https://en.wikipedia.org/wiki/Prediction_market
Unfortunately, they are generally illegal in the US.
https://en.wikipedia.org/wiki/Prediction_market
Unfortunately, they are generally illegal in the US.
Prediction markets aren't magic, they aggregate views. If they say there's a 15% chance of something happening, by golly, sometimes it will happen. The point is (1) they are self-correcting and (2) they have a much better track record that polls. Not even the "market can stay irrational longer than you can stay solvent" rule applies. With prediction markets, you know exactly when and how the outcome will be settled.
There's a simple fallacy here - you can't use a market mechanism to define the probability of the outcome of a one-off event like a major referendum.
All you have are hopes and fears about the probability, which are always - by definition - limited and subjective.
In fact you're simply quantifying opinions about opinions. That's one step removed from what polls do. Polls fail because of practical sampling errors (and sometimes because of political bias) - not because no one understands the theory of polling.
There's no way a prediction market mechanism can be self-correcting, because by the time you have real data the question is settled.
Some people may find this kind of thing useful. I'm not convinced it's any more than entertainment.
> That makes prediction markets non-falsifiable.
Why ? We can just make a sufficiently big list of prevision which are estimated around 15% now and check in one year that around 15% of them happened.
No, it means you can't assess prediction markets, or any mechanism for making probabilistic predictions, from a one-off event. I was responding to tobias3 regarding Brexit.
> There's no way a prediction market mechanism can be self-correcting, because by the time you have real data the question is settled.
You misunderstand. I'm not claiming that the market for a single event is self-correcting. I'm claiming that if someone discovers a bias in prediction markets as a whole (e.g., the pop-science biases discussed in the Economist article), then traders can profit off this bias and it will be removed from the market's prediction.
There's a lot wrong with your other points, but it would take too much discussion to draw it out.
It was 10% in the hours after the polls closed.
The odds for a Leave victory in England were out of step with the odds for a UK wide leave vote for many hours (even though overall Leave was never going to be happen without England going Leave).
Consider a perfect prediction market. This means that if it says "the probability of X is 0.6" 100 times, then approximately 60/100 times X will occur. If you wrongly interpret that to mean "X will occur", a perfect prediction market will be "wrong" 40/100 times.
That's your failure to interpret the market, nothing else.
Of course this is difficult to measure, because there are not enough natural experiments (i.e. 100x Brexit) and there are many variables at play here.
We'll have to wait for papers with regards to Brexit + predicition markets for a more detailed analiysis, but still, for the time my trust in prediction markets is down a bit.
(the reason I did this is because I'm pro Remain)