Prediction Market
Arbitrage in 2026
Prediction market arbitrage means buying YES on the same outcome at a lower price on one platform and selling NO (or buying YES) at a higher price on another platform, locking in a guaranteed profit regardless of how the event resolves. As the US market matures with Kalshi, Polymarket, and PredictIt running simultaneously, price discrepancies are real and exploitable. This guide covers how to find them, execute them, and what the risks actually are.
What prediction market arbitrage looks like
Prediction markets on the same event can trade at different prices across platforms because they have separate liquidity pools and different trader bases. When a price difference creates a guaranteed profit after fees, that's an arbitrage opportunity.
PredictIt: "Will Candidate X win?" — NO trading at 44¢
Arb trade:
→ Buy YES on Kalshi at 52¢ (cost: $520 for 1,000 contracts)
→ Buy NO on PredictIt at 44¢ (cost: $440 for 1,000 contracts, capped at $850)
If Candidate X wins: Kalshi YES pays $1,000 (+$480) — PredictIt NO loses $440 → net +$40
If Candidate X loses: PredictIt NO pays $1,000 (+$560) — Kalshi YES loses $520 → net +$40
Before fees: guaranteed $40 profit on $960 invested (~4.2% return). After fees: ~$5–20 profit depending on Kalshi fee rate and PredictIt's 10% profit + 5% withdrawal.
This is pure arbitrage: the profit is independent of the outcome. The challenge is finding these windows before they close, executing both legs fast enough, and making sure fees don't erase the spread.
Where price discrepancies actually occur
Arb opportunities
The highest-volume arb pair. Polymarket has significantly more international trader volume on political and geopolitical markets: their pricing can diverge 2–6 cents from Kalshi on the same event. Check both orderbooks on US election markets, Fed rate markets, and major political races. Polymarket's QCEX platform has joined this arb ecosystem since December 2025.
Arb opportunities
US political markets are the primary overlap. PredictIt's $850 position cap limits arb size but the price discrepancies can be larger (5–10 cents) because PredictIt's retail-heavy user base prices differently from Kalshi's more sophisticated traders. Best for niche congressional and state races where PredictIt has the only liquid market.
Arb opportunities
Both active on US political markets but very different audiences: Polymarket is crypto-native global traders, PredictIt is US politics retail. Discrepancies occur especially around breaking political news when one platform reprices faster than the other. USDC vs USD settlement means you need accounts on both and a float in each currency.
Arb opportunities
Robinhood uses Kalshi's same orderbooks for most contracts: prices are identical or near-identical. True arb between these two is rare. However, there are occasionally markets where Robinhood's interface shows different pricing due to UI rounding or order type differences.
How to execute a prediction market arb
- 1 Maintain funded accounts on both platforms
You need capital deployed on both platforms before you see an opportunity. An arb window often lasts minutes or less: transferring funds after you spot the discrepancy is too slow. Keep a float on Kalshi, Polymarket, and PredictIt simultaneously.
- 2 Monitor the same market on multiple platforms
Check the same contract on each platform side by side. Tools: use each platform's web app in separate browser tabs, or build a simple script using Kalshi's REST API and Polymarket's CLOB API to compare prices automatically. Manual monitoring is feasible for 5–10 markets; systematic monitoring requires API work.
- 3 Calculate the arb after fees
An apparent 5-cent discrepancy may not be profitable after fees. Kalshi charges 0–7% of profit on winning trades; PredictIt charges 10% of profit + 5% withdrawal. Calculate both legs: (price spread × position size) − (Kalshi fee on winning leg) − (PredictIt fee on winning leg). If the net is positive, it's profitable.
- 4 Execute both legs as simultaneously as possible
Place both orders as close together in time as possible. The slower leg creates risk: if the market moves between the first and second trade, you may have a partially-hedged directional position rather than a pure arb. Limit orders on both platforms help; market orders are faster but may fill at worse prices.
