How prediction market fees work
What is a prediction market fee?
A prediction market fee is the amount the operator or liquidity provider deducts from a filled order in exchange for matching, settlement, and market-making services. On a central limit order book venue the fee is charged as a percentage of notional. On an automated market maker venue the fee is embedded in the price as a spread. Some venues charge only takers; others charge both sides. The fee is separate from the payoff of the contract itself.
Why do prediction market fees change with the probability of the outcome?
Fees on an AMM prediction market typically peak at 50/50 markets and drop toward the extremes. The underlying math is fee = notional × base_rate × p × (1 − p). At 50/50 the multiplier is 0.25 (its maximum). At 90/10 it is 0.09. At 99/1 it is 0.0099. The formula captures the intuition that uncertain markets need more compensation for liquidity risk while extreme markets do not. This is the industry-standard shape for symmetric prediction market fees.
Why do fees differ between users on the same trade?
Almost every trading venue offers tier-based fee discounts. On PredictAsiaX two independent programs stack compound: the VIP tier is based on 30-day trading volume, and the PAX Holder tier is based on how much PAX token you hold and how long you have held it. A user with Diamond VIP and Founding PAX pays 70% less than a base-tier user on the same trade. This is disclosed in the order preview before every trade.
How does an AMM prediction market actually charge the fee?
The house edge is built into the LMSR pricing curve. When you buy 100 shares at $0.55, the market's fair value estimate might be $0.545 — the extra $0.005 per share is the fee. You pay it up front as part of the price. At settlement the winning contract still pays $1.00 face value; there is no additional fee at settlement. Because the fee is priced in, users see it as a widening of the spread rather than a separate line item — but modern venues (including PredictAsiaX) surface the effective fee in the order confirmation UI so it is not hidden.
How does a CLOB prediction market split its taker fee?
A CLOB market makes explicit which counterparties contributed to a fill. PredictAsiaX splits the 30 basis point taker fee across five actors: the builder that acquired the user (acquisition_builder), the builder that executed the trade (execution_builder), the platform operator (operator), the market creator (market_creator), the liquidity provider (lp), and a residual platform_net bucket. Bps are calibrated so builders and creators earn attribution rewards, LPs are compensated for depth, and the operator covers infrastructure. Progressive tier bumps shift bps toward builders at higher builder tiers.
What is the difference between a taker fee and a maker rebate?
A taker order is one that immediately crosses the current best bid or offer and takes liquidity off the book. A maker order rests on the book and provides liquidity until a later taker matches against it. Because makers create market depth, most venues either waive their fee or pay them a small rebate. PredictAsiaX pays makers 2 basis points and charges takers 30 basis points. This is a common structure across derivatives and prediction market exchanges.
Are fees the same across all products on a platform?
Usually not. Fees are calibrated per venue and per product type based on liquidity depth, competition, and target user profile. On PredictAsiaX the AMM fast market fee (0 to 1.25% depending on price) is distinct from the CLOB taker fee (30 basis points flat). Some venues also carve out specific market categories (geopolitical events, weather, science) for a lower rate to encourage information aggregation on socially valuable questions.
How should I compare prediction market fees to traditional broker or exchange fees?
Traditional securities broker commissions have collapsed toward zero in most markets, but the true cost of a trade is typically the bid-ask spread paid to the market maker, not the visible commission. In equities this spread runs 1 to 30 basis points on liquid names. In options and futures it runs 5 to 100 basis points. Prediction market fees at 30 basis points sit within this range for CLOB venues and are lower than most sports betting exchanges (which typically charge 200 to 500 basis points of gross winnings).
Do fees get charged on losing trades?
Yes. Fees are charged at fill time and are independent of the eventual settlement outcome. A user who buys a losing position pays the fee at purchase and loses the entire cost of shares at settlement. A user who buys a winning position pays the same fee at purchase and receives the face value at settlement. Fees are never refunded based on outcome — they compensate for the execution service, not the correctness of the prediction.
How is the fee formula versioned and disclosed?
Sound practice is to version-control the fee formula and stamp every trade with the version in effect at fill time. PredictAsiaX uses semantic versioning (currently v2.0.0), announces any planned change at least 30 days in advance, and preserves each historical version so that past trades remain governed by the version under which they were executed. A user disputing a fee can request the exact version and inputs used to compute their specific charge.
Can I model my expected fee cost before placing a large order?
Yes. Most venues publish a pre-trade fee estimator endpoint. PredictAsiaX exposes GET /api/fees/quote-live for session-authed clients (session cookie) and GET /api/v1/fees/estimate for API-key clients (builders and quant systems). Both return the identical 5-actor breakdown that will be applied at fill time. Institutional traders should hit the endpoint before each order to size positions against the effective bps for their tier.
Where is my fee history recorded?
Every fee event is written to an append-only financial ledger with an idempotency key of the form fee_split:{trade_id}:{event_type}. The ledger is immutable — corrections happen only via compensating entries with a matching trade_id reference. Users can query their own history via the personal fee dashboard at /settings/fees or export it as CSV. The ledger is designed to be reconcilable end-to-end against on-chain payout transactions.