Polymarket arbitrage means exploiting price disagreements — either within Polymarket itself or between Polymarket and other prediction market platforms like Kalshi. When two prices on the same outcome don’t agree, buying the cheap side and selling (or shorting) the expensive side locks in a risk-free profit.

This guide covers how arbitrage actually works on Polymarket in 2026, the types of opportunities that still exist, realistic profit expectations, and ranked reviews of the best arbitrage bots available.

How Arbitrage Works on Polymarket

Arbitrage on Polymarket comes in three forms:

YES/NO Mispricing (Intra-Market)

Every Polymarket binary market has YES and NO shares that should sum to $1.00. When they don’t — say YES is $0.52 and NO is $0.46 — buying both locks in a guaranteed $0.02 profit per share regardless of the outcome. These mispricings are rare on liquid markets but appear regularly on thinner books, especially during high-volatility events.

Correlated Market Drift (Intra-Platform)

Polymarket often lists related markets that should be mathematically linked. For example, if there’s a “Will the Fed cut rates in June?” market and a “Will the Fed cut rates before August?” market, the second should always be priced higher than the first. When they drift apart — perhaps because one market attracts heavy one-sided action — that drift is an arb opportunity.

Cross-Platform Arbitrage

The widest and most consistent arb opportunities in 2026 are between platforms. An event priced at $0.55 on Polymarket might trade at $0.60 on Kalshi. Buying YES on Polymarket and effectively shorting the equivalent on Kalshi captures that $0.05 spread. Cross-platform arb involves more complexity (different settlement mechanics, separate capital pools, timing differences) but offers spreads that are 2–5x wider than intra-Polymarket opportunities.

For the technical infrastructure behind cross-platform arb, see the unified API comparison and the cross-market arbitrage guide.

Is Polymarket Arbitrage Still Profitable in 2026?

Yes, but the landscape has changed significantly since 2024.

What’s tightened. Intra-Polymarket YES/NO mispricings that used to persist for minutes now close in seconds. Bot competition has intensified — dozens of arb bots monitor the same order book, and the fastest bot captures the opportunity while everyone else gets nothing. The average intra-market spread has compressed from 3–5 cents in early 2025 to 1–2 cents in March 2026.

What still works. Cross-platform arb remains viable because capital is fragmented across platforms with different settlement mechanisms. A $0.04–0.08 spread between Polymarket and Kalshi on the same event can persist for hours when liquidity is thin on one side. Correlated-market drift also persists because Polymarket’s market creation is permissionless — anyone can create markets, and related markets don’t always attract the same trading population.

Realistic numbers. Serious arb traders with $5,000–20,000 in capital, using a fast cross-platform bot, report net monthly returns of 2–5% after fees. Intra-market-only strategies typically return 0.5–2% monthly with higher variance. Below $2,000 in capital, subscription costs for a decent arb bot ($80–400/month) eat most of the profit.

The fee math matters. Polymarket charges taker fees, Polygon has gas costs, and cross-platform arb ties up capital on both sides. A 3-cent spread might look profitable, but after 1 cent in Polymarket fees, 0.3 cents in gas, and potential slippage on one leg, the net capture is under 2 cents. Bots that don’t do fee-adjusted scanning will report profits that evaporate in practice.

The execution, however, is anything but simple. Spreads are thin, competition from other bots is fierce, and a slow arb scanner consistently arrives after the opportunity has already closed. Any tool you use needs to be fast, accurate at identifying genuine mispricings (not just noise), and reliable enough to execute both legs of a trade before the spread disappears.

An honest caveat before the tools. There is no independently verifiable, turnkey paid arbitrage bot for Polymarket that we can recommend by name. The real market is open-source scanners, a free cross-platform calculator, and developer frameworks you assemble yourself. The sections below cover those verifiable tools — with their licenses, GitHub activity, and source links — and how to build your own executor on top of them.

