2.6. Prediction Markets¶
A prediction market is a specialized gpAsset such that total debt and total collateral are always equal amounts (although asset IDs differ). No margin calls or force settlements may be performed on a prediction market asset. A prediction market is globally settled by the issuer after the event being predicted resolves, thus a prediction market must always have the global_settle permission enabled. The maximum price for global settlement or short sale of a prediction market asset is 1-to-1.
Table of Contents
In the following, we denote a positive outcome as a predication market that resolves to true (i.e. a price feed of 1) and a negative outcome to resolve to false (i.e., a price feed of 0)
If the bet resolves to true (i.e. a price feed of 1), then the PM-asset can be settled release the collateral to the holder of the asset.
If, instead, the bet resolves to false (i.e. a price feed of 0), then those that sold the PM-asset on the market and went short, made a profit since it PM-asset became worthless.
Prediction markets are assets that trade freely and can be borrowed from the market at a 1:1 ratio with the backing asset (which could be any other asset, including GPH, USD, GOLD).
A user can take either bet on a positive outcome, or a negative outcome. We here show how this works, technically.
If you are confident that the bet will resolve positive, you want to hold that particular PM-asset since it allows you to settle it for it’s collateral on a 1:1 basis.
In order to get hold of those tokens, you can put a buy order for them at any price (between 0 and 1) and wait for it to be filled, or buy at market rates. By this technique, a user can pre define at which odds to buy shares.
For instance, if you think that the bet resolves positively at a probability of 80%, you can put your buy order at a price of 0.8. If the bet resolves positively (price feed of 1), then you can settle your shares at 1 and make a 20% profit.
If you can buy tokens at a price of 0.2 (i.e. market participants think it is unlikely to resolve positively), then you could make 80% profits at a risk of loosing with 80% probability.
After closing of the bet, a user can claim his profits by settling his borrow position and taking out the collateral:
Settlement in the CLI wallet:
>>> settle_asset <account> <amount> <symbol> True
Borrowing in the GUI wallet: A settlement button is available when hovering the asset in your account’s overview.
In order to bet for a negative outcome (bet resolves to false with a price feed of 0), you need to sell the tokens. In order to get them, you should not buy them at the market, but instead borrow them from the network by paying collateral at a 1:1 ratio.
For example, in the PM.PRESIDENT2016 if you want to bet on a negative outcome with 100k GPH, you can borrow 100k PM.PRESIDENT2016 by paying 100k GPH to the network.
Since PM-Assets can technically be pegged by any other asset, you may need to pay USD (or anything else) instead of GPH.
Once you borrowed the token, you can sell them at any price between 0 and 1. If you thing the probability of a negative outcome is 20%, you should consider selling your tokens at 0.2.
If the bet resolves negatively (price feed of 0), your debts is worth debt = amount * price = 0 GPH, you can reclaim your collateral at zero cost, and get to keep 20% profits from selling the token at 0.2. If instead the bet resolves positively and you sold all tokens, you cannot close your borrow position to redeem your collateral. However, your total loss is reduced by 20% for selling the tokens at the market.
If, by the end of the bet, you still have some of the tokens left, you can of course close your borrow position partly and redeem the corresponding percentage of the collateral.
A price feed needs to be published for the prediction market by the issuer or feed producer. It is essentially a global settlement which will set the parameters of the asset such that the holders of the asset can settle at the outcome of the bet (0, or 1). The details are shown in the guide pm-close-manual (ref: docs.gph.ai material)