Market Maker Behavior in Event Markets

Market maker behavior in event markets is not the same animal you'd study on the NYSE or in FX. On Kalshi and Polymarket, the "market makers" are often a mix of automated bots, sharp individual traders, and a handful of quant shops running spread-capture strategies against retail flow. Understanding how these participants set quotes, manage inventory, and react to news is the difference between reading a price as truth and reading it as a signal shaped by someone else's risk book. If you trade event contracts — election markets, Fed decisions, sports outcomes — you need to know whose liquidity you're trading against, because the spread you see is a function of their incentives, not the market's "true" probability.

How Market-Maker Quoting Works in Prediction Markets

In a traditional order book, a market maker posts two-sided quotes — a bid and an ask — and earns the spread by turning over inventory quickly. Event markets complicate this because contracts settle at exactly $0 or $1, so a market maker's inventory risk isn't symmetric like it is with a stock that drifts continuously. A market maker holding YES contracts near expiry on a binary event isn't managing gradual price decay; they're managing a step function. This is why you'll often see spreads widen sharply as an event approaches resolution — the market maker isn't being lazy, they're pricing in the jump risk of a single information event (a vote count, a game clock hitting zero, a data release) flipping the contract's value discontinuously.

On Kalshi specifically, registered market makers get fee rebates and obligations to maintain two-sided quotes within a certain width during active hours. On Polymarket, liquidity is largely permissionless — anyone can post limit orders, and much of the "market making" is done by automated bots running fairly simple inventory-neutral strategies. If you're trying to understand mechanics at a platform level before you trade either, How Kalshi Works is worth reading first, since the rebate structure changes how aggressively makers quote around news.

Spread Behavior and Quant Signals Around News Events

Watch what happens to the bid-ask spread in the 10 minutes before a scheduled catalyst — a jobs report, a debate, a game's final minutes. Spreads widen because market makers are quant-driven: they're running models that estimate the probability distribution of the next tick and they pull quotes or widen them when that distribution gets fat-tailed. This is not noise. A widening spread is itself information — it tells you the makers think uncertainty just increased, independent of where the midpoint sits. If you see a market's midpoint hold steady but the spread balloon from 2 cents to 8 cents, that's the maker telling you they expect a jump, not a drift.

You can use this mechanically. Spread width as a standalone feature is a decent proxy for implied volatility in a market that doesn't have an options chain to back out volatility directly. Traders comparing venues for this reason should look at Kalshi vs Polymarket 2026 — spread dynamics differ meaningfully between the two because of different maker incentive structures and different retail flow composition.

Order Flow Toxicity and Adverse Selection

Market makers price for adverse selection — the risk that the person trading against them knows something they don't. In event markets this shows up starkly around insider-adjacent flow: a sudden burst of one-directional volume on a corporate earnings market or a niche political market can trigger a maker to pull quotes entirely rather than widen them, because the marginal cost of being picked off by an informed trader exceeds any spread they could charge. You'll notice this as a market that simply goes quiet — no fills, stale-looking quotes — right before a large move. That silence is a market maker recognizing toxic flow and stepping back, not indifference.

For you as a trader, this means liquidity is not a constant. It evaporates precisely when you need it most, right before a resolving event, because that's exactly when adverse selection risk peaks for anyone quoting two-sided markets. Treat displayed depth as a snapshot, not a guarantee — a $500 order that looks fillable at the mid can move the book by three or four cents once you actually submit it, because the resting size includes maker orders that get pulled the instant flow looks informed.

Inventory Management and Skewed Quotes

A market maker who has accumulated a large YES position will skew their quotes — lowering both bid and ask — to encourage sell-side flow and offload risk, even if their own probability estimate hasn't changed. This is standard inventory-risk management, and it means the quoted price at any given moment reflects the maker's position, not solely the market's collective belief. If you're trying to extract the "true" implied probability from a quote, you need to account for this skew, especially in thinner markets where a single large maker dominates the book.

This is one of the more counterintuitive parts of reading event-market odds correctly: a moving price doesn't always mean new information arrived. Sometimes it just means a maker got heavy on one side and is repricing to rebalance. For a framework on separating signal from inventory noise when reading quotes, see How to Read Prediction Market Odds — it goes through the mechanical adjustments worth making before you treat a quote as a probability.