The Real Polymarket Arbitrage Tools (2026)

ToolTypeLicense / StarsPlatformsPriceSource
EventArbCross-platform arb calculator (not a bot)Open-source, ~28★Polymarket, KalshiFreeeventarb.com · GitHub
polymarket-arbitrageOpen-source arb scannerNo license declared, ~165★PolymarketFreeGitHub
pmxtUnified prediction-market API (build-your-own)MIT, ~1,831★Polymarket, Kalshi, LimitlessFreeGitHub
OctoBot Prediction MarketOpen-source bot (arb module in development)GPL-3.0, ~88★PolymarketFreeReview · GitHub

GitHub star counts verified May 2026. None of these execute fee-adjusted, two-legged cross-platform arbitrage automatically out of the box: EventArb surfaces opportunities for manual execution, the scanners detect spreads, and pmxt is a library you build on.

For broader context on the prediction market bot ecosystem, see the overall bot rankings and the buyer’s guide.

What to Look for in an Arbitrage Bot

Not all arb bots are created equal. Before evaluating specific products, here are the criteria that matter most:

1. Detection speed and accuracy. The bot needs to identify genuine arbitrage opportunities quickly — ideally within seconds of a spread opening. False positives (flagging normal price movement as arb) waste capital and erode trust. The best bots use multiple confirmation signals before triggering a trade.

2. Execution reliability. Finding an arb is only half the job. The bot must execute both legs of the trade before the spread closes. This means tight integration with Polymarket’s CLOB API, proper order management, and graceful handling of partial fills or failed transactions. A bot that detects arbs but fails to execute them is worse than useless — it gives you false confidence.

3. Cross-platform reach. Intra-Polymarket arbs still exist, but the most consistent spreads in 2026 are between platforms. Bots that scan Polymarket, Kalshi, and other prediction markets simultaneously find more opportunities and wider spreads. If a bot only looks at Polymarket, it is competing in the most crowded lane.

4. Risk management. Good arb bots let you set maximum position sizes, stop trading when spreads narrow below a threshold, and alert you when one leg of a trade fails. Without these controls, a bot can accumulate large one-sided positions during volatile markets.

5. Fee awareness. Arbitrage profits are often small — sometimes just 1-3 cents per share. A bot that does not factor in Polymarket’s trading fees, gas costs, and potential slippage will report profitable arbs that actually lose money after costs. Fee-adjusted scanning is non-negotiable.

Choosing the Right Arbitrage Tool

Because no turnkey paid bot exists, your choice depends on how much you want to build:

  • Want to spot opportunities without coding? Use EventArb’s free calculator to see live cross-platform spreads between Polymarket and Kalshi, then place the trades manually.
  • Comfortable running open-source code? The polymarket-arbitrage scanner detects intra-Polymarket mispricings; you supply the execution and the wallet.
  • Building a real automated executor? pmxt gives you one unified API across Polymarket, Kalshi, and Limitless, so you write a single integration instead of one per platform.
  • Already running OctoBot? Its arbitrage module is under active development — usable for experimentation, not yet a finished product.

None of these hands you fee-adjusted, two-legged cross-platform execution out of the box. That part you build or do by hand.

The Tools in Detail

EventArb

EventArb (eventarb.com) is a free, web-based arbitrage calculator covering Polymarket, Kalshi, and other event-contract platforms. It is a calculator and scanner, not an automated bot — it surfaces price discrepancies across platforms so you can evaluate and execute them yourself. The underlying logic is open-source (github.com/akhan280/event-contract-arbitrage, ~28★, verified May 2026).

Because it never holds keys or executes trades, there is no custody risk; the trade-off is that you place every order manually and the spread can close before you do. For traders who want visibility into genuine cross-platform spreads without writing code, it is the most accessible starting point — and the honest replacement for the “cross-platform arb bot” category, which has no verifiable turnkey product.

polymarket-arbitrage (ImMike)

github.com/ImMike/polymarket-arbitrage is an open-source scanner focused on intra-Polymarket arbitrage — detecting YES/NO mispricings and correlated-market drift within Polymarket. At ~165 GitHub stars (verified May 2026) it is one of the more popular community arb tools, though it carries no declared open-source license, so review the terms before any commercial use.

It is a detection tool, not a managed product: you run it yourself, supply your own Polygon wallet and API credentials, and handle execution. Expect to read the source. For technical traders who believe intra-Polymarket spreads still offer enough opportunity, it is a concrete, free starting point — with none of the invented performance claims that surround paid “arb bots.”

pmxt

pmxt (github.com/pmxt-dev/pmxt, MIT, ~1,831★, actively maintained as of May 2026) is “CCXT for prediction markets” — a unified API that normalizes Polymarket, Kalshi, and Limitless behind one interface. It is not an arbitrage bot itself; it is the backbone you build one on.

