A market showing 99-cent Yes contracts and 1-cent No contracts may look like a solved question. It may also be a thin order book with one stale quote, a wide effective spread, and a settlement rule doing far more work than the headline. Market making on prediction markets sits at that uncomfortable intersection of forecasting, trading infrastructure, and contract interpretation.
For traders, a market maker is often just the counterparty on the other side of an order. For the person or firm providing quotes, the job is less glamorous: continuously offer to buy and sell contracts, manage the risk of being wrong, pay platform costs, and hope the spread is wider than the damage caused by better-informed traders. It is not free yield. It is a small business built beside a newswire.
What market making actually means
A prediction-market contract commonly pays $1 if a defined event occurs and $0 if it does not. A Yes contract trading at 62 cents implies a market probability near 62%, subject to fees, spreads, contract design, and the usual caveat that markets can be wrong with impressive confidence.
A market maker posts bids and asks rather than waiting to take whatever price appears. If they quote a bid of 60 cents and an ask of 64 cents for Yes, they are willing to buy at 60 and sell at 64. The 4-cent difference is the quoted spread. In theory, filling both sides creates a gain before fees and risk. In practice, orders rarely arrive with the symmetry of a textbook diagram.
On many platforms, a trader can create a complete set by buying one Yes and one No contract for a combined $1, or can use a corresponding conversion or redemption process. That relationship should limit obvious mispricing between outcomes. But the exact mechanics vary. Some venues use central limit order books; others use automated market makers; some offer platform incentives for liquidity; and some offer interfaces that make a thin market look busier than it is.
The underlying principle remains simple: liquidity providers make it easier for others to enter or exit. Their reward, if any, comes from the spread, rebates, incentives, or favorable inventory adjustments. Their risk comes from everything else.
Why the spread is not profit
The first mistake in evaluating market making on prediction markets is treating the visible spread as income. A 4-cent spread is not a 4-cent profit if one side gets filled moments before new information moves the fair price by 12 cents.
This is adverse selection. A trader who buys your Yes offer may know more than you do, react faster than you do, or simply have noticed a public development before your pricing model has. Election results, court orders, economic releases, injury reports, and official statements can all reprice an event contract very quickly. In a market tied to a scheduled data release, resting quotes are effectively invitations for faster participants to trade against stale prices.
The probability of adverse selection rises as a contract approaches resolution. It also rises when the market wording is ambiguous, because informed participants may have read the rulebook more carefully than the quote provider. The desk checks the facts - someone should also check the settlement source.
Market makers therefore widen spreads when uncertainty is high, reduce order size, cancel quotes around known news events, or avoid markets whose resolution criteria they cannot model confidently. These choices may make the market less pleasant for takers. They may also be the difference between a sustainable quote and a donation box for anyone with a fast browser tab.
Inventory is a directional position
A liquidity provider is rarely flat for long. Suppose buyers repeatedly lift the Yes ask at 64 cents. The market maker now holds a growing short Yes position, which is economically similar to a long No exposure. If later information raises the fair probability to 80%, the apparent spread collected on earlier fills can disappear rapidly.
Inventory management means adjusting quotes to attract trades that reduce that imbalance. A maker short Yes may lower its bid for Yes and raise its ask, while improving prices on No if the platform's mechanics support it. In active financial markets, firms hedge inventory in related instruments. Prediction markets often offer fewer clean hedges. A contract on inflation may relate to futures or economic data, but the payout condition, deadline, and resolution source may not match closely enough to eliminate risk.
That is why a market maker can be a forecaster whether it wants to be or not. Quoting both sides does not erase exposure when flow is one-sided.
Fees, rebates, and the real execution cost
Published fee schedules matter more than glossy claims about liquidity. A maker may pay trading fees, earn rebates, face withdrawal charges, or incur blockchain transaction costs on an on-chain venue. A taker may pay a per-contract fee, a percentage of profits, or a spread embedded in an automated pricing curve. The names differ; the economics do not become friendlier because the interface uses confetti.
Before assessing a strategy or a venue, calculate the all-in cost of a round trip. That includes the difference between the price shown and the price actually filled, platform trading fees, any required conversion costs, deposit and withdrawal charges, and the opportunity cost of capital tied up until settlement. If incentives are involved, check their terms, duration, eligibility rules, and whether they compensate for meaningful inventory risk or merely encourage quote spam.
Displayed depth deserves similar skepticism. Ten thousand contracts quoted at a price may be useful liquidity, or it may be cancelable liquidity that vanishes when a large order approaches. Order books change. Automated market makers can provide guaranteed executable prices, but price impact may become severe as trade size grows. Neither model is automatically superior. The relevant question is what a trader can execute, at what total cost, during the conditions that matter.
Contract rules can outweigh market skill
Prediction-market market makers price contracts, not broad narratives. The distinction is essential. “Will inflation fall?” is not a tradable proposition until the platform specifies the index, observation period, release source, threshold, timezone, revision policy, and what happens if the source is delayed or changed.
A contract may resolve on an initial government release rather than a later revised figure. It may use a named media outlet, a particular sports league's official record, or a platform-appointed determination process. It may contain language governing postponed events, candidate replacements, market disruption, or outcomes that are technically true but outside ordinary intent.
For a passive trader, missing those details can cause a bad trade. For a market maker quoting both sides all day, it can cause systematic mispricing. A seemingly minor ambiguity creates a valuable edge for participants who understand the rule better. That edge is not always clever forecasting. Sometimes it is just reading.
Market structure changes the job
Centralized, regulated event-contract exchanges may impose participant eligibility requirements, geographic restrictions, account verification, position limits, and specific fee schedules. Their contracts may have clearer formal oversight, but availability is not universal and product coverage can be narrow.
Crypto-native on-chain markets may offer different access patterns, custody arrangements, settlement mechanisms, and liquidity incentives. They can also introduce smart-contract risk, wallet and transaction costs, token volatility, oracle dependencies, and legal uncertainty that does not vanish because the interface loads quickly. A US reader should not infer legal availability from an app being technically reachable.
Play-money forecasting platforms are a separate category. They can be useful for learning calibration or testing a pricing process, but they do not reproduce the discipline of capital at risk, real withdrawal mechanics, or every feature of live order flow. Treating a game leaderboard as proof of a viable market-making operation would be ambitious. So would treating paper gains as rent money.
A practical framework before providing liquidity
Anyone considering liquidity provision should start with four questions: Can I legally access and use this venue from my jurisdiction? What exactly determines settlement? What is my worst-case exposure if only one side fills? And what are the documented all-in costs after fees and withdrawals?
Then test execution mechanics at small size. Observe how quickly orders fill, whether displayed depth persists, how cancellations work, and whether the platform's stated contract rules match the market interface. Keep a record of fills and inventory rather than relying on a chart that happens to look flattering after a quiet week.
Market making may suit participants with disciplined pricing, fast monitoring, enough capital to absorb uneven fills, and a genuine understanding of the venue's rules. It is a poorer fit for someone whose entire thesis is that the visible spread looks wide. On prediction markets, the spread is often widest precisely where the unanswered questions are most expensive.