- 5 Manage until resolution
Hold both positions until the market resolves. Don't exit early: early exit introduces new execution risk. At resolution, one leg wins and one loses; your net profit is the price spread minus fees that you locked in at entry.
- 6 Track and report each platform separately for taxes
Each platform issues its own 1099-MISC. Your Kalshi profits and PredictIt profits are reported separately. The arb may show a profit on one platform and an offsetting loss on the other: both are reportable. Work with a tax professional familiar with derivatives if you're doing this at volume.
Why arb trades fail — the real risks
Execution risk
Both legs must fill at the expected prices. In fast-moving markets, your second leg may fill at a worse price: turning a theoretical profit into a loss. Use limit orders to control your fill price, but accept that limits may not fill at all if the market moves.
Resolution dispute risk
Kalshi's unilateral market resolution (e.g. the Khamenei market) can break an arb if one platform resolves differently from another. If Kalshi resolves "No" but PredictIt resolves "Yes" due to a definitional dispute, both legs could lose. Read each platform's resolution criteria carefully before trading.
Fee erosion
On small spreads (2–3 cents), fees can eliminate the entire arb margin. At 5% Kalshi profit fee + 10% PredictIt profit fee + 5% PredictIt withdrawal fee, you need a spread of at least 6–8 cents to net a real profit on most trade sizes.
Capital lock-up
Both legs tie up capital for the full duration of the contract, which could be months for political markets. Calculate the annualized return to assess if the locked-up capital is earning its keep vs other opportunities.
Building an API-based arb system
Manual monitoring of 10+ markets across 3 platforms is unsustainable. Serious arb traders use the platforms' APIs to automate price monitoring and trigger alerts or orders:
- Public API at
trading-api.kalshi.com - Real-time orderbook data via WebSocket
- Programmatic order placement
- Market metadata: resolution criteria, end date
- Central limit orderbook API for QCEX
- WebSocket feeds for real-time prices
- On-chain settlement: programmatic execution possible
- Market matching via condition IDs
To build a basic arb scanner: pull Kalshi and Polymarket prices for the same event (match by event title and resolution date), compare YES prices plus (1 − NO prices), flag any pair where the sum exceeds 1 + estimated fees, and alert for manual review or auto-execute. The hardest part is event matching: the same question is described differently on each platform.
Common questions
Is prediction market arbitrage legal? +
Yes: fully legal. You're trading on multiple licensed CFTC exchanges simultaneously. Holding offsetting positions across platforms is standard practice in financial markets. Report profits from each platform accurately on your 1099-MISC forms.
What's the best platform pair for prediction market arb? +
Kalshi ↔ Polymarket for political and macro markets: highest volume, most frequent discrepancies. Kalshi ↔ PredictIt for niche US political markets where PredictIt has deeper liquidity in congressional and state races. PredictIt's $850 cap limits position size on that leg.
How much can you make arbitraging prediction markets? +
Realistic retail arb: 2–6 cent spreads on political markets with $500–2,000 positions per arb. A 4-cent spread on a $1,000 position yields roughly $20–40 pre-fee gross profit. After fees on both legs, net is $5–20 per round trip. Systematic API traders doing 50–100 arbs/month can generate meaningful income; manual retail arb is supplementary.
Do fees eliminate prediction market arb profits? +
On small spreads, yes. You need a spread of at least 6–8 cents to net a real profit after Kalshi's profit fee and PredictIt's 10%+5% structure. Kalshi ↔ Polymarket arb is more fee-efficient because Polymarket's 2% taker fee is charged at entry (not on profit), so the net fee is more predictable.
More strategy guides
Prediction market strategy
Kelly Criterion, finding edge, common mistakes, and systematic approaches to trading event contracts.
ComparisonKalshi vs Polymarket
The two primary arb platforms compared in depth: fees, orderbooks, settlement, and API access.
ComparisonPolymarket vs PredictIt
The second major arb pair: fee structures, caps, and which markets are only on one platform.