Cross-Platform Prediction Arbitrage: Real $10k Case Study
Cross-platform prediction arbitrage is the practice of simultaneously buying and selling the same event outcome on different prediction market platforms to lock in a risk-free profit from price discrepancies. In this case study, a $10,000 starting portfolio was deployed across three major platforms over 90 days, generating a net return of 8.3% — with detailed breakdowns of every trade, mistake, and lesson learned along the way.
What Is Cross-Platform Prediction Arbitrage?
Prediction market arbitrage exploits the fact that platforms like Polymarket, Kalshi, and Manifold don't always price the same event identically. When one platform shows a 58% probability for a political outcome while another prices it at 45%, there's a mathematical opportunity to profit regardless of how the event resolves.
This is fundamentally different from speculative trading. You're not betting on who wins — you're betting on the gap between platforms closing.
Think of it like sports betting arbitrage, but with legally clearer structures (depending on your jurisdiction) and generally more liquid markets during high-profile political and economic events.
Why Price Discrepancies Exist
- Different user bases with different information and biases
- Liquidity gaps causing markets to move at different speeds
- Platform-specific mechanics (order books vs. automated market makers)
- Time zones and trading hours affecting when prices update
- News spreading unevenly across communities
For a deeper strategic foundation before diving into execution, the Trader Playbook: Natural Language Strategy for Q2 2026 is worth reading alongside this case study.
Portfolio Setup and Platform Selection
Starting Conditions
| Parameter | Detail |
|---|---|
| Starting Capital | $10,000 |
| Duration | 90 days (Q1 2026) |
| Primary Platforms | Polymarket, Kalshi, Metaculus |
| Market Focus | U.S. political events, economic indicators |
| Tools Used | Manual monitoring + API alerts |
| Risk Per Trade | Maximum 5% of portfolio ($500) |
Platform Comparison
Before placing a single dollar, it's critical to understand how each platform operates:
| Platform | Market Type | Fees | Min Trade | Liquidity |
|---|---|---|---|---|
| Polymarket | AMM + Order Book | ~2% spread | $1 | High (political) |
| Kalshi | Order Book | 7% of profit | $5 | Medium-High |
| Manifold | Play money + real | Varies | Minimal | Low-Medium |
| PredictIt | Exchange | 10% profit + 5% withdrawal | $5 | Medium |
Kalshi was selected as the primary platform for regulatory-compliant U.S. political contracts, while Polymarket served as the counterpart for most arbitrage legs. Getting properly set up across all of these platforms takes time — the KYC & Wallet Setup for Prediction Markets: Algorithm Guide walks through the verification process efficiently.
The 90-Day Execution: Month-by-Month Breakdown
Month 1: Finding the Right Markets
The first 30 days were deliberately conservative. Only 6 arbitrage trades were executed, with an average spread of 4.2% before fees.
Key trade example — Federal Reserve Rate Decision:
- Kalshi priced "Fed holds rates in March" at 62 cents (62% implied probability)
- Polymarket priced the same outcome at 71 cents (71% implied probability)
- Strategy: Buy NO on Polymarket (equivalent to buying "rates change" at 29 cents), buy YES on Kalshi at 62 cents
- Net position: Guaranteed profit if spread narrows OR event resolves at either extreme
After fees, this trade netted $47 on a $400 position — a 11.75% gross return, but only 6.8% net after platform fees.
Month 1 Results:
- Trades executed: 6
- Gross P&L: +$312
- Fees paid: $89
- Net P&L: +$223 (2.23% portfolio growth)
Month 2: Scaling Into Political Markets
Political events created the widest and most frequent spreads. The 2026 midterm cycle generated enormous activity on both Kalshi and Polymarket, and the Kalshi Trading Quick Reference After the 2026 Midterms provided essential context for reading market structure during this period.
