Kalshi-Polymarket Cross-Platform Arbitrage: My Exact Method and Returns

Kalshi-Polymarket arbitrage is one of the few genuinely repeatable edges available to retail traders in prediction markets — when you know exactly what you're looking for and how to execute against it. Because Kalshi and Polymarket price overlapping events independently (different user bases, different liquidity, different settlement rules), the same underlying question can trade at meaningfully different implied probabilities across both venues at the same moment. This guide walks through the exact framework for identifying, sizing, and executing these gaps, along with the structural risks that erase most people's theoretical edge before it ever gets realized.

What Kalshi Polymarket Arbitrage Actually Is (And Isn't)

True arbitrage means locking in a spread with zero directional risk — buying "yes" on one platform and "no" on the equivalent market on the other, where the combined cost is less than $1.00 in guaranteed payout. In practice, pure riskless arb across Kalshi and Polymarket is rare and fleeting. What you're actually hunting for most of the time is probability divergence: two markets tracking the same real-world event where the implied probabilities diverge by more than transaction costs and settlement risk justify.

This distinction matters because it changes how you should think about position sizing. A genuine locked arb (rare) deserves near-max size within your risk limits. A probability-divergence trade (common) is a directional bet with a statistical edge, not a locked profit, and should be sized like any other structured position — informed by your read on which platform is mispricing, not just the spread itself.

Finding the Cross Platform Arbitrage Setup

The mechanical part of cross platform arbitrage is straightforward once you build the habit: you need markets on both platforms that resolve on the same underlying event with the same resolution criteria. The hard part is verifying resolution equivalence — Kalshi and Polymarket often use subtly different source data, cutoff times, or tie-breaking rules for what look like identical questions.

• Match on resolution source, not just headline. A Fed rate decision market on Kalshi settling off FOMC statement language can diverge from a Polymarket contract settling off a specific data provider's feed.
• Check settlement timing. If one platform settles same-day and the other has a 24-48 hour dispute window, you're carrying extra risk that isn't priced into the spread you're seeing.
• Confirm contract size and fee structure. Kalshi's per-contract fee schedule and Polymarket's gas/slippage costs both eat into a spread that looks wider on the screen than it is after execution.

Once you've confirmed true equivalence, the actual scan is simple: pull implied probability on both sides, net out fees, and flag anything where the gap exceeds your minimum threshold (most experienced traders use somewhere between 3-6 cents as a floor, depending on liquidity and time-to-resolution). If you're doing this manually across dozens of markets a day, you will miss most of the good windows — this is exactly the kind of repetitive, data-heavy screening that PillarLab AI is built to automate, since it already pulls live order book data from both venues.

Sizing a Prediction Market Arb Position Correctly

Position sizing on a prediction market arb trade should account for three separate risk layers, not just the headline spread:

• Execution risk. The spread you see when you scan is not the spread you get when you fill. Thin order books on either side mean your entry moves the price against you mid-execution, especially on Polymarket where depth can be shallow outside major markets.
• Correlation risk. If you're not holding a perfectly matched pair (same resolution source, same cutoff), you're exposed to a scenario where one platform resolves "yes" and the other resolves "no" on a technicality — turning your arb into a double loss instead of a locked gain.
• Capital lockup. Prediction markets tie up capital until resolution, sometimes weeks or months out. A 4-cent spread that takes six weeks to resolve is a very different risk-adjusted trade than the same spread resolving in six days.

A disciplined approach caps any single arb pair at a small percentage of total bankroll (many pros stay under 5-8% per pair) regardless of how attractive the spread looks, precisely because the correlation risk above is hard to fully eliminate even with careful matching.

Execution Mechanics That Determine Whether the Edge Survives

The gap between a spread you identify and a spread you actually capture comes down to execution discipline:

• Leg in simultaneously where possible. Placing both sides in sequence exposes you to the first fill moving the market before your second leg executes, especially in low-liquidity contracts.
• Use limit orders, not market orders. Market orders on either platform during a thin book can consume enough depth to erase the entire spread you were trying to capture.
• Track your realized fill price against your scanned price. If your realized spread consistently comes in 1-2 cents worse than your scan showed, your minimum threshold needs to move up, not your process.

