowicz (mail):
anonymous coward -- I agree on the standard definition, but I'm using the term in the same way as this article, which I discuss in the linked to section.


Gerg -- Thanks for the point. Bluffing might have been a more straightforward analogy -- a "pokur" player surely takes into account an opponent's bets in assessing the probability of victory, even though knowing that sometimes these will mislead.
1.30.2008 10:52am
Mr. Liberal:

If I am playing poker, and think that another player has a tell, I might rationally take this into account. Sometimes, the tell is a fake, but I'm better off looking for tells and taking them into account than wearing a blindfold.


You might be better on average. But you might be very much worse off. And that is only in Poker.

If a bluff ensures that you lose a very large pot in poker, in that particular case, you are worse off for taking into account the signals of others.

We might consider an election to be entirely different than a game of p*ker. After all, if you lose a couple of hundred dollars this weekend, you can always come back next weekend. Elections, especially major elections, unlike p*ker games, are not held every week. They are unique events and the consequences of successful manipulation would be very high indeed.


Nonetheless, if you're unconvinced, or if you think that manipulation might undermine confidence in government, that's no reason to abandon prediction markets altogether.


This, I believe, is just hype. The idea that prediction markets are going to determine the outcomes of elections gives them far too much importance.
1.30.2008 10:53am
Sean O'Hara (mail) (www):
The use of arbitrage here makes sense, it just isn't entirely clear. Normally arbitrage is a discrepancy in pricing between markets, such that a trader could make an easy profit by buying a stock in New York and immediately selling in London. With futures markets, arbitrage can also be a discrepancy between pricing and reality -- say a bunch of Paulnuts bid his stock up to $.80, then anyone with sense could make a profit by shorting the stock and waiting -- either the price will drop, or Paul will drop out/lose.
1.30.2008 11:26am
MDJD2B (mail):
The areas in which manipulation might be a serious possibilities are those in which the amont of money invested in the predictions market is much smaller than the amount of money that rides on the outcome.
1.30.2008 11:29am
J. F. Thomas (mail):
If officials deciding whether to expand a highway use a prediction market to forecast traffic in 20 years

Now this idea really disturbs me? Bet on what traffic patterns will be like in 20 years and use that to design roads. The market is going to be superior to the professional judgment of experts? How does this possibly make sense? They should have put the fasteners used in the big dig in Boston into the prediction market or the structural stability of the I-35 bridge in Minneapolis. I bet disaster could have been averted. After all actual engineering testing and education is so expensive and time consuming.

Next you'll tell me that rather than clinical trials of drugs we should just have prediction markets to decide whether they are safe and effective. Merck can create a new drug, say it cures acne, and a prediction market can decide whether it is safe and effective. That way we can get rid of the FDA and all those expensive trials.
1.30.2008 11:50am
J. F. Thomas (mail):
Is there a prediction market for when the next hurricane will strike and where. We can get rid of NOAA and all those expensive weather satellites too.
1.30.2008 11:51am
Steven Lubet (mail):

I'm better off looking for tells and taking them into account than wearing a blindfold.


That is true only if you are good at recognizing tells. If you aren't good at it, then you are indeed better off wearing a blindfold

The same is true about bluffing. If you aren't good at recognizing bluffs, then you are better off calculating the pot odds and betting for value.

In reality, of course, most p*ker players overestimate their abilities to predict other players' actions, that being the key difference between good and bad players.
1.30.2008 12:00pm
Curt Fischer:

If a bluff ensures that you lose a very large pot in poker, in that particular case, you are worse off for taking into account the signals of others.


The pokêr analogies are rampant here in this thread, but it doesn't seem like many commenters are very familiar with the game. For example, a bluff could never ensure that you lose a large pot. Only a fold could do that.
1.30.2008 1:41pm
Wayne Jarvis:

Now this idea really disturbs me? Bet on what traffic patterns will be like in 20 years and use that to design roads. The market is going to be superior to the professional judgment of experts?


