Welcome to PM Academy

Module 09 · Advanced · ~18 min

Arbitrage.

By the end of this module, you’ll be able to spot prices that disagree across markets and lock in the difference, the cleanest edge in trading, when you’re fast enough to grab it before someone else does.

Risk-free profit exists, if you’re fast enough to capture it. Learn to spot price gaps across platforms and extract guaranteed value.

Quick answer

What is arbitrage in prediction markets?

Arbitrage is buying and selling the same outcome simultaneously across different markets to lock in a price discrepancy: buy a contract where it’s cheap, buy the opposite side where it’s expensive, and collect the guaranteed difference at settlement. The gaps exist because liquidity is fragmented: platforms have different user pools, news propagates at different speeds, and retail-heavy venues misprice events that sharp-money venues get right. Four archetypes show up in PMs: cross-platform gaps (the most common), YES + NO parity breaks inside one market, multi-outcome markets whose probabilities don’t sum to 100%, and correlation arbs between logically linked markets. The execution rhythm is always the same: spot, buy low, buy the opposite, collect. But “risk-free” is a textbook word: execution latency, settlement-source mismatches, fee drag, thin depth, and counterparty risk all turn paper arbs into losses, so net out the costs before sizing up.

Section 01

What is arbitrage?

Exploiting price differences

Arbitrage is the simultaneous purchase and sale of the same asset across different markets to profit from a price discrepancy. In prediction markets, this means buying a contract on one platform where it’s cheap and selling (or buying the opposite side) on another where it’s expensive.

Fragmented liquidity

Different platforms have different user pools, creating natural price divergence.

Slow information propagation

News travels at different speeds to different platforms, creating temporary mispricings.

Different user bases

Retail-heavy platforms often misprice events that sharp-money platforms get right.

Price gap scanner

Platform A

YES price $0.42

Platform B

YES price $0.55
Platform B YES price 55¢
Spread 13¢
Profit per pair 13¢

Buy YES on Platform A at $0.42, buy NO on Platform B at $0.45. Guaranteed 13¢ profit per pair.

Section 02

Types of PM arbitrage.

Four flavors of arbitrage live in PMs, and they show up at different frequencies. Cross-platform is the most common. Parity (YES + NO ≠ $1) is the easiest to spot inside a single market. Multi-outcome and correlation arbs are rarer but pay better when they appear, because fewer traders are watching for them. Pick one to specialize in, the scanner that catches all four is your own attention, and you can’t split it.

Cross-platform

Same question, different prices on different exchanges. Buy cheap on one platform, sell expensive on another. The most common and accessible form of PM arbitrage.

YES/NO mispricing

When YES + NO don’t sum to $1.00, buying both sides locks in guaranteed profit. This happens when order books are thin or during high volatility events.

Multi-outcome

Markets with 3+ outcomes where total implied probability exceeds or falls below 100%. Buy all outcomes when the sum is under $1.00, or sell all when it’s over.

Correlation arb

Two logically linked markets priced inconsistently. If A implies B, but B is priced differently, you can exploit the logical relationship for guaranteed profit.

Section 03

The parity rule.

In any binary market, YES + NO must equal $1.00. When they don’t, free money exists. This is the most fundamental arbitrage in prediction markets.

YES price (cents) 45¢
NO price (cents) 50¢
Position size (shares)
All-in fee rate 2.0%

Combined trading fees, gas, and slippage as a percentage of capital deployed. A paper arb with a 3¢ spread dies at a 3% fee, slide this up and watch it happen.

Combined price

$0.95

BUY BOTH

Guaranteed profit of 5¢ per pair

Gross profit

$5.00

on $95.00 invested

Net after fees

+$3.10

$1.90 lost to fees (2.0%)

Section 04

Cross-platform execution.

Knowing the arb exists is the easy part. Capturing it before it closes, across two platforms, two order books, and two sets of fees, is where the actual work happens. The four-step pattern below is the same regardless of which arb type you’re trading. The only thing that varies is how fast you have to move.

Section 05

When “risk-free” isn’t.

Arbitrage is called “risk-free,” but only in textbooks. In real markets, three things can turn a paper-perfect arb into a loss. Recognize them before you size up, none of these failure modes are visible in the price gap itself.

Execution risk

Price moves before you complete both legs of the trade. The gap you spotted vanishes while you’re still placing the second order.

Settlement risk

Platforms disagree on the outcome or use different resolution sources. What resolves YES on one platform could resolve NO on another.

Fee drag

Transaction fees, gas fees, and withdrawal fees eat the spread. A 5¢ arb becomes a 1¢ loss after platform fees on both sides.

Liquidity risk

Can’t fill your full size at the quoted price. Partial fills reduce profit and leave you with unhedged exposure on one side.

Counterparty risk

Platform insolvency or frozen withdrawals lock your capital. Your profit is only real if you can actually withdraw it. Diversify across trusted platforms and never overallocate to any single venue.

Common questions

PM arbitrage: what people ask

Each answer also ships invisibly as schema.org FAQ data for search engines and AI assistants. Tap a question to expand.