Cross-platform arbitrage requires reading order books and placing orders on multiple venues, and pmxt removes the biggest chunk of that work — a single integration instead of one per platform. If you are serious about building your own fee-aware, two-legged executor, this is the foundation to start from. See the unified API comparison and the Polymarket API guide.

OctoBot Prediction Market

OctoBot Prediction Market (github.com/Drakkar-Software/OctoBot-Prediction-Market, GPL-3.0, ~88★) is a free, open-source, self-hosted Polymarket bot. Its live, flagship strategy is copy trading; an arbitrage module is under active development and marked as a work in progress in the repository, so treat it as experimental rather than a finished arb product.

The appeal is full self-custody — keys never leave your machine — plus a visual UI and a paper-trading mode for risk-free testing. If you are already running OctoBot or want an open-source base to extend, it is worth watching as the arbitrage module matures. See the full OctoBot review for deployment details.

How to Evaluate an Arbitrage Tool

Before committing real capital to any arb tool, run through this testing checklist:

  • Paper trade or dry-run for at least two weeks. OctoBot offers a paper-trading mode, EventArb lets you inspect spreads without trading, and the open-source scanners can be run in observe-only mode. Use whichever applies. Arb performance varies significantly with market conditions, and two weeks gives you a reasonable sample.
  • Verify fee-adjusted returns. Ask the bot to show net returns after Polymarket fees, gas costs, and any slippage. If the dashboard only shows gross arb spread captured, do the math yourself — you may find that many “profitable” arbs are break-even or negative after costs.
  • Test failure handling. Deliberately trigger edge cases: what happens when one leg of an arb fails? Does the bot unwind the other leg, hold, or do nothing? Poor failure handling is the single biggest risk in arb trading.
  • Check execution speed against competition. During your trial, track how many flagged opportunities are still available when the bot tries to execute. If fewer than 50% of arbs are capturable, the bot is too slow for the current competitive environment.
  • Review the support and update cadence. Polymarket’s API and fee structure change. A bot that was last updated three months ago may be using stale fee calculations or broken API endpoints. Check the changelog or commit history.
  • Compare against manual arb scanning. Spend a few hours manually scanning Polymarket for arbs alongside the bot. This gives you ground truth for whether the bot is finding opportunities you would miss, or just flagging the same obvious spreads you can see yourself.

For a deeper framework on evaluating prediction market agents, see the verification guide and the buyer’s guide.

Setup Guide: Getting Started with Arb Trading on Polymarket

If you have selected a bot, here is the general sequence to go from zero to live arb trading:

Step 1: Set up your Polymarket account and wallet. You need a funded Polymarket wallet with USDC on Polygon. If you are using cross-platform arb, you will also need funded accounts on the other platforms. See the Polymarket bot quickstart for wallet setup details.

Step 2: Generate API credentials. Polymarket’s CLOB API requires API keys for programmatic trading. Generate these through the Polymarket interface and store them securely. For guidance on API setup, see the Polymarket API guide.

Step 3: Connect the tool. For self-hosted scanners and bots (polymarket-arbitrage, OctoBot, or anything you build on the pmxt API), configure your API credentials in a local environment file. For the EventArb calculator, no key connection is needed — it only reads public prices. Never share API keys or commit them to version control.

Step 4: Configure risk parameters. Set your maximum position size per arb, minimum spread threshold (we recommend starting at 3 cents or higher to ensure fee-adjusted profitability), maximum concurrent positions, and daily loss limit. Start conservative — you can always loosen parameters after you have confidence in the bot’s behavior.

Step 5: Validate detection before risking capital. Run your scanner (or OctoBot’s paper-trading mode) for at least one to two weeks and confirm the spreads it flags are real and still profitable after Polymarket fees and gas before you execute live.

Step 6: Go live with small size. When you are ready to trade live, start with 10-20% of your intended capital. Monitor closely for the first few days, checking that fee-adjusted PnL matches what you saw in paper trading. Scale up gradually as you build confidence.

Step 7: Monitor and iterate. Arb opportunities shift as markets mature and competition changes. Review your bot’s performance weekly, adjust spread thresholds as needed, and stay current on Polymarket API changes that might affect execution.