Prediction Market Fees: Inside the Fee Wars

The prediction market industry has no standard fee structure. While traditional brokerages converged on zero-commission trading years ago, prediction markets range from 0.01% to over 15% in total costs — a difference of more than 1,000x between the cheapest and most expensive platforms.

This disparity matters as the exchange sector heats up. Polymarket re-entered the US market in late 2025 with fees designed to undercut Kalshi, while recently introducing 15-minute taker fees. PredictIt, which nearly shut down after a CFTC enforcement action, still charges the highest fees in the industry. Meanwhile, FanDuel and DraftKings have entered through CME-regulated exchanges, adding another layer of fee structures to an already fragmented market.

Key stats to know

• Most expensive: PredictIt charges 10% of gross profits plus a 5% withdrawal fee — the highest combined rate among political markets.
• Cheapest: Polymarket US charges 0.10% on trades. A $1,000 position incurs just $0.10 in fees.
• Most complex: Kalshi uses a probability-weighted formula where fees peak at 50/50 odds and drop toward the extremes.

We collected and documented the fee structures across all prediction market apps comparing real costs across trade sizes and how these fees can affect market efficiency and behavior.

A note on terminology:

“Taker” fees apply when you place an order that fills immediately against an existing order in the book—you’re taking liquidity. “Maker” fees apply when you place an order that sits and waits for someone else to fill it—you’re making liquidity. Platforms typically charge makers less (or nothing) because they want orders on the book for others to trade against.

The different fee models used

The prediction market industry has fractured into distinct fee philosophies. The table below breaks down fees across all prediction markets spanning US-regulated exchanges, global crypto markets, and emerging DeFi protocols. We verified each fee structure against first-party documentation where available and flagged discrepancies with third-party sources.

What the table doesn’t capture

Raw fee percentages miss important context:

• Spread costs: Platforms with thin liquidity often have wider bid-ask spreads that dwarf stated fees. A “0% fee” market with a 5-cent spread costs more than a “2% fee” market with penny-wide quotes.
• Deposit friction: Kalshi’s 3-30 day security holds on new deposits can lock capital during fast-moving markets. PredictIt’s 60-day international hold plus $40 wire fee effectively prices out non-US traders.
• Gas fees: Decentralized exchanges charge blockchain transaction fees that vary by network congestion. Polygon (Polymarket) typically runs $0.01-$0.50. Base (Limitless) and BNB Chain (Opinion, Myriad) are similarly cheap. Ethereum mainnet can spike to $20+ during volatility.
• Slippage: Large orders on illiquid markets move prices against you. This hidden cost often exceeds explicit fees on smaller platforms.

How the different fee models work

• Formula-based (Kalshi): Fees scale with contract price and quantity using the formula 0.07 × contracts × price × (1-price). This means 50/50 markets cost the most while extreme odds cost almost nothing. A $1,000 position at even odds runs about $35 in taker fees.
• Profit-based (Polymarket Global, PredictIt): You pay nothing to trade, but the platform takes a percentage when you win. Polymarket Global charges 2% of net profits at withdrawal. PredictIt charges 10% of gross profits on every winning trade plus 5% on withdrawals—a combined drag that can exceed 14% on successful positions.
• Flat per-contract (Robinhood, DraftKings, ForecastEx): Simple and predictable. Robinhood charges $0.02 per contract ($0.01 commission + $0.01 exchange fee). ForecastEx builds its $0.01 fee directly into the spread, so Yes + No always equals $1.01.
• Payout-based (FanDuel Predicts): The 2% fee applies to your potential payout at checkout, not your cost basis. If you’re buying a contract that could pay $100, you’re charged $2 regardless of whether you win.
• Dynamic/adaptive (Kalshi formula, Limitless, Opinion): Fees change based on market conditions. Limitless rewards early conviction with 0.03% fees that climb to 3% as resolution approaches. Opinion peaks at 2% when probability sits at 50% and drops toward zero as outcomes become more certain.

Platform-specific notes and discrepancies

Several markets have a few fee quirks worth noting.

The official Polymarket US fee schedule at polymarketexchange.com shows a 0.10% taker fee (10 basis points), not the 0.01% reported in some November 2025 coverage. At 0.10%, a $1,000 trade costs $1.00 in fees. This is still far below Kalshi but 10x higher than initially reported on launch.

Kalshi’s S&P 500 and Nasdaq-100 markets use a halved fee multiplier (0.035 instead of 0.07), cutting trading costs by 50% compared to standard markets, while ForecastEx stands alone in paying interest on your collateral while you hold positions. The rate tracks 0.5% below the Fed Funds rate, effectively subsidizing long-term positions.

Crypto.com waives its $0.10 technology fee when your contract settles in-the-money, leaving only the $0.10 exchange fee on winning positions. And the ~3% figure for Myriad comes from DefiLlama protocol data; Myriad’s official documentation describes fees as going to liquidity providers and protocol development but we couldn’t find a specific rate during our research.

