Market Making Strategies: Spreads, Inventory and One-Sided Risk
Understand prediction-market market making with a worked spread example, three quoting strategies, and the risks of fees and one-sided fills.
Market making means offering prices at which other traders can buy or sell, rather than simply betting on one outcome. In an order-book prediction market, a maker typically posts a bid to buy and an ask to sell around an estimate of fair value.
The aim is to capture a spread. The risk is that only one side fills—or that it fills because someone else has better information. A four-cent quoted spread is not four cents of guaranteed profit.
Start with the contract, not the spread
Before choosing quotes, identify exactly what makes Yes win: the resolution source, deadline, time zone and treatment of edge cases. Polymarket’s resolution documentation identifies the source, end date and edge cases as parts of a market’s rules; winning tokens redeem for $1 and losing tokens become worthless.
Consider a hypothetical contract: “Will Agency X publish Report Y on its official website by 5 p.m. UTC on Friday?” Assume the full rules say a press interview does not count. You are pricing timely publication, not whether the report’s findings are likely to be announced somewhere else.
For that exact event, suppose your probability estimate is 50%. In a simplified $1-or-$0 contract, that implies a 50-cent expected payout before costs. It does not mean the market’s midpoint is necessarily correct. Polymarket’s market-making guidance explicitly distinguishes the book midpoint from the maker’s assessment of fair value.
One spread, two very different results
Assume you post a bid for 100 Yes contracts at 48¢ and an ask for 100 at 52¢. You already have sufficient cash and Yes inventory to support both orders. On Polymarket, buy orders require available funds and sell orders require the corresponding outcome tokens—not an unsupported promise to sell.
For illustration, assume a hypothetical fee of $0.50 per 100-contract fill. This is not a platform fee quote.
| If both orders fill completely | Calculation |
|---|---|
| Purchase cost | 100 × $0.48 = $48 |
| Sale proceeds | 100 × $0.52 = $52 |
| Gross spread captured | $52 − $48 = $4 |
| Assumed fees | $0.50 + $0.50 = $1 |
| Net result from those fills | $3 |
That result assumes both fills occur at the stated prices, with no additional costs. Your Yes inventory returns to its starting level; any risk from the inventory you already held remains.
Now suppose news makes timely publication less likely. Before you cancel your stale 48¢ bid, someone sells into it, filling all 100 contracts. Your 52¢ ask remains unfilled. The executable exit bid is now 36¢. If you can sell all 100 newly acquired contracts there:
100 × ($0.36 − $0.48) = −$12 before fees.
With the same assumed fees on purchase and exit, the loss is $13. If you instead hold those additional contracts and the market resolves No, their $48 purchase cost is lost, plus the purchase fee.
This illustrates adverse selection: someone trades against your quote when it has become favorable to them and unfavorable to you. The unfilled 52¢ ask does not compensate you.
Three market making strategies—and their limits
1. Fixed-spread quoting
Post a bid and ask at a chosen distance from your fair-value estimate, as in the 48¢/52¢ example.
This is easy to understand, but a fixed spread does not mean fixed risk. If your estimate changes, the quotes need to change too. A wider spread may reduce trading interest; a narrower one leaves less room for fees and unfavorable fills. Check market depth, not just the best displayed prices.
A limit order is not automatically a maker order: it can execute immediately against an existing quote. Polymarket’s post-only option rejects an order that would match immediately, ensuring it adds liquidity instead.
2. Inventory-aware quoting
Adjust prices and sizes according to the exposure you have accumulated.
If repeated bid fills leave you with more Yes contracts than intended, you might lower the bid, reduce its size, or stop buying. You might also lower the ask to make selling more attractive. These are exposure controls, not a claim that the event’s probability has fallen.
Polymarket’s official guidance recommends inventory-based quote adjustments and size limits. The important beginner distinction is between your probability estimate and the prices you are willing to offer given your existing risk.
3. Event-aware quoting
Reduce size, widen quotes or withdraw them around scheduled announcements and deadlines. Update or cancel quotes when material information arrives.
This addresses the stale-quote problem in the example. Polymarket’s market-making guidance recommends canceling stale orders and using expiring orders before known catalysts. Expiry helps manage exposure, but it cannot protect an order that fills before it expires.
Fees and rewards belong in the calculation
Do not assume providing liquidity is free. As of October 9, 2026, Kalshi’s fee guidance says some markets charge maker fees when resting orders execute, and that fees differ across markets. Check the applicable schedule before quoting.
Incentives are separate from spread capture. Polymarket describes liquidity rewards for qualifying resting orders and maker rebates for eligible executions. Neither should be treated as a guaranteed payment or protection against inventory losses.
Before placing a quote, be able to answer: What resolves the contract? What if only my buy fills? How much total exposure can all my open orders create? What will trigger cancellation?
If those answers are unclear, use the pre-trade reading checklist before risking money. This is education, not a recommendation to run a market-making strategy.