Absolutely. The data supporting this point is pretty convincing. The Wisdom of Crowds by James Surowiecki is an excellent primer.
1.30.2008 1:57pm
Nathan_M (mail):
There is a crucial difference between prediction markets and other markets which Mr. Abramowicz ignores in this post.

In the stock market, overwhelmingly the incentives traders face are the incentives of the market. There are the occasionally, usually small, exceptions, but in general the way to make money with the stock market is to make money from the stock market.

With a prediction market that isn't necessarily the case. When a highway construction company decides to invest in the "what will congestion be in 20 years" prediction market (leaving aside that it's hard to imagine how such a market would give information to policy makers unless they can commit to not build new roads) it's likely that the company will see it can make money overall by losing money in the prediction market to skew the prices there.

This would create arbitrage opportunities for other traders, but the opportunity would be highly risky, and only available to traders who traders with the necessary knowledge -- which is specialized and highly costly to acquire.

All of which is to say there are fundamental reasons, which Mr. Abramowicz is not addressing, to think prediction markets would not be useful in guiding policy makers. I would suggest studies done of existing prediction markets, where there is no potential financial benefit from manipulating the market, are not necessarily applicable.
1.30.2008 2:01pm
Michael Abramowicz (mail):
J.F. Thomas -- You said, "Next you'll tell me that rather than clinical trials of drugs we should just have prediction markets to decide whether they are safe and effective." Well, sort of -- see this section. What you're missing is that with sufficient subsidies (on par with the amounts currently spent on evaluating new drugs), the experts would be the ones participating.

Nathan_M: You make good points, and I agree that we need more experiments of manipulation (though preliminary experiments support my case). I think you overestimate the riskiness of the arbitrage opportunities in these relatively low-stakes markets. At least some firms will gather reputations for being good predictors, and if someone anonymous bets against them, lots of people will pour money behind the reputable predictors. Finally, as I indicated, we could have prediction markets with only trusted participants if you're still unconvinced.
1.30.2008 2:10pm
Elliot123 (mail):
In the traffic model - how does the contract expire? What is the event? Does it simply expire at the last price on the last day? Is there some event? What corresponds to delivery?
1.30.2008 2:31pm
Michael Abramowicz (mail):
Elliot123 -- Lots of possibilities here. Could be the number of vehicles on a particular highway on a particular date (perhaps conditional on a given level of highway funding). Admittedly one challenge is getting people to invest now for distant payoffs (even if we award interest on the payoff price).
1.30.2008 3:15pm
Elliot123 (mail):
"This would create arbitrage opportunities for other traders, but the opportunity would be highly risky, and only available to traders who traders with the necessary knowledge -- which is specialized and highly costly to acquire."

Let's consider what happens at the limits of a market that is allowed to move from 0 to 100. If I want to push the price up to 100, I will face shorts who can take a position for nothing. Bill Gates can sit there all day long paying 100, while I calmly sell to him for zero.

The risk to both of us is zero. If he keeps the market at 100, he gets back all his money. He pays 100, he gets back 100.

I also face no risk. I pay zero for a short at 100. If it expires at 100, I lose zero.

But Gates has to keep coming up with money, while I have to do nothing. He will eventually run out of money, and I will then sell at .99, forcing him into a loss on all his contracts, and giving my position a corresponding profit. If he comes up with more money and bids 100, I sell to him s long as he can keep getting new money.

I concede examples at the limit are extreme, but the point is the huge risk the maniplator must take to overcome the much smaller risk his opposition faces.

The obvious solution here is to remove the upper and lower limits limit, but it can still be a very expensive risk for someone trying to force the market in one direction. However, removing the limits also exposes everyone to potential unlimited loss. (At 0 to 100, my loss per contract is limited.)

Many have tried to use financial power push futures markets in their direction, and have lost very badly. However, a huge difference is that the futures markets are based on actual physical commodities, financial instruments, or measurable events.

