APMM - Automated Predictive Market Maker
The APMM is Signa's automated market-making model for Yes/No prediction markets.
Core idea
Instead of treating the two sides of a market as unrelated instruments, the APMM prices them as complements:
P(Yes) + P(No) = 1
This keeps the market aligned with probability semantics.
Intuition behind the model
Pricing depends on the state of the market, including:
- L = Liquidity depth parameter
- S_yes = Yes Token supply
- S_no = No Token supply
At a high level:
P(Yes) = f(L, S_yes, S_no)
The exact implementation can vary, but the important design constraint is the complement relationship between the two outcomes.
Why use an APMM for prediction markets?
| Dimension | Signa APMM | Traditional CLOB | Traditional LMSR |
|---|---|---|---|
| Continuous liquidity | Quote available from the curve | Depends on resting orders | Available by design |
| Probability semantics | Native Yes/No complement | Requires interpretation | Often less intuitive for users |
| Capital structure | One bilateral market design | Efficient for active books | Can be costly in some setups |
| Initialization | Supports non-50/50 priors | Flexible quoting | Often centered around fixed defaults |
Example
Initial state:
P(Yes) = 0.40P(No) = 0.60
If users buy more Yes, the Yes price rises and the No price falls. If demand later shifts toward No, the curve adjusts in the opposite direction.
The purpose is not just continuous execution. It is to keep execution aligned with interpretable probability pricing.