Current prediction market thinking is focused solely on discrete mechanism. That is, there is a moment in time when prediction market must be settled. Still this approach failed to gain traction. So it is probably a good idea to ask why?
My assumption is that reality is changeable in nature, and the truth deviate in time. So even seemingly improbable events such as death could be reversed in a mind of the crowd.
I failed find any insights on this topic. So let us start discussion on continuous prediction market idea in which trading could last indefinitely expressing current estimation of the probability in such a way that participants could profit based on existing beliefs and not only on future beliefs.
The design came to my mind is shockingly simple and is a mix between 2 brilliant mechanisms: conviction voting and bonding curve:
- Let us assume that we have some fact. Let say
- Participant could vote for
falseby buying tokens in quadratic bonding curves
- Estimator is proportion between
truereserve pool and
- Periodically loosing reserve pay some
payoff percentto the winning reserve
The question I ask is how
payoff percent could be defined? I like the idea of adaption quorum biasing. This concept could be applied to our case: the more contradictory is opinion the more time it needs to converge.