Take a look at the current situation in Minneapolis wheat for a real world example of a slugfest.
1.30.2008 3:27pm
Elliot123 (mail):
"Elliot123 -- Lots of possibilities here. Could be the number of vehicles on a particular highway on a particular date (perhaps conditional on a given level of highway funding). Admittedly one challenge is getting people to invest now for distant payoffs (even if we award interest on the payoff price)."

I doubt there would be any more interest in holding a contract to expiration in the highway contract than there is now in a normal futures contract. Probably less, in fact, if they have to wait fo delivery. Normally, nobody is left holding at expiration unless they are interested in taking delivery, made a mistake, or are totally screwed. Everybody else is out before expiration, and they have used the market for whatever their purpose at various points in the life of the contract.

I would further suspect that the highway authorities wouldn't care where the contract settled either. They would be watching the behavior of the market over time to get whatever info they could glean from it.
1.30.2008 3:42pm
J. F. Thomas (mail):
For example, the rules might provide that a drug that has at least a 90 percent chance of eventual approval can be sold immediately but would need to be pulled off the market if the probability later fell below 50 percent or if an adverse decision were ultimately reached. There are dangers associated with such a regime: patients and others might stockpile a drug

Umm, what about the danger that patients might die or serious side effects would be discovered? You seem to be missing the point of the FDA approval process.

"Gee, people think our new jet engine design is good, it's got a 90% chance of approval on the prediction market. We can start attaching it to planes. If the approval drops below 50%, we can always go back to the old engines."

I want to fly on a plane designed by aeronautical engineers who think like that.
1.30.2008 5:45pm
J. F. Thomas (mail):
Could be the number of vehicles on a particular highway on a particular date

This is silly. It is one thing to bet on who will be elected president or will win the Superbowl and by how much (and the point spread is there to make sure everyone sane just doesn't bet on the Patriots), but on what traffic patterns will look like in twenty years. That may not be a certainty but based on solid empirical data and research you can make a pretty good prediction about such things. Having a bunch of people who know nothing about a particular city, the growth patterns, or do not have professional training in predicting such things bet on it because they don't like the line on the Lakers game is not going to improve your decision making process one bit.
1.30.2008 5:53pm
Elliot123 (mail):
Just look how well the USSR did with all those experts making economic predictions.
1.30.2008 9:10pm
Michael Abramowicz (mail):
J.F. Thomas -- (1) The whole point is to weigh the serious dangers you are considering with the also serious danger that delayed approval of a drug might lead to patient deaths. (2) You seem again to be making the mistake of assuming that the same Joes who play tradesports are going to be making highway predictions. In a world with many prediction markets like this, subsidized at high levels, firms would specialize in this. As Elliot123 implies, this could be a lot better than having official experts make decisions.

Elliot123 -- Yes, good point, there are lots of ways that we could effectively pay people off shortly after their bets well before delivery. A simple one is that after the market settles, people could sell their interests to those willing to wait for delivery.
1.31.2008 11:34am
J. F. Thomas (mail):
(1) The whole point is to weigh the serious dangers you are considering with the also serious danger that delayed approval of a drug might lead to patient deaths.

How can you weigh these dangers if you don't have the results of the clinical trials? The point of the approval process is to have a scientifically rigid procedure to ensure safe and effective drugs. No market can substitute for safety testing. The quote I provided was from your article. You are the one who ignored the danger of adverse health effects, not me.

My second point was partly facetious. The point is that predicting something like traffic patterns twenty years in the future is a pretty exact science--an accurate prediction market isn't going to tell you anything a competent professional evaluation will and is therefore a waste of money (because I assume someone is collecting a commission on these bets).

Certainly, you are not proposing limiting the pools only to competent gamblers are you? Anyone will be able to get in on the action, so while professional firms who know what they are doing in predicting traffic patterns may indeed bet, so can anyone who is bored with betting on football or basketball.
1.31.2008 5:12pm