Polymarket Liquidity Rewards: Scoring, Maker Rebates and Net Economics

Polymarket runs two separate maker programs, and they pay for different things. Liquidity Rewards pay for resting orders that sit close to the midpoint and meet a market's size and spread thresholds, whether or not those orders ever trade. Maker Rebates return a share of taker fees when a resting order actually fills. Both pay daily in pUSD with a $1 minimum, and both are discretionary. Neither is a subsidy for taking inventory risk blindly.

The distinction matters for strategy. A quote that never fills can earn liquidity rewards but produces no rebate. A quote that fills earns a rebate but also acquires the position the taker wanted to shed. The first is a payment for displaying depth; the second is a payment for absorbing flow. A maker who confuses them will misprice both.

Liquidity Rewards: the scoring rule

Each incentivized market publishes a minimum qualifying order size, a maximum qualifying spread and a daily reward rate. The order book is sampled once per minute at a random offset, so an epoch — one UTC day — contains up to 1,440 samples. An order outside the maximum spread, or below the minimum size, does not score at all.

For an order at spread s inside a maximum spread v, the score is S = ((v - s) / v)^2 × b, where b is an in-game multiplier. Size is measured in shares. The formula is quadratic in the distance from the midpoint: an order at half the allowed spread scores a quarter of an order at the midpoint, not half.

Both sides of the market are scored separately, and the market score is the minimum of the two sides. A single-sided quote still scores when the midpoint is between $0.10 and $0.90, but at one-third of its one-sided value. Outside that range, only two-sided liquidity scores. The design favors balanced quoting, not a large order parked on one side.

Your share is your Q_min divided by the sum of all makers' Q_min in the sample, summed across the day's samples, then normalized again. In plain terms: depth, tightness and two-sidedness all matter, and size alone does not win.

A worked score

Take a market with a maximum spread of 3 cents. A 100-share bid one cent from the midpoint scores ((3 - 1) / 3)^2 × 100 = 44.4. A 200-share bid two cents away scores ((3 - 2) / 3)^2 × 200 = 22.2. A 100-share ask on the complementary outcome, one cent away, scores 44.4. The first side total is 111.1.

On the other side, suppose the maker has 100 shares at 1.5 cents (25.0), 100 shares at 2 cents (11.1) and 200 shares at 0.5 cents (138.9), for a total of 175.0. The minimum of the two sides is 111.1. The single-sided adjustment — one-third of each side — is 37.0 and 58.3, so the minimum remains 111.1.

If that maker's Q_min is 10% of the market's total in a sample, the maker earns 10% of that sample's reward allocation. If the market pays $100 per day and the share stays at 10%, the payout is about $10 per day. The 10% share is not stable: other makers can tighten, add size, or leave, and the pool can be reduced.

Maker rebates: the fee-sharing layer

Maker rebates are funded by taker fees. For each filled maker order, the fee equivalent is C × feeRate × p × (1 - p), the same curve used to charge takers. Your daily rebate is your fee equivalent divided by the market's total fee equivalent, multiplied by the rebate pool. Totals are calculated per market, so you compete only with other makers in that market.

The rebate percentages are 20% for crypto, 15% for sports, and 25% for politics, finance, economics, culture, weather, other/general, mentions and tech. Geopolitics has no taker fee and no rebate. A market with a temporarily zero fee also pays no rebate for the duration of the holiday. Payouts are daily in pUSD, with a $1 minimum.

An example: 1,000 shares of a politics market at $0.40 generates a fee equivalent of 1,000 × 0.04 × 0.40 × 0.60 = $9.60. If the day's rebate pool for that market is $1,000 and your fee equivalent is 5% of the market total, your rebate is $50. That is separate from any liquidity reward and from the taker rebate tiers, which reward volume rather than resting depth.

The two programs can stack on the same resting order, but not automatically. A resting order that scores for liquidity rewards and later fills can earn both, provided the market is fee-enabled and the order met the reward thresholds while it rested. A fill on a fee-free market earns no rebate, even if the order scored for rewards.

The economics that bite

The headline rate is not the return. Consider a maker with $2,000 of inventory quoting one cent wide on a market that pays $100 per day. A 10% share is $10 per day, or 0.5% of inventory per day. One adverse two-cent move on 1,000 shares produces a $20 mark-to-market loss — two days of rewards. Filled repeatedly on the wrong side, the strategy loses more than it earns.

Adverse selection is the core problem. Your resting bid is most likely to fill when someone with better information wants to sell. The reward is compensation for that risk, not a risk-free yield.

Capital is the second constraint. Rewards are paid on quoted size, but the capital supporting those quotes is locked until the orders are cancelled or filled. A market that pays 20% annualized on $10,000 of quotes is not paying 20% on the $100,000 of inventory you need to survive a run of fills. Compare reward income with the capital at risk, not with the notional of the quotes.

Competition is the third. Reward pools are fixed per market and per day. More makers dilute the share; a tighter quote wins more of the sample but fills more often. The equilibrium is not stable, and a market that looks profitable in a snapshot can be unprofitable a week later. Rates and configurations can change, and rewards can be removed for wash trading or self-matching.

Finally, the payout mechanics matter. The $1 minimum means a small maker may accrue rewards that are never paid. Rewards are distributed in pUSD, so they are another position in the same collateral, not a withdrawal to a bank. A quote that drifts outside the spread band stops earning even though it remains on the book.

An operational playbook

Start by reading the market's actual configuration, not its headline. The reward-markets endpoint returns the minimum size, maximum spread, daily rate and date range. Do not infer an active reward from a market's appearance in a list.

Quote both sides, keep the order inside the spread band, and use post-only so a marketable order is rejected rather than turned into a taker fill. Size to the minimum with a buffer, because a partial cancel can drop you below the threshold. Use GTD around scheduled events so a stale quote does not sit through a catalyst. Track fills separately from rewards: a high reward rate with a low fill rate is a display business, while a high fill rate with a low reward rate is a flow business.

Reconcile daily. The rewards endpoint reports earnings by date and asset; the rebates endpoint reports maker rebates by market. Compare those totals with your inventory PnL over the same period. A maker who reports $500 of rewards and $900 of inventory losses has not earned $500.

Risks

• Adverse selection and inventory risk. Fills cluster when the market moves against you.
• Reward and rule changes. Rates, spreads, minimum sizes and eligibility can change at Polymarket's discretion.
• Liquidity and exit risk. A market that is deep enough to quote may not be deep enough to exit at the same spread.
• Operational risk. Stale quotes, failed cancels, API rate limits and engine restarts can leave unwanted exposure.
• Compliance risk. Wash trading, self-matching and other inauthentic activity can disqualify a maker from rewards.

Treat rewards as a rebate on providing a service, not as yield. The service is liquidity, the cost is adverse selection, and the edge is execution discipline.