Polymarket Fee Structure: Mathematical Implications for Systematic Traders

On March 30, 2026, Polymarket expanded its fee structure from just Crypto and Sports to nearly every market category on the platform. The change was announced through a quiet documentation update - no blog post, no tweet. Most traders found out when their next trade cost more than expected.

The fee model is not a flat percentage. It is a probability-weighted curve that charges the most when uncertainty is highest and the least when outcomes are nearly certain. That design choice has real implications for how and where systematic traders should operate.

The Formula

Polymarket calculates taker fees using:

fee = C × feeRate × p × (1 - p)

Where C is the number of shares traded, feeRate is a category-specific constant, and p is the share price (which represents the implied probability of the outcome).

Makers pay zero fees. Only takers - traders who lift existing orders from the book - are charged.

The critical term is p × (1 - p). This is the variance of a Bernoulli distribution. It peaks at p = 0.50 where it equals 0.25, and it approaches zero as p moves toward 0 or 1.

A trade at 50¢ generates the maximum fee. A trade at 10¢ or 90¢ generates 36% of the maximum. A trade at 5¢ or 95¢ generates just 19%.

Category Fee Rates

Each market category has its own feeRate parameter:

The "After Rebate" column reflects the net cost if you factor in the Maker Rebate program's redistribution. Crypto makers receive 20% of collected taker fees. All other fee-enabled categories receive 25%.

Geopolitical and world events markets remain completely fee-free. Polymarket has stated it does not profit from trading activity on those markets.

The Fee Curve Shape

The p × (1 - p) term creates a symmetric parabola. A trade at 30¢ generates the exact same dollar fee as a trade at 70¢. The shape means:

• High-uncertainty markets (40¢–60¢): Maximum fee extraction. This is where the platform generates the most revenue per share traded.
• Moderate-probability markets (20¢–40¢ or 60¢–80¢): Fees drop meaningfully. A trade at 20¢ costs 64% of the 50¢ fee.
• Near-certain outcomes (<10¢ or >90¢): Fees become negligible. A trade at 5¢ costs just 19% of the peak.

For a 100-share Crypto trade, this translates to $1.80 at 50¢, $1.15 at 20¢, and $0.34 at 5¢.

The practical consequence: traders who specialize in high-conviction positions near the extremes pay structurally less than traders who operate in uncertain, contested markets.

Fee Precision and Edge Cases

Fees are rounded to 5 decimal places. The smallest possible fee is 0.00001 USDC. Anything below that rounds to zero, which means very small trades at extreme prices may incur no fee at all.

Fees are collected in shares on buy orders and in USDC on sell orders. This asymmetry matters for position tracking - your share count after a buy is slightly less than the raw calculation suggests.

Fees apply only to markets deployed on or after the category's activation date. Pre-existing markets in the same category remain fee-free. The feesEnabled flag on the market object is the authoritative check.

Maker Rebates: The Other Side

Polymarket does not keep all collected fees. Taker fees fund a Maker Rebates program that redistributes a portion daily in USDC to liquidity providers.

Rebates are proportional to your share of executed maker liquidity within each market. You compete only with other makers in the same market, not across the platform.

The rebate creates a two-sided incentive: takers pay fees that fund makers who provide the liquidity takers need. Deeper books mean tighter spreads, which theoretically offsets some of the fee cost through better execution prices.

What This Means for Systematic Strategies

The fee structure creates distinct zones of fee efficiency:

Low-fee zone (p < 0.15 or p > 0.85): Fees are under 50% of the peak. Strategies that identify mispriced tail events - outcomes the market prices as very likely or very unlikely - operate in the cheapest part of the curve. This is where contrarian and arbitrage strategies have the highest fee advantage.

Peak-fee zone (0.40 < p < 0.60): Maximum fee extraction. Strategies that trade contested outcomes pay the full cost. Scalping in this zone requires enough edge to clear the fee hurdle on every round trip.

Category arbitrage: Crypto markets carry 2.4× the fee rate of Sports markets. A systematic trader running the same strategy across categories will see meaningfully different returns purely from fee differential. Finance and Politics at 0.04 feeRate represent the middle ground.

Maker advantage: Since makers pay zero fees and receive rebates, any strategy that can express its views through limit orders rather than market orders transforms from fee-payer to fee-receiver. This is the single largest structural edge available.

The Revenue Math

At roughly $1 billion in daily trading volume, the fee structure is projected to generate $800,000 to $1 million per day for Polymarket. That estimate depends on the probability distribution of trades across the fee curve - volume concentrated at 50¢ generates more revenue than volume at extremes.

For context, Polymarket operated with zero trading revenue for most of its existence. The fee rollout that began with Crypto in January 2026 and expanded to nearly all categories by March 30 represents the platform's first real monetization model.

Observations

The fee formula is elegant. By tying cost to uncertainty rather than notional value, it charges the most where information is most valuable and the least where outcomes are already decided.

For systematic traders, the implications are structural:

• Tail-event strategies pay less in fees than contested-outcome strategies
• Maker strategies receive rebates; taker strategies absorb the full cost
• Category selection affects returns independently of prediction accuracy
• Geopolitical markets remain a fee-free zone for any strategy that can find edge there

The fee structure is not static. Polymarket's changelog shows rates have been adjusted multiple times since January 2026, and the documentation explicitly warns API users not to hardcode fee assumptions. The feeSchedule object on each market is the source of truth.

None of this is a recommendation to trade on Polymarket or any prediction market. These are observations about fee mechanics and their mathematical properties.

Interested in exploring Polymarket? You can sign up through our referral link - it costs you nothing extra and helps support SwapHunt's independent research.