Prediction markets are fragmented. The same outcome may trade across Kalshi, Polymarket, and other venues, each with separate liquidity, participants, fees, and settlement rules.
Equity markets also trade across multiple exchanges, but their infrastructure connects those venues through common clearing, consolidated market data, and rules designed to protect displayed quotations. Prediction markets do not yet have comparable market structure.
This fragmentation can create several potentially profitable opportunities.
Cross-venue arbitrage
The simplest case involves the same outcome trading at different prices on two venues.
Suppose:
- Kalshi: Candidate X wins — Buy Yes at 44¢
- Polymarket: Candidate X wins — Buy No at 52¢
Buying both contracts costs 96¢. If the contracts are truly equivalent, one side must settle at $1. The combined position therefore returns $1 regardless of the outcome, producing a gross profit of 4¢ per pair before fees.
The difficult part is confirming that the contracts are genuinely equivalent, matching events across thousands of markets, and monitoring prices in real time. Resolution criteria, deadlines, cancellation provisions, and unusual edge cases can all create meaningful differences.
That is what ArbEx is designed to address.
Relative value trading
Some markets are economically related even though they do not resolve on the same outcome.
Drawing from the equity markets, a useful comparison is Coca-Cola and PepsiCo. A strong earnings report from Coca-Cola may cause PepsiCo shares to rise because investors update their expectations for the broader beverage industry that quarter. The two securities are related, but they are not interchangeable.
Prediction markets contain similar relationships. For example:
- Candidate X wins the vote
- Candidate X is elected
The two outcomes are closely connected, but they are not equivalent. A candidate can win the popular vote without winning the election. A difference between their prices is therefore not automatically an arbitrage, but it may present a relative value opportunity when the implied relationship appears inconsistent with theoretical probabilities, current information, or other markets.
Liquidity may also be concentrated in one contract, causing a related market to adjust more slowly. The most liquid markets are usually the ones to move first.
Synthetic markets
Some outcomes can be replicated using combinations of other contracts.
Consider a CPI market in which one venue lists cumulative thresholds:
- CPI above 3.0%
- CPI above 3.1%
while another venue lists an exact outcome:
- CPI equals 3.1%
Assuming the contracts resolve to the same tick size, the exact 3.1% outcome can be represented as:
The synthetic position is therefore long (buy yes) the lower threshold and short (sell yes or buy no) the adjacent higher threshold. It pays only when CPI falls within the interval represented by the exact value contract.
ArbEx identifies many of these constructions automatically. Traders can also create custom synthetic tickers, define multi-leg relationships, and monitor their pricing in real time.
Why these opportunities exist
Prices can remain out of line for several reasons. A large order may temporarily move one venue without affecting another. Liquidity may be concentrated in a single contract. Market makers may use different models, face different inventory constraints, or simply fail to identify a related market.
Capital also remains fragmented. Traders must maintain separate balances on each venue, and moving funds between platforms can be slow. These limitations prevent obvious price differences from disappearing immediately.
ArbEx brings these markets into a single view. Its matching engine identifies equivalent and related contracts, constructs synthetic outcomes, and streams executable cross-venue edge in real time.