How do Polymarket liquidity rewards work?

The scoring formula, in plain terms

Polymarket's market-maker documentation publishes the arithmetic rather than describing it loosely, which makes the program unusually easy to check. An individual order scores S(v,s) = ((v-s)/v)² · b, where v is the market's configured max spread, s is the distance in cents between the order and the adjusted midpoint, and b is the order size. Because the bracket is squared, the score falls away faster than the order drifts. An order half of the way to the cutoff keeps a quarter of the score it would have earned sitting on the midpoint, not half.

Scores are not measured continuously. The documentation describes the book being sampled every minute using random sampling, which works out to 10,080 samples across a weekly epoch. An order that appears for thirty seconds and vanishes may never be sampled at all. At the end of the day each address's accumulated score is divided by the sum of every participant's score in that market, and that fraction is the fraction of the pool the address receives. It is a relative contest, so the same order earns less on a day when more size is quoting tightly beside it.

Two further rules govern whether an order scores at all. Each incentivized market defines a minimum qualifying order size, below which an order is ignored. And when the midpoint sits between 10 and 90 cents, one-sided quoting scores at a reduced rate, divided by a scaling factor the documentation puts at 3.0 across all markets. Outside that band, where the midpoint is under 10 cents or over 90 cents, the help center is blunt: liquidity has to be two-sided to score anything.

What 92 live markets actually carried

The reward parameters are public per market, so the program can be measured rather than assumed. On July 25, 2026 we asked Polymarket's public Gamma API about the first 120 market slugs in our July 24 snapshot and it returned 92 live markets, reading the rewardsMinSize, rewardsMaxSpread and clobRewards fields on each. Every figure below is as of that date and will drift as pools are refunded.

Forty-one of the 92 carried a non-zero daily reward rate and 51 carried none. Six of the 41 were nominal, a tenth of a cent a day, all of them July Bitcoin threshold markets, and a rate that size cannot clear the $1 daily payout floor for anybody. Setting those aside leaves 35 of 92, about 38 percent, with a funded pool of a dollar a day or more. Across those 35 the median pool was $150 a day and the total came to roughly $11,129 a day. The largest single pool in the sample was $2,000 a day on the market on whether the Israel and Iran ceasefire continues through July 25, followed by $1,250 a day each on two July 2026 Federal Reserve rate markets, including whether the Fed leaves rates unchanged after the July meeting.

The eligibility parameters were more uniform than the pools. Max spread was 4.5 cents on 62 of the 92 markets, 3.5 cents on 10, 2.5 cents on 8 and 5.5 cents on 5, with 7 markets carrying no reward parameters at all. Minimum qualifying size clustered at 200 shares on 39 markets and 50 shares on 30, with 100 shares on 10 more and a scatter at 20, 59 and 300. Worth noting that a market can carry parameters and no funded pool: the tightest cutoff we saw, 2.5 cents, appeared on eight markets, most of which had no daily rate attached.

The most striking pattern was where the pools were not. The five markets with the largest reported liquidity figures in the sample carried no funded pool at all, and all five were long-dated legs of 2028 US election fields, among them whether Beto O'Rourke wins the 2028 Democratic nomination, which reported roughly $2.6 million of liquidity with YES quoted near a third of a cent. The funded pools sat instead on near-dated questions with live news attached. That is consistent with a program aimed at tightening quotes where trading is actually happening, and it is a reminder that a headline liquidity figure on a long-tail leg is a different animal from a book somebody is being paid to keep tight.

Liquidity rewards are not maker rebates

These two get run together constantly, including by sites that write about the platform, and they are separate mechanisms with separate funding. Liquidity rewards are a per-market pool scored on where resting orders sit, and they existed through the years when Polymarket charged no trading fee at all. Maker rebates arrived with the fees themselves during 2026, are funded from taker fees collected in a category, and return a published percentage of those fees. Both flow to the same kind of participant, which is why the confusion persists, but they are measured, funded and capped in completely different ways.

The fee column is the one that changed most recently. Polymarket ran a crypto fee rollout in January 2026, piloted a market-order fee on selected sports events from February 18, 2026, and published a broader schedule after that, with geopolitics left fee-free. The per-category caps and the formula behind them are set out in Polymarket fees explained, and the underlying roles are covered in makers versus takers on Polymarket.

What the pool is compensating

A daily pool reads like free money on a page that only quotes the pool size. It is worth being precise about what the money is attached to. An order resting inside a 4.5 cent band around the midpoint is an order that anybody can fill at that price, which means it fills fastest precisely when somebody arrives who knows something the book does not. That asymmetry is the standard cost of quoting, and it is the reason venues fund pools like this in the first place: the pool has to be worth more than the losses to informed flow, or nobody quotes.

The $1 daily floor also does more work than it looks. Rewards are proportional shares of a fixed pool, so on a market with a $50 daily pool and heavy competition, a modest order can score a real fraction of the total and still land under a dollar, at which point it pays nothing and does not carry forward. Combined with the minimum qualifying size, which was 200 shares on the largest share of markets we sampled, the program is structurally aimed at participants quoting continuously in size rather than at occasional resting orders.

None of that is a judgement about whether the arrangement is worthwhile for anyone. It is the shape of the trade the parameters describe, and it lines up with the research finding this desk covers in who actually wins on Polymarket, where a wallet's maker volume share was the strongest single predictor of positive performance among the behaviours measured. A statistical association across many wallets is not a promise about any one account, and the pools described here are one input among several.

Checking a market's reward parameters yourself

Nothing in this piece required an account or a key. Polymarket's Gamma API returns rewardsMinSize and rewardsMaxSpread on every market record, along with a clobRewards array whose rewardsDailyRate carries the funded pool and whose start and end dates bound it. Querying a single market by slug returns all of it in one response. Polymarket's own interface also surfaces the reward available on an individual market order book and a rewards page showing the current day's accrual rather than a running balance.

The wider set of free endpoints, and their limits, is covered in does Polymarket have an API. For what this desk collects, at what cadence, and what it deliberately leaves out, see our methodology page. Cent Signals does not currently carry reward parameters in its snapshot: the figures above were read live from the API for this piece and are dated accordingly. The markets index carries the volume and liquidity we do collect, and the glossary defines each term as this desk uses it.

Frequently asked questions