Prediction Market Design: Liquidity and Fees
A prediction market with no liquidity gives useless prices. Here is how market design, liquidity provision, and fees decide whether your odds mean anything.
A prediction market is only as good as its prices, and its prices are only as good as its liquidity. You can have a perfectly worded question and a trustworthy resolver, and still get garbage odds if nobody can trade without moving the price ten points. Liquidity is what lets the market convert scattered opinions into a sharp probability. Fees are what pay for that liquidity to exist. Get these two wrong and your market produces numbers that look like forecasts but mean nothing. This is the plumbing nobody wants to talk about, and it decides everything.
Why thin markets lie
In a thin market, a single modest trade swings the price wildly, because there is almost nothing on the other side. That means the displayed odds reflect the last person who showed up, not the crowd's real estimate. A market showing 70% might flip to 55% the instant someone with a few hundred dollars trades, which tells you the 70% was never a real signal.
This is the difference between a market and a poll dressed as one. The whole reason prediction markets can beat expert forecasts is that money forces honesty and depth aggregates many views into one price. Strip out the depth and you lose the mechanism. You are left with a poll that costs money to answer.
The two ways to bootstrap liquidity
New markets have a cold-start problem: nobody wants to trade where there is no liquidity, but liquidity comes from trading. Two designs break the loop.
An automated market maker. Instead of matching buyers to sellers, a formula always quotes a price and takes the other side of any trade. The market maker guarantees you can always trade, at a price that moves predictably with volume. This is how most modern prediction markets bootstrap: the AMM provides liquidity from day one, and the curve is designed so the price reflects the ratio of money on each side. The cost is that the market maker can lose money if the outcome surprises it, which someone has to fund.
A traditional order book. Buyers and sellers post limit orders and the market matches them. This gives tighter prices in liquid markets but dies in thin ones, because an empty book means no trades. Order books work when you already have active traders; they are a poor fit for a fresh, low-volume question.
Most community-scale markets should start with an AMM. Order books are for markets that have already earned their volume.
How fees make or break the market
Fees are not just revenue. They are a design lever with sharp tradeoffs.
Charge too much and you kill volume, because every trade has to overcome the fee to be worth making. High fees widen the effective spread, which pushes casual traders out and leaves you with the thin market you were trying to avoid. Charge too little and you cannot compensate whoever is providing liquidity, so the liquidity leaves. The market maker or liquidity providers are taking real risk, and the fee is how they get paid for it.
The design question is who the fee funds. If fees go to liquidity providers, you are subsidizing depth, which is usually what a young market needs. If fees go purely to the operator, you had better have organic depth already, or you are taxing a market you are also starving. Publish the fee structure before anyone trades, the same way you publish the resolution rule, because a fee changed mid-market reads as a rug pull.
Design for the market you actually have
The honest operator move is to match the mechanism to the volume you can realistically attract. A niche community question will never have order-book depth, so give it an AMM and a fee that pays for liquidity. A high-traffic public market can support an order book and tighter fees. Do not copy the design of a market a hundred times your size.
On MintVote I treat liquidity and fees as the difference between a market that produces a real probability and one that produces a decorative number. The question and the resolver get the attention, but the liquidity design is what makes the price worth believing. Build the depth, price the fee honestly, and pair it with a verifiable, auditable resolution so the number people trust was both fairly priced and fairly settled. A market without liquidity is not a market. It is a rumor with a percentage sign.