Before committing capital to any strategy, the obvious question is: what can I actually expect to make? For prediction market arbitrage, the honest answer is: it depends on capital size, market conditions, and how selective you are about which signals you act on — and the numbers are real but modest, not the double-digit monthly returns some sites imply.
Typical spread sizes
Net arbitrage edges on Polymarket/Kalshi pairs — after both platforms' fees are already deducted — are typically well under 1% per position. The distribution is skewed:
- Most opportunities: comfortably under 1% net edge, close to the minimum worth trading
- High-activity events (elections, Fed decisions): occasionally 1.5%–3% net edge
- The median measured net edge, across live detections, is around 0.7%
Only opportunities above a minimum net-edge threshold are worth the two manual orders it takes to place them — the platform's default filters out the rest.
Frequency of opportunities
Across thousands of actively monitored markets, the number of genuinely tradeable signals per day varies with market conditions — some days bring several, some bring none. Not all of these can be filled at the quoted price due to thin order books on one leg or the other. The live numbers show the current measured rate rather than a marketing estimate.
Capital requirements
There's no minimum, but practical considerations suggest:
- Under $1,000: Transaction costs and small contract sizes limit scalability. Returns in absolute terms will be modest.
- $1,000–$10,000: The sweet spot for getting started. Enough to fill most opportunities at quoted prices without moving the market.
- $5,000: Larger capital runs into liquidity limits on thin markets. Position sizing must account for order book depth.
Realistic returns
Unlike a strategy that turns capital over monthly, an arbitrage position stays committed from the moment it's opened until the underlying market resolves — about 230 days at the median, not weeks. Assuming a $5,000 bankroll and a 0.7% net edge held to resolution:
- P&L on that position: $5,000 × 0.7% = $35, realised at resolution
- Annualised across resolutions: roughly 1% on deployed capital at the median (0.70% over about 230 days)
These are historical medians from live detection data, not a projection — see /proof for the current numbers. Returns scale with capital up to the liquidity ceiling of each market,
Factors that affect performance
Market activity: More news = more price divergences = more opportunities. Election cycles, major economic events, and breaking news all create spikes in arbitrage opportunities.
Selectivity: Acting only on the highest-edge signals, rather than everything that clears the minimum threshold, raises the average return per position at the cost of fewer trades.
Fee optimisation: Maker vs taker orders, tier discounts, and choosing which side to place on which platform can reduce fee drag.
Is it worth it?
For a market-neutral strategy — one that doesn't require predicting anything — a few percent a year with minimal directional risk is a real, if modest, return; being selective about which signals to act on pushes that meaningfully higher for the same capital. The main constraint is market liquidity, not strategy quality. Join the Arbitrage Agent waitlist to get access.
Related articles
- How Kalshi Fees Are Calculated: Formula, Maker Fees and Rounding
- Complete-Set Arbitrage on Polymarket vs Cross-Venue Arbitrage
- Prediction Market Arbitrage Strategy: The Complete 2026 Guide
- Best Polymarket Arbitrage Agent: Automated vs Manual Comparison
- 84% of Polymarket Traders Lose Money — Here's Why
- Is Prediction Market Arbitrage Drying Up? How Much Edge Is Left in 2026
Key terms: Prediction Market Arbitrage · Spread · CLOB (Central Limit Order Book)