Prediction market fees, with the math

What does a $50 trade cost?

Take 100 contracts at 50 cents each. The position costs $50 before fees. These examples use published schedules checked on 24 September 2026. They exclude the spread, slippage and any deposit or withdrawal charges.

The last row isolates the frontend's fee. It is not a quote for the total Hyperliquid trading cost, and 0.1% is an example rather than a required rate. The same position size does not mean the venues offer the same market or execution.

Kalshi fees depend on the contract

Kalshi's general taker formula is M × 0.07 × contracts × price × (1 − price), with rounding defined in its schedule. M defaults to 1. Specific series have different multipliers, and some charge makers too. The $1.75 example above uses M = 1 and needs no rounding adjustment.

There is no settlement fee under the current general schedule. Check the official fee schedule for the series you plan to trade, rather than applying one percentage to every market.

Polymarket fees vary by category

Polymarket's formula is shares × fee rate × price × (1 − price). Its published sports rate is 0.05, giving $1.25 in this example. The crypto rate is 0.07, giving $1.75. Makers pay no trading fee, and geopolitical markets are fee-free under the current schedule.

Read the Polymarket fee rules and the individual market's parameters. “Polymarket is free” and “Polymarket charges 7%” both leave out information that changes the bill.

Hyperliquid has a separate frontend fee

Hyperliquid's outcome fee rules distinguish opening, closing and settlement. Its HIP-4 documentation currently describes zero protocol fees during initial testing. That is a temporary setting, not a promise of permanent free trading. Builder codes can still charge on eligible sells.

A frontend's builder fee is separate from protocol and market-deployer charges. See the HIP-4 specification and how Hyperliquid prediction markets work.

Start with the right fee

The protocol’s trading fee and your builder fee are different charges. A trader approves a maximum builder fee for the builder code. That approval is a limit, not proof that every trade or settlement produces a fee for your site.

For Floatout’s current outcome path, estimate builder earnings from eligible sell activity. Do not count settlement proceeds as builder revenue. Hyperliquid’s protocol fee rules describe closing and settlement charges, but protocol revenue is not automatically operator revenue.

Use eligible volume, not headline volume

Separate perps and outcomes in your model. Use the builder rate configured for each product and volume that actually bears that fee. Reconcile estimates to recorded builder-fee receipts once you have activity.

For a hypothetical $100,000 of eligible volume at a 0.1% builder fee, gross builder fees are $100. That is before Floatout’s share, partner commissions, trader kickbacks, and the costs of running your community. It is an example, not a revenue forecast or a recommended fee.

A larger rate can reduce a trader’s willingness to use the site. Compare the complete trading experience and costs, not just what you can charge.

What the operator retains

Retained revenue = gross builder fees − platform share − affiliate commissions − trader kickbacks.

Your operating result also subtracts promotion, moderation, support, and other business costs. Use the product-specific calculator with explicit assumptions. Avoid counting the same activity twice across products or confusing a share of builder fees with a share of all exchange fees.

On the free plan, builder fees go to Floatout. Activation applies to future eligible activity. Pre-activation fees are not an owner balance and are not promised as activation credit.

Decide when activation makes sense

Divide the activation price by a plausible monthly net operating surplus to estimate a simple payback period. If the surplus is zero or negative, there is no payback under those assumptions. A calculation with constant volume does not account for growth, churn, or unusual event schedules.

Compare plans using both the share difference and features you will actually use. An extra five percentage points requires $100,000 of cumulative gross builder fees to produce an extra $5,000, before other differences. That arithmetic alone does not value support or product features.

Record real activity through the pilot scorecard, then revisit the estimate. Never ask members to churn trades to improve the model.