Market Making on Prediction Markets: A Step-by-Step Deep Dive
Market making on prediction markets means simultaneously posting buy and sell orders on both sides of a binary outcome, capturing the spread between them as profit while providing liquidity to other traders. Done well, it's one of the most consistent — and underused — edges in the prediction market space. This guide walks you through exactly how it works, what the risks look like, and how to build a repeatable process from scratch.
What Is Market Making and Why Does It Matter?
In any market, someone has to be willing to buy and sell at any given moment. That's the market maker — the participant who posts resting limit orders on both sides of the order book and profits from the difference between the bid and ask price.
On traditional exchanges, market making is dominated by high-frequency trading firms with millions in infrastructure. But prediction markets — platforms like Polymarket, Kalshi, and Manifold — are still inefficient enough that individual traders and small teams can compete profitably.
Why does market making matter for the ecosystem? Without it, prediction markets suffer from:
- Wide spreads that make entry and exit expensive for directional traders
- Low depth, meaning large orders move prices dramatically
- Stale prices that don't reflect current information quickly
Market makers solve all three problems simultaneously. In exchange, the platform and its users reward them with the spread — the tiny difference between what buyers pay and what sellers receive.
According to a 2023 analysis of Polymarket order books, average bid-ask spreads on active political markets ranged from 2 to 8 cents per share, with less liquid markets showing spreads above 15 cents. That's a significant edge for a disciplined market maker.
Understanding the Core Mechanics
Before placing a single order, you need to understand the fundamental math.
The Bid-Ask Spread
Imagine a market asking "Will the Fed cut rates in September?" The current fair probability might be 52%. As a market maker, you might post:
- Bid: 50¢ (you'll buy YES shares at $0.50)
- Ask: 54¢ (you'll sell YES shares at $0.54)
If both orders get filled, you've earned 4 cents per share with zero directional exposure — as long as you've correctly estimated the fair value in between.
Inventory Risk
The biggest operational risk for market makers isn't losing the spread — it's accumulating one-sided inventory. If a flood of informed traders hits your ask (buying YES from you), you're suddenly short YES shares without intending to be. Your position is now directional, and if the probability moves against you, those losses can dwarf your spread income.
This is why inventory management is central to every professional market making operation.
The Role of Fair Value
Everything in market making pivots on your fair value estimate — the probability you believe is closest to the truth. Your bid sits below it; your ask sits above it. The wider you post, the more spread you collect but the fewer fills you receive. The tighter you post, the more volume you do but the thinner your margin.
This tradeoff is the fundamental optimization problem of market making.
Step-by-Step: How to Start Market Making on Prediction Markets
Here's a numbered process you can follow from zero:
Choose your market carefully. Start with liquid, binary markets — political elections, economic indicators, or sports outcomes. Check out the beginner's guide to sports prediction markets for context on which markets tend to have volume.
Establish a fair value model. Before posting, you need an independent estimate of the true probability. This can be a simple weighted average of external forecasts, a Bayesian model, or a more sophisticated ML approach. Without a model, you're posting blind.
Calculate your spread. As a starting rule of thumb, set your spread at 2× your model uncertainty. If your model is confident to ±3%, post a 6% spread (bid at -3%, ask at +3% from fair value).
Post your initial orders. Use limit orders on both sides simultaneously. On platforms with an API, this can be automated — see how traders automate this process in the Fed Rate Decision Markets via API playbook.
Monitor inventory in real time. Track which side is getting filled. If you accumulate more than 20% of your intended max position on one side, begin leaning your quotes to rebalance.
Update quotes when your model updates. If new information hits — a speech, data release, news event — reprice your fair value immediately. Stale quotes are the market maker's version of leaving money on the table (or worse, getting picked off by someone who knows something you don't).
Review P&L by position, not just overall. Understand which markets are generating spread income vs. which are generating directional losses from inventory.
Scale gradually. Start with small position sizes ($50–$200 per market) until you understand how your quotes interact with the flow on each specific platform.
Key Strategies for Prediction Market Makers
Passive Market Making
The classic approach: post resting orders on both sides, let fills come to you, collect spread. Best suited for higher-volume markets where you'll get natural two-way flow. This works well on well-established political markets like U.S. elections — strategies explored in the automated presidential election trading guide are directly applicable here.
Adaptive Quote Shading
Instead of symmetric quotes around your fair value, shade your quotes based on order flow signals. If buy flow is dominant, move your ask down slightly (to sell more) and your bid down (to reduce long exposure). This turns your market making into a mild directional strategy layered on top of spread capture.
Cross-Market Hedging
Advanced market makers hedge positions across correlated markets. For example, if you're long YES on "Will Party A win the Senate?" you might short YES on "Will Party A win the House?" if those outcomes are historically correlated. This reduces inventory risk without having to cancel and repost orders. AI agents can accelerate this kind of analysis — the AI agents in prediction markets comparison covers several tools built for exactly this.
Volatility-Adjusted Spreads
Markets near resolution (e.g., an election two days away) have very different volatility profiles than markets with six months to run. Use wider spreads on long-dated markets (more uncertainty means more adverse selection risk) and tighter spreads close to resolution (uncertainty collapses, spreads can tighten).