Bot-Driven Liquidity and Automated Market-Making Strategies

Both Kalshi and Polymarket have seen a rise in automated market-making bots that run relatively simple strategies: quote a fixed spread around a reference price (often derived from a correlated market — a sportsbook line, a polling average, an options-implied probability), adjust size based on realized fill rate, and pull quotes on volatility spikes detected via a simple threshold on recent trade velocity. These bots are not sophisticated in the options-market-maker sense — most aren't running full stochastic volatility models — but they are fast, and they dominate liquidity provision in high-volume markets like sports and macro data releases.

The quant angle worth internalizing: because these bots reference external correlated prices, event markets often lag or lead the reference market by seconds to minutes depending on bot latency and platform API rate limits. In sports markets specifically, this lag creates short windows where a live line move hasn't fully propagated into the event-market quote yet. Traders building automated strategies around this should look at how Best AI for Sports Betting tools handle latency and cross-reference pricing, since the same lag dynamics apply whether you're trading manually or systematically.

Cross-Platform Arbitrage and Maker Response

When the same event trades on both Kalshi and Polymarket, price discrepancies attract arbitrage capital, and market makers on each platform respond by tightening or loosening quotes based on how much of that cross-platform flow they're absorbing. A maker who notices persistent one-way flow coming from arbitrageurs will adjust their model to assume the "true" price sits closer to the other venue's quote, effectively importing information from the competing platform. This is one reason prices across Kalshi and Polymarket converge faster on liquid, high-attention markets than on niche ones — the maker incentive to close the gap scales with volume and rebate value.

For a side-by-side breakdown of where these platforms currently diverge on fees, liquidity depth, and settlement mechanics, Best Prediction Market 2026 lays out the practical differences that affect how aggressively makers close arbitrage gaps on each.

How PillarLab AI Fits Into This

Reading market-maker behavior manually — spread width, quote skew, cross-platform lag, order-flow toxicity — is a lot to track across dozens of live Kalshi and Polymarket contracts simultaneously. PillarLab AI is built to do this systematically. It runs a structured 9-pillar analysis on every market you query, pulling real-time data directly from Kalshi and Polymarket order books and combining it with external reference signals — polling data, sportsbook lines, macro releases — the same inputs the bots described above are quoting against.

The pillars break down liquidity depth, spread behavior, momentum, cross-platform pricing gaps, and sentiment into distinct scored components, so instead of eyeballing a widening spread and guessing whether it means volatility or inventory skew, you get a direct read on which pillar is driving the move. When PillarLab flags a divergence between a market's quoted price and its underlying pillar scores, that's often exactly the kind of maker-driven mispricing — inventory skew, stale quotes, adverse-selection pullback — described above, surfaced as an actionable edge rather than something you have to reconstruct by hand from raw order book snapshots.

Because it watches Kalshi and Polymarket concurrently, PillarLab also catches the cross-platform lag windows where one venue's makers haven't yet repriced against the other. For traders who want a systematic view of market-maker positioning without manually monitoring order books across two platforms, PillarLab AI is the tool built specifically for that job.

Frequently Asked Questions

Do Kalshi and Polymarket have official market makers?

Kalshi has a registered market-maker program with fee rebates and quoting obligations. Polymarket's liquidity is largely permissionless, provided by independent bots and traders rather than a formal designated program.

Why do spreads widen right before an event resolves?

Market makers widen spreads to price in jump risk — the chance a single resolving event flips the contract's value discontinuously from near-$0.50 to $0 or $1, increasing their inventory risk near expiry.

Can a moving price mean nothing changed about the actual odds?

Yes. Market makers skew quotes to manage inventory, so a price move can reflect a maker offloading a large position rather than new information changing the true probability.

Are prediction-market bots as sophisticated as options market makers?

Generally no. Most run simpler strategies referencing correlated external prices like sportsbook lines or polling averages, with basic volatility-triggered quote withdrawal rather than full stochastic models.

How can I track market-maker behavior without watching order books manually?

Tools like PillarLab AI score liquidity, spread, and cross-platform pricing gaps automatically across Kalshi and Polymarket, surfacing maker-driven mispricings as a structured edge signal.

Start free with 10 credits