Key trade example — House Seat Competitive District:
- Polymarket: Democratic candidate wins District X at 38%
- Kalshi: Same race at 29%
- Spread: 9 percentage points
This 9-point gap is unusually large. Here's why it existed:
- A local poll had just released on one platform's community forum but hadn't propagated to the other
- The Polymarket AMM had absorbed a large one-directional bet, artificially inflating the price
- Weekend liquidity was thin, exaggerating the move
Position taken: $450 on Kalshi YES + $380 on Polymarket NO. The spread closed within 48 hours as arbitrageurs flooded in.
Net profit: $61 on $830 deployed (7.3% return, 4 days)
Month 2 Results:
- Trades executed: 11
- Gross P&L: +$541
- Fees paid: $134
- Net P&L: +$407 (3.9% on remaining portfolio)
Month 3: Automation Attempts and Lessons
Month 3 introduced partial automation using API monitoring. Alerts were set to trigger whenever the same contract on two platforms differed by more than 5 percentage points.
This is where things got educational. Automated scanning revealed 47 apparent arbitrage opportunities — but only 14 survived after accounting for:
- Bid-ask spreads eating the margin
- Contract definition mismatches (subtle wording differences)
- Withdrawal timing that would leave capital locked past resolution
- Slippage on larger position sizes
For traders serious about automation, Automating Polymarket vs Kalshi with Limit Orders covers the technical execution in detail, including how to handle order routing and timing.
Month 3 Results:
- Trades executed: 14
- Gross P&L: +$389
- Fees paid: $98
- Net P&L: +$291 (2.6% net)
Step-by-Step: How to Execute a Cross-Platform Arbitrage Trade
Here's the exact process used for each qualifying trade in this portfolio:
- Identify candidate contracts — scan both platforms for the same real-world event with a measurable price gap (minimum 5% after estimated fees)
- Verify contract equivalence — read the resolution rules on both platforms word for word; "wins the election" vs. "wins the popular vote" are not the same contract
- Calculate net expected value — subtract both platforms' fees, estimate slippage based on your position size and current liquidity
- Check timing alignment — confirm both contracts resolve on the same date/trigger and that you can exit or receive resolution before capital is locked
- Size the position — use no more than 5% of portfolio per trade; larger positions cause slippage that destroys the edge
- Place orders simultaneously — leg risk (entering one side before the other) is the biggest operational risk; use limit orders pre-loaded before confirming
- Monitor and exit if spread widens abnormally — sometimes one platform's price moves further against you; pre-define your stop
- Record everything — track gross P&L, fees, resolution timing, and effective annualized return for each trade
Risk Management: What Almost Went Wrong
Two trades in this case study came dangerously close to losses:
Near-miss #1: Contract Wording Mismatch A Supreme Court case market appeared to show a 7-point spread. Only after taking both positions did close reading reveal that one platform resolved on "ruling issued" while another resolved on "ruling upheld on appeal." These are very different events. The position was closed at a small loss (-$28) rather than holding through resolution risk.
The Supreme Court Ruling Markets: Deep Dive for Q2 2026 article specifically covers how resolution criteria vary across platforms for legal and judicial events — essential reading before trading this category.
Near-miss #2: Liquidity Evaporation A mid-sized position ($600) in an economic indicator market lost its counterpart liquidity when Polymarket's market suddenly went illiquid. The position had to be closed at a worse price on Kalshi, resulting in a $34 loss instead of a projected $55 gain.
Key risk management rules that protected the portfolio:
- Never exceed 5% of portfolio in one arbitrage pair
- Always verify liquidity depth at your intended position size before entering
- Maintain 20% cash reserve to cover unexpected margin or withdrawal delays
- Read resolution criteria on both platforms before any trade
Final Portfolio Results: 90-Day Summary
| Month | Trades | Gross P&L | Fees | Net P&L |
|---|---|---|---|---|
| Month 1 | 6 | +$312 | -$89 | +$223 |
| Month 2 | 11 | +$541 | -$134 | +$407 |
| Month 3 | 14 | +$389 | -$98 | +$291 |
| Total | 31 | +$1,242 | -$321 | +$921 |
- Starting portfolio: $10,000
- Ending portfolio: $10,921
- Net return: 9.21% gross / 8.3% net of fees
- Annualized equivalent: ~33%
- Win rate: 29/31 trades profitable (93.5%)
- Average hold time: 4.2 days
The win rate is high because genuine arbitrage has a mathematical edge — the two near-misses were operational errors, not market losses. Scaling this strategy requires automating house race predictions and political market monitoring to catch opportunities faster than manual scanning allows.