Many traders build a running log of scanned-versus-realized spreads by market category (politics, macro, sports) because slippage patterns differ meaningfully by category — sports markets on Polymarket, for instance, tend to have thinner books than flagship political markets, which changes your effective execution cost even when the headline spread looks similar. This is the same category-level pattern recognition covered in the Kalshi vs Polymarket comparison, which is worth reading before you commit real size to cross-platform positions.

Risk Controls Specific to Prediction Market Arb

Because this strategy depends on two separate platforms with two separate rulebooks, the risk controls differ from single-platform trading:

• Resolution criteria audits before entry, not after. Read both platforms' rulebook language for the specific market, not just the market title, every single time — even for market types you've traded before, since rules get amended.
• Platform-level exposure limits. If either Kalshi or Polymarket experiences a liquidity event, regulatory action, or withdrawal delay, capital parked there is temporarily inaccessible regardless of how "locked" your arb looked on paper.
• Time-decay awareness. Spreads that look attractive weeks before resolution can compress or widen unpredictably as new information arrives — don't assume today's gap holds until settlement.

If you're building out a full cross-platform workflow, it's worth comparing your current toolset against what's covered in the best prediction apps for Kalshi and Polymarket roundup — most manual scanning setups underperform a structured, API-driven process once you're tracking more than a handful of markets at once.

How PillarLab AI Fits Into This

PillarLab AI was built around the exact problem cross-platform arbitrage exposes: manual scanning doesn't scale, and the markets that matter move faster than a spreadsheet can track them. The tool runs a structured 9-pillar analysis on any market you feed it, pulling real-time data directly from Kalshi and Polymarket order books rather than relying on delayed or cached snapshots. That matters specifically for arb work, because the entire strategy lives or dies on catching spreads before they compress.

The 9-pillar framework covers the dimensions that matter for this kind of trade: current implied probability on each platform, liquidity depth, resolution-criteria alignment, time-to-resolution, historical volatility of the specific market category, and several other structured inputs that would otherwise require manually cross-referencing two separate rulebooks and two separate order books every time you want to evaluate a pair. Instead of guessing whether a 5-cent gap is a real opportunity or a stale quote, you get a structured, actionable output that tells you where the edge actually sits and how confident the underlying data supports it.

For traders running cross platform arbitrage as a repeatable process rather than an occasional lucky find, this is the difference between reacting to spreads after they've already narrowed and catching them while they're still live. It won't do the trade for you, and it won't manufacture edge that isn't there — but it removes the manual bottleneck that causes most people to miss the good windows entirely.

Frequently Asked Questions

Is Kalshi-Polymarket arbitrage legal?

Yes. Trading matched positions across two independently regulated platforms is legal; you're simply taking positions on both venues based on public pricing, not exploiting any system vulnerability.

How much capital do you need to start cross platform arbitrage?

There's no fixed minimum, but smaller accounts struggle to overcome per-contract fees and gas costs. Most practical setups start with enough capital to diversify across several pairs, not just one.

Why do Kalshi and Polymarket prices diverge on the same event?

Different user bases, liquidity levels, and information flow speeds mean each platform prices probability independently, creating temporary gaps until enough traders on either side close them.

What's the biggest risk in prediction market arb?

Resolution mismatch — two markets that look identical but settle on different criteria, timing, or data sources, turning what looked like a locked position into a directional double loss.

Can this strategy be automated?

Scanning and data aggregation can be automated effectively; execution and risk judgment still benefit from human oversight, especially around resolution-criteria verification.

If you want to stop manually cross-referencing order books and rulebooks every time a spread catches your eye, start free with 10 credits and run a full 9-pillar analysis on the next Kalshi-Polymarket pair you're considering — you'll see the liquidity, resolution, and probability data laid out in one structured view instead of two open browser tabs.