2. Prediction Markets
Overview
• Two types of markets currently available on Zeitgeist: Categorical ("How
will event X play out?") and scalar ("What will measurement A be in the
future?").
• In the future: Combinatorial markets ("How will event X play out
contingent on event Y?").
• Traders (informants) buy tokens according to which outcomes they think
are likely to occur. They are rewarded if they get it right and punished if
they get it wrong.
• Liquidity providers make these markets to gather information.
4. Categorical Markets
• Suppose we have a categorical market with outcomes
• Suppose an informant believes has a probability of . When would we expect them to buy
?
• A good way to
f
igure this out is to calculate the expected value. If the current price of is ,
then my expected pro
f
it of buying is...
• .
• This is positive if and only if .
• Informants are expected to buy if and only if the current price (i.e. the current prediction) is
lower than their prediction .
A, B, …
A p
A
A q
A
𝔼
= p ⋅ $1 − q
p > q
q
p
5. Scalar Markets
• Scalar markets a range and two assets: LONG and SHORT.
• They are used to predict the value of some future quantity or measurement in
that range.
• For example: "What will the USD Currency Index be at the end of Q2 2023 (UTC)?"
• Let p be the price of SHORT and q the price of LONG. Then the prediction is
.
• Recall that . Thus, the higher the price of LONG (resp. SHORT), the
closer the prediction is to (resp. ).
[a, b]
v = pa + qb
p + q = 1
b a
6. Scalar Markets
• Suppose is the current prediction and is the value that you predict.
When do we expect an informant to buy LONG or SHORT?
• A little bit of calculating gives... and
.
• An informant is expected to buy SHORT if and LONG if .
v w
𝔼
long = (w − v)/(b − a)
𝔼
short = (v − w)/(b − a)
w > v w < v
7. Summary
• Users are incentivized to "
f
ix" the prediction of the market.
• They are rewarded if they are correct and they go broke if they're wrong.
• Where do the rewards come from...?
9. Liquidity Providers
Who are the Traders Buying from?
• Anyone can provide liquidity to a market on Zeitgeist... This means taking the
opposite side of the bet (it's informants vs. LPs).
• Why would you want to provide liquidity?
• Liquidity providers "make the market". They pay traders to make predictions
and receive information.
• Liquidity providers can collect fees. Largest allowed fee is 10%.
• "Everybody's wrong." If the LPs have inside information, they can pro
f
it off of
everybody else's ignorance.
• Nevertheless, LPing is inherently risky.
10. Risk Mitigation
• Set good initial predictions. The better the initial prediction, the less the
informants can pro
f
it off of making corrections.
• Prepare for information shock by con
f
iguring your market with long grace
periods.
• Withdraw liquidity on information shock if necessary.