Scaling Beyond $10k: What Changes
A $10,000 portfolio can execute most trades without significant slippage. At $50,000+, the math shifts:
- Slippage becomes material — your own orders move the market against you
- Fewer qualifying opportunities — the 5% rule means positions need larger spreads to stay viable
- API access becomes essential — manual scanning can't compete at scale
- Cross-platform capital allocation — keeping equal balances on each platform without leaving idle cash is a full-time optimization problem
For institutional or larger retail traders, AI Agents for Prediction Markets covers how algorithmic tools are reshaping execution at higher capital levels.
Frequently Asked Questions
Is cross-platform prediction arbitrage legal?
Prediction market arbitrage is generally legal in jurisdictions where the underlying platforms are licensed to operate. Kalshi is regulated by the CFTC, while Polymarket operates under different legal frameworks. Always confirm the regulatory status in your specific jurisdiction before trading, as rules vary significantly by country and state.
How much capital do you need to start prediction market arbitrage?
You can technically start with as little as $500-$1,000, but the practical minimum for meaningful returns after fees is around $3,000-$5,000. Below this threshold, flat fees and minimum trade sizes on platforms like Kalshi eat too much of the margin. The $10,000 level used in this case study is a comfortable starting point.
What are the biggest risks in cross-platform arbitrage?
The three primary risks are contract mismatch (resolution criteria differ between platforms), liquidity risk (one side becomes untradeable before you can exit), and timing risk (one platform resolves before the other, leaving you with a directional position). Careful pre-trade verification eliminates most of these, but they can never be fully automated away.
How do prediction market fees affect arbitrage profitability?
Fees are the single biggest drag on arbitrage returns. In this case study, $321 in fees reduced gross returns by 25.8%. Kalshi charges approximately 7% of profit, while Polymarket's AMM spread functions as an implicit fee of 1-3%. Always calculate net-of-fee returns before entering any position — a 4% gross spread can become 0% or negative after fees.
Can you automate cross-platform arbitrage completely?
Partial automation is very effective — API-based price monitoring and alert systems dramatically improve opportunity capture. Full automation, including order execution, is achievable but requires careful handling of leg risk, slippage management, and contract verification logic. Most serious arbitrageurs use semi-automated systems where alerts are automated but final execution involves human confirmation.
How do I find arbitrage opportunities between platforms?
The most effective methods are: (1) API monitoring with price-comparison scripts, (2) community forums and Discord servers where traders share mispricings, and (3) systematic manual scanning during high-volatility news events when prices diverge fastest. The opportunities are most frequent and widest during breaking news, earnings reports, and major political announcements.
Start Your Own Arbitrage Strategy with PredictEngine
This 90-day case study proved that cross-platform prediction arbitrage is a viable, repeatable strategy — but execution quality is everything. The difference between 8.3% net returns and breaking even came down to fee awareness, contract verification, and position sizing discipline.
PredictEngine is built for exactly this kind of systematic prediction market trading. With real-time market monitoring, cross-platform analytics, and tools designed for both manual and automated execution, it gives you the infrastructure to run this strategy at any scale. Whether you're starting with $2,000 or $200,000, the edge is in the tooling — explore PredictEngine's full platform and see how it fits your arbitrage workflow today.
Run your own quant desk
Felix lets you research, backtest and run automated trading strategies, with risk controls built in.
Try Felix