Prediction market arbitrage scanners and bots all promise to find the same thing: the same event priced differently on two venues. What differs is how much of the work between a detected gap and a profitable trade the tool actually does. This checklist gives you eight questions to ask of any scanner, including ours, and a way to check each answer yourself.
1. How does it decide two markets are the same event?
Matching is the hardest part. Titles that look alike can resolve on different sources, deadlines or edge cases, and a mismatch turns an arbitrage into two directional bets. Ask whether the tool compares resolution rules or only titles, and whether you can see both rule texts from the opportunity. See why the same event has different prices and Resolution Risk.
2. Does it show the net edge, or only the gross spread?
Fees come out of the gap. Polymarket documents a category-based taker fee, fee = C × feeRate × p × (1 − p), and Kalshi publishes its own schedule, so the cost differs by market and price. A tool that shows only the gross spread overstates every opportunity. Worked numbers are in Prediction Market Arbitrage After Fees.
3. Which prices does it use?
Last-traded price, mid price and the price you would actually pay to buy are three different numbers. For an arbitrage, only the executable price on each leg counts. Ask which one the scanner displays, and whether it is the best ask for the side you would buy.
4. How old are the prices?
Spreads can close quickly. A scanner should show when each price was read, and a displayed opportunity is only a snapshot of that moment. Treat the timestamp as part of the opportunity, and re-check the live book before you act.
5. Does it check order book depth?
A price quoted for 5 contracts says nothing about 500. Ask whether the tool reads the size available at the quoted price, because depth, more than spread width, usually caps what you can deploy. See Prediction Market Liquidity.
6. What does it say about execution and leg risk?
If one leg fills and the other does not, you hold a directional position. Ask what the tool does about that, whether it executes at all or only alerts, and whether it can run in a simulated mode first so you can compare detections with the live books before risking money. See Two-Leg Execution.
7. Does it publish measured results?
Claims about profit or speed are easy to write and hard to verify. Prefer tools that publish what they measured, with the method, over ones that publish a headline return. Our own measured numbers are on /proof, and we do not promise any return.
8. Can you use it in your jurisdiction?
Both venues restrict access by country and the lists change. A scanner cannot change whether you may open accounts on both. Check your own country first; see Is Prediction Market Arbitrage Legal? and Can Europeans Trade Prediction Markets?.
A five-minute test
- Pick one opportunity from the tool.
- Open both markets and compare the displayed prices with the live order books.
- Read both resolution texts and decide whether they truly match.
- Recompute the net edge with each venue's own fee schedule and the size you would trade.
- If any step contradicts the tool, treat that as information about the tool.
Arbitrage Agent is built around these checks: matched pairs, net edge after fees and a simulated mode. See how it compares with other approaches or view the plans.
Related articles
- Best Polymarket Arbitrage Agent: Automated vs Manual Comparison
- How to Arbitrage Polymarket vs Kalshi
- Prediction Market Arbitrage After Fees: A Worked Example
Key terms: Prediction Market Arbitrage · Spread · CLOB