Fee example on $100 in trades, favorite to win

.

*Fee estimated; platform may not offer this market. †2% of profit, charged at withdrawal. ‡10% profit fee + 5% withdrawal fee. *Polymarket Global charges 2% of net profits at withdrawal, not per-trade. †PredictIt’s 10% profit fee applies only if you win; this assumes a $50 profit.

What this means:

At small sizes, flat per-contract fees (Robinhood, Crypto.com) take a larger bite than percentage-based models. A $100 Robinhood trade pays the same $4 whether you’re trading a 1% longshot or a 99% lock. Polymarket’s zero-fee structure gives casual traders the most runway to learn without fee drag eating into limited capital.

Coin-flip scenario

You bet $100 on a market trading near even odds. This is where Kalshi’s formula-based fees begin to peak, while PredictIt’s percentage model looks less brutal—until you factor in the withdrawal fee eating another 5%.

DraftKings, FanDuel and Fanatics have highest fees:

DraftKings, Fanatics, and FanDuel charge flat per-contract fees that look small but compound fast. At $50,000, you’re paying over $1,000 in fees—regardless of odds or profit potential. These platforms work fine for casual $20 trades. At size, the math breaks down.

Large position fees

At $10,000 in positions, the fee models diverge dramatically. Polymarket US charges $10. Robinhood, DraftKings, and FanDuel charge $256 with no volume discount. PredictIt takes $958, nearly a third of your $3,158 profit.

*Fee estimated; platform may not offer this market. †2% of profit, charged at withdrawal. ‡10% profit fee + 5% withdrawal fee

What the four tables show:

• Polymarket US wins at every probability and position size
• At $10,000 on a 76% favorite, flat-fee platforms charge $263 vs Polymarket’s $10
• MetaMask’s 4% flat fee ($400) now costs more than the per-contract platforms
• PredictIt takes $958 from a $3,157 profit—30% of your winnings gone

Real-world case study: The 50% maker fee

In June 2025, Rufus Peabody documented what happens when Kalshi’s fee structure meets low-probability markets. He placed $1,000 in maker orders at $0.02 per contract (+4900 odds) on Tommy Fleetwood to win a golf tournament.

The result: 12.52% average fees across his fills, with some individual contract matches incurring 50% fees.

Here’s why: Kalshi’s maker fee at the time was a flat $0.0025 per contract, but the exchange rounded up at the individual trade level. When Peabody was matched on a single contract at $0.02, his fee rounded from $0.0025 to $0.01—a 50% charge on that fill. On contracts matched in larger batches, the fee averaged out to something reasonable. But on fills that trickled in one contract at a time, the rounding multiplied his costs.

Kalshi changed its maker fee formula in July 2025 to scale with probability (similar to its taker fees), but the lesson still holds: fee structures that look clean in documentation can produce ugly results in practice, especially on longshots and illiquid markets.

The turnover multiplier

The scenarios above show single-trade costs. Active traders also face a compounding problem: fees multiply with every position entered and exited.

Consider the following two traders with $10,000 accounts who each generate 10x annual turnover ($100,000 in trades):

Trader A (Polymarket US):

• Per-trade fee: 0.10%
• Annual fee drag: $100
• Fees as % of account: 1%

Trader B (Kalshi, averaging across odds):

• Per-trade fee: ~1.2% (based on 2025 revenue data)
• Annual fee drag: $1,200
• Fees as % of account: 12%

Both traders need positive expected value to profit, but Trader B needs a 12-point edge just to cover fees before showing any return. At higher turnover rates, this gap widens further.

Early exits double your fee exposure

Selling a position before resolution doubles your fee exposure on most platforms. You pay to enter, then pay again to exit.

This creates a strategic tension: holding through resolution avoids the second fee but locks you into a position that might move against you. On platforms with round-trip costs above 3-4%, actively managing positions becomes expensive enough that many traders default to buy-and-hold strategies even when selling would be optimal.

Polymarket’s zero-fee model (on most markets) sidesteps this entirely—you can adjust positions freely without fee penalty. This explains why Polymarket sees higher per-market trading velocity despite lower overall volume than Kalshi.

Where do fees go from here

The prediction market fee wars are just starting.

Polymarket’s 0.10% taker fee forced Kalshi to introduce a new rebate program, VIP for high-limit traders and promotional waivers—a response that would have been unthinkable when Kalshi was the only US exchange. As more exchanges enter and compete for volume, downward pressure on fees will intensify.

What happens next depends on which business model wins. If Polymarket proves that near-zero trading fees can sustain a profitable exchange, the industry will follow. If Kalshi’s formula-based approach becomes the regulatory template, competitors will have to justify why they charge more or less. Either way, the current 1,000x spread between cheapest and most expensive platforms won’t survive a maturing market.