  1. What types of arbitrage exist in prediction markets?
    Four. Cross-platform: the same question priced differently on different exchanges, the most common and accessible form. YES/NO mispricing: the two sides of one market don’t sum to $1.00, usually in thin books or during high-volatility events. Multi-outcome: a market with 3+ outcomes whose total implied probability strays from 100%, so you buy all outcomes under $1.00 or sell all over it. Correlation arb: two logically linked markets priced inconsistently.
  2. How does YES + NO parity arbitrage work?
    A YES/NO pair always settles to exactly $1.00, so if you can buy both sides for less, the difference is locked in. Buy YES at 45¢ and NO at 50¢ and the 95¢ pair pays $1.00: 5¢ guaranteed gross profit per pair. Fees decide whether it survives, a paper arb with a 3¢ spread dies at a 3% all-in fee rate.
  3. Why do risk-free arbitrage trades lose money?
    Five reasons the price gap doesn’t show you: execution risk (the gap vanishes while you’re placing the second leg), settlement risk (platforms use different resolution sources and can resolve the same question differently), fee drag (a 5¢ arb becomes a 1¢ loss after fees on both sides), liquidity risk (partial fills leave unhedged exposure on one side), and counterparty risk (insolvency or frozen withdrawals make the profit theoretical).
  4. What checks should you run before sizing up an arbitrage trade?
    Five: the question wording is identical on both platforms; the resolution sources match, since different oracles can produce different outcomes; the all-in cost including trading fees, gas, slippage, and withdrawals still leaves profit; liquidity depth on both sides covers your size without erasing the spread; and the capital lockup is worth it, a 3% return over 6 months is only 6% annualized.
  5. Why do arbitrage opportunities exist at all?
    Fragmented liquidity. Different platforms have different user pools, so prices diverge naturally; news travels to different platforms at different speeds, creating temporary mispricings; and retail-heavy venues misprice events that sharp-money venues price correctly. The window is short, the gap you spotted can vanish while you’re still placing the second order, which is why the four-step rhythm (spot, buy low, buy the opposite, collect) is mostly about speed.

Section 06

Build your process.

Five checks before you size up. The Continue button unlocks at 5/5.

Advanced tier wrap-up

Take it live.

You’ve completed all three Advanced-track modules, analysis, portfolio construction, and cross-platform arbitrage, capping a seven-module journey from Fundamentals to Alpha. The next thing the curriculum can teach you is what your own fills feel like. Open a small position with what you’ve built; the rest of the academy makes more sense once your book has touched the live market.

Start trading on Limitless

Advanced tier complete

Gaps captured.

You can find money lying on the ground. When two sibling markets disagree or YES + NO doesn’t add to a dollar, you’ll know how to close the gap and pocket the difference, the closest thing in trading to a free lunch.

Concretely, you’ve completed all three Advanced-track modules, analysis, portfolio construction, and cross-platform arbitrage. Three things you walk away with from the arbitrage capstone:

01

The ability to scan a market for the four arb archetypes, cross-platform, YES + NO ≠ $1.00 parity breaks, multi-outcome, and correlation, and know which one you’re actually looking at.

02

A four-step execution rhythm (spot → buy low → buy the opposite → collect) so you can close both legs before the gap closes on you.

03

A pre-flight checklist for the risks that make “risk-free” arbs lose money, execution latency, settlement-source mismatch, fee drag, liquidity depth, and counterparty exposure, so you only size up when the math still works after costs.

Quick recall

Without scrolling back, can you answer these?

Five questions across the Advanced tier. Click each to reveal. The test is whether you can answer first.

  1. You see best-bid $0.54 and best-ask $0.58. What’s the spread telling you, and where will a market order actually fill?
    The 4¢ spread is your round-trip cost of impatience. You pay $0.58 to enter and receive $0.54 to exit, so the market needs to move 4¢ in your favour just to break even. A market order to buy YES fills at $0.58 against the best ask first, then walks up the book if your size exceeds the top level. Limit orders sit on the book; market orders cross it.
  2. Before you place a trade, what one sentence in the market description deserves the most attention, and why does it determine whether you get paid?
    The resolution source: the specific data feed, oracle, or authority that decides the outcome. “Bitcoin above $70k” settled by Coinbase spot at UTC 23:59 is a different market than the same question settled by Binance perp at NY close. The price you’re trading is only as good as that source. If you can’t name it cleanly, you don’t know what you’re betting on.
  3. You priced a market at 0.55; it’s trading 0.42. What’s the expected value of $100 at risk, and why isn’t EV alone enough to size the position?
    EV per dollar = (0.55 × $1.00) − $0.42 = +$0.13, so $100 at risk has roughly $31 of expected value (you buy ~238 shares at 0.42; 238 × ($1.00 − $0.42) × 0.55 + 238 × (−$0.42) × 0.45 ≈ $31). EV alone ignores variance and correlation. That’s why Module 03 layered Kelly on top, and why Module 08 sizes across positions, not just within one.
  4. Platform A has YES at $0.42; Platform B has NO at $0.55. Is there an arb, and what kills it most often in practice?
    Yes, the parity check fails: $0.42 + $0.55 = $0.97, so buying both sides costs $0.97 and pays $1.00, a 3¢ gross arb. What kills it: fees on both legs (often 1–2% each), execution latency between platforms, settlement-source mismatch (the two platforms might use different oracles), and depth: the quoted price disappears once you size up. After costs, a 3¢ gap is usually break-even at best.
  5. How does Module 08’s correlation lens change the way you’d execute a Module 09 arb?
    Naively, an arb is correlation-neutral: both legs settle together and cancel. The synthesis: legs look paired but execute on different platforms, oracles, and timestamps. The legs are correlated almost +1 only on a clean settlement, and can decouple under stress (one platform halts, one oracle prints a different number). So you size arbs the way Module 08 sizes a portfolio: cap exposure per platform, never assume −1.0 correlation between legs you can’t close simultaneously.

Next up: the Football tier, how the sportsbook world prices games, where its biases leak into PMs, and how to translate xG into edge.

Continue to Module 10

Complete the checklist above to unlock