Comparing Market Making Approaches
| Strategy | Complexity | Capital Required | Spread Collected | Inventory Risk |
|---|---|---|---|---|
| Passive symmetric quoting | Low | $100–$500 | Moderate | Medium |
| Adaptive quote shading | Medium | $500–$2,000 | Higher | Lower |
| Cross-market hedging | High | $2,000+ | High | Low |
| Automated algorithmic MM | High | $1,000+ | Highest | Varies |
| Manual single-market MM | Low | $50–$200 | Low | High |
The sweet spot for most individual traders is adaptive quote shading — it's learnable, doesn't require heavy infrastructure, and meaningfully reduces the inventory risk that kills passive strategies during news events.
Managing Risk as a Market Maker
Risk management isn't a feature you add to a market making strategy — it is the strategy.
Adverse Selection
Adverse selection happens when the traders filling your orders know more than you do. In prediction markets, this often looks like informed traders buying your ask right before a major announcement. To protect yourself:
- Monitor unusual volume spikes before news events
- Set hard position limits per market
- Use time-based cancellations (cancel all orders X minutes before a scheduled announcement)
Correlation Risk
Markets that seem uncorrelated can move together in crisis. A geopolitical shock might simultaneously move election markets, economic indicator markets, and energy price markets. Keep total capital exposure across correlated categories bounded. The geopolitical prediction markets quick reference guide is useful for understanding which categories tend to co-move.
Platform Risk
Prediction markets are still a relatively new asset class with real regulatory and operational risk. Diversify across platforms, understand the terms of each platform's resolution rules, and don't concentrate capital in markets with ambiguous resolution criteria.
Tools and Automation for Market Makers
Manual market making is viable at small scale, but to compete seriously — or to run across multiple markets simultaneously — you need automation.
Modern approaches include:
- API-based order management: Most major prediction platforms offer REST APIs for order placement and cancellation
- Fair value models: Rule-based models can be coded in Python in a weekend; ML models take longer but offer a significant edge
- Alerting systems: Set up notifications for large fills, inventory threshold breaches, and upcoming news events
- Backtesting frameworks: Before going live, test your quoting strategy on historical order book data
Platforms like PredictEngine offer tools purpose-built for active prediction market participants, including automated quoting assistance and portfolio tracking. For traders interested in how AI can turbocharge these workflows, reviewing the AI agents for prediction markets beginner's guide provides a solid foundation.
If you want a comparison of how different earnings-focused prediction markets work — relevant for understanding how volume and spreads behave around data events — the Tesla earnings predictions comparison and NVDA earnings predictions case study both offer concrete examples.
Realistic P&L Expectations
Let's set honest expectations. A well-run manual market making operation on prediction markets with $1,000 in capital across 5–10 active markets might generate:
- Spread income: 3–8% per month of deployed capital in ideal conditions
- Inventory losses: 1–4% per month in adverse selection and directional losses
- Net: roughly 2–5% per month on active capital — impressive but not guaranteed
The most consistent market makers are those who obsess over model quality (accurate fair values = fewer adverse selection hits) and risk discipline (hard position limits = no single blow-up event ruins months of spread income).
Frequently Asked Questions
What Is a Market Maker in Prediction Markets?
A market maker in prediction markets is a trader who simultaneously posts buy and sell limit orders on both sides of a market, earning the difference between the bid and ask price. They provide liquidity that makes it easier for other participants to enter and exit positions quickly.
How Much Capital Do I Need to Start Market Making on Prediction Markets?
You can start with as little as $100–$200 by focusing on one or two active markets and keeping position sizes small. Most serious market makers operate with $1,000 to $10,000 to spread across multiple markets and absorb inventory fluctuations without over-concentrating risk.
What Is the Biggest Risk in Prediction Market Making?
The biggest risk is adverse selection — informed traders filling your orders right before a major event moves the market against your position. Managing this requires pre-event order cancellations, tight position limits, and continuously updated fair value models.
Can I Automate Market Making on Prediction Markets?
Yes, and automation is increasingly essential for competitive market making. Most major platforms provide APIs that allow automated order placement, cancellation, and inventory monitoring. Tools available on PredictEngine can help streamline this process considerably.
How Do I Calculate the Right Spread to Post?
A practical starting rule is to set your spread at 2× your model's uncertainty. For example, if you estimate a 52% probability with ±3% confidence, post a bid at 49¢ and an ask at 55¢. Tighten your spread in high-volume markets and widen it around uncertain events or long-dated contracts.
Is Market Making on Prediction Markets Legal?
In most jurisdictions, trading on regulated prediction market platforms like Kalshi (which is CFTC-regulated) is fully legal. Platforms operating in decentralized or offshore contexts have different legal statuses depending on your country. Always verify the regulatory status of any platform before committing capital.
Start Market Making with the Right Tools
Market making on prediction markets is one of the most intellectually engaging and financially rewarding strategies available to retail traders today — but it rewards preparation and discipline above all else. The traders who succeed aren't necessarily the ones with the most capital; they're the ones with the most accurate fair value models and the most rigorous risk management habits.
If you're ready to move from theory to practice, PredictEngine gives you the infrastructure to quote markets, track inventory, and analyze performance — all in one place. Whether you're running a simple passive strategy on political markets or building a fully automated multi-market system, it's the platform designed for serious prediction market participants. Start your free account today and put your edge to work.
Run your own quant desk
Felix lets you research, backtest and run automated trading strategies, with risk controls built in.
Try Felix