notes-econ-predictionMarkets

SchellingCoin

"

Note that, as mentioned in the post on P + epsilon attacks, there is a refinement by Paul Sztorc under which minority voters lose some of their coins, and the more “contentious” a question becomes the more coins minority voters lose, right up to the point where at a 51/49 split the minority voters lose all their coins to the majority. " -- [1]

Subjective SchellingCoin

Designed to answer 51% attacks against SchellingCoin?.

"

Prediction markets to fund science

Prediction markets might be used to fund science research. Here's how it would work. People would make bets on whether X would be invented, or Z question would be resolved, by various dates. When scientists actually make progress in discovery, they then have surprising information; they can place their own bet on the market. After they publish, the consensus odds will adjust in their direction, and their bet will pay off.

Prediction pseudo-market

Could you make a prediction pseudo-market, where instead of buying/selling assets, ppl just asserting probabilities and certainties, and got a prediction score?

You still need a mechanism to calculate what the 'market price' of questions is, so that you can reward ppl who correctly bet against the trend more. Also, maybe weight questions which have higher bet volume more? Eg you get more points for predicting questions of wide interest and great import than for predicting who will win your local little league championship.

This could be an Ethereum project.

Corruption resistance

One issue with prediction markets (or any market) is that people with power over the outcome of a question can do something they are not supposed to do in order to make a certain outcome take place. Eg the way a sports game or match might be 'fixed' where one predetermined side would intentionally 'throw the game' in order to make money betting onn it.

Can you make a prediction market that is resistant to match fixing? Perhaps via some sort of asymmetric question structure?

I'm not sure but my guess is yes; note that you can remove matching (in the market sense; ie the need to match buyers and sellers) by moving to a pseudo-market, which, as seen in the previous section, you can probably do if the goal is purpose is purely to predict (and if external incentives for participation exist), rather than if there are actual limited goods being distributed by a market.

For example, a company might like to use a prediction market to predict how many widgets it will sell in fiscal quarter 2. But in a traditional prediction market the VP of sales might bet that no more than X wigets will be sold, and then have the power to make sure that happens, which would be bad for the company. If you only let people make bets that pay off when they sell "X or more widgets" then they can never have an incentive that conflicts with the company's. This would not work in an actual market, because trades would have to be matched because the number of goods must be conserved, but in a pseudo-market it's possible.

For example, you could specify that all questions in the market must be phrased so that working towards the 'yes' answer benefits the company, and then prohibit shorting. There would be a marketmaking bot that you initially buy from.

Markets that explain their reasoning

Currently markets are like a black box -- they give you an answer but they don't tell you why. In machine learning classifiers, we see that although black box classifiers can perform well, in many applications there is also a demand for classifiers that can explain their reasoning in an intelligible way.

Can we create a market that can explain its reasoning?