Prediction markets have been around for decades in academic circles, but 2024-2026 turned them into mainstream alternatives to traditional sportsbooks. If you've heard the names Kalshi or Polymarket but aren't sure what they actually are, this guide is for you.
By the end, you'll understand: - What a prediction market is and how it differs from a sportsbook - How prices map to probabilities ("cents = percent") - Why markets often produce more accurate odds than a single sportsbook - How to start using one in 2026 (Kalshi for US, Polymarket for international) - How to think about edge and expected value when trading
The One-Sentence Definition
A prediction market is a public exchange where users buy and sell contracts that pay $1 (or 100 cents) if a specific event happens and $0 if it doesn't.
That's it. Everything else is mechanics.
Example: "Will the Dallas Cowboys Win the Super Bowl This Year?"
On a prediction market, this event has two contracts:
- YES contract: Pays $1 if Dallas wins the Super Bowl, $0 if they don't.
- NO contract: Pays $1 if Dallas doesn't win, $0 if they do.
Let's say the market is currently trading at: - YES: 15 cents (you can buy a YES contract for $0.15) - NO: 85 cents (you can buy a NO contract for $0.85)
If you buy 100 YES contracts at 15 cents, you pay $15. If Dallas wins the Super Bowl, you get $100 back (100 × $1 each). If they don't, you get $0.
The price IS the probability. The market is saying Dallas has a 15% chance of winning. That's the collective estimate of everyone trading that market.
Why "Prices Are Probabilities" Is Powerful
On a traditional sportsbook, you see odds like "Dallas +550." This translates to an implied probability of about 15.4%, but the sportsbook has already baked in a 5-7% margin for itself (the "vig"). The actual fair probability the sportsbook thinks is closer to 17-18%.
On a prediction market with zero fees (like Polymarket), 15 cents means 15% — no vig, no margin. If you buy YES for 15 cents, your break-even rate is 15%. Anything above that is profit in expectation.
This is the structural reason prediction markets matter: the price is the probability, not the probability plus a house edge.
The Four Components You Need to Understand
1. Contracts
The basic unit is a contract that pays $1 on one outcome and $0 on the other. Markets are binary (YES/NO) by default. Some platforms offer multi-outcome markets (e.g., "Which team wins the division?") where you can buy shares in any candidate team.
2. Order book
Every prediction market has a limit order book — a list of buy orders and sell orders at various prices. Just like a stock exchange. If you see YES quoted at 15c bid / 16c ask, that means: - Someone is bidding 15c for YES contracts (wants to buy them cheap) - Someone is asking 16c to sell YES contracts
You can either take the ask (buy 16c, fill immediately) or place a bid (offer 15c, wait to get filled).
3. Spread
The gap between bid and ask (16c - 15c = 1c in this example) is the "spread." Tight spreads (0.5c-1c) happen on liquid markets. Wide spreads (5-10c) happen on thin markets and make trading more expensive.
4. Resolution
When the event actually happens, the market resolves. If Dallas wins the Super Bowl, every YES contract pays $1 and every NO contract pays $0. Losers lose their entire stake; winners collect.
How Prediction Markets Differ From Sportsbooks
Five key differences, each with a practical consequence:
| Dimension | Sportsbook | Prediction Market |
|---|---|---|
| Counterparty | The book | Another trader |
| Fees | 5-10% vig baked in | 0-2% explicit (often 0 on Polymarket) |
| Can you sell before resolution? | No | Yes — just sell at current market price |
| Who sets the price? | The bookmaker's oddsmakers | Collective trading activity |
| Beat-the-system strategy | Find soft lines early | Arbitrage and information edge |
The practical upshot: on a prediction market, you can "lock in" a profit by buying low and selling before the event resolves. On a sportsbook, once you bet you're stuck with the outcome.
This makes prediction markets much more like stock trading and much less like casino gambling.
Why Prediction Markets Are Often More Accurate Than Single Sportsbooks
Public prediction market prices aggregate information from many diverse traders. Economists have shown this works because:
-
Diverse sources of information. Some traders know the injury report. Some follow coaching tendencies. Some run ML models. The market aggregates all of them.
-
Money is on the line. Traders have skin in the game. Bad predictions cost them. This filters out casual/bad takes and rewards accuracy.
-
Active arbitrage pressure. If a market says 20% on Dallas winning but the "true" probability is 15%, someone profits by selling YES. That pressure pushes the price toward fair.
Academic studies (including work done on PredictIt and the Iowa Electronic Markets) consistently find that prediction market prices are within 2-5% of the true outcome probability on well-liquidated markets.
But: thin, illiquid markets can be wildly off. If only 3 people are trading a market for Bowling Green vs. Eastern Michigan football, you might see prices that don't reflect anything meaningful. Always check liquidity before trusting a price.
How to Start Using One (US vs. International)
If you're in the United States: Kalshi
Kalshi is the largest US CFTC-regulated prediction market. It offers sports events (NFL, NBA, MLB, NHL), political events, macroeconomic events, and weather.
Onboarding: 1. Sign up at kalshi.com 2. Complete KYC (SSN, bank linkage — standard US finance onboarding) 3. Deposit funds via ACH (takes 1-3 days) or debit card (instant) 4. Browse markets and place trades
Fees: 0-2 cents per contract depending on the market. Higher than Polymarket but the cost of being legal in the US.
If you're international: Polymarket
Polymarket is the largest global prediction market. Deepest liquidity, best for sharps, runs on the Polygon blockchain with USDC-denominated contracts.
Onboarding: 1. Connect a crypto wallet (MetaMask, WalletConnect) 2. Fund with USDC on Polygon 3. Browse markets and trade directly via web UI
Fees: 0% trading fee + gas (typically under $0.01 on Polygon).
If you want to practice risk-free: Manifold
Manifold is a play-money prediction market. No KYC, no money, but real forecasters and real leaderboards. Ideal for testing a strategy before committing real capital.
Understanding Edge and Expected Value
This is where prediction markets become interesting for serious traders. Here's the core math:
If a market says 40% but you think the true probability is 50%, and you can buy YES contracts at 40 cents:
- You pay $0.40 per contract
- True value is $0.50 (your model's probability × $1.00 payout)
- Your edge is $0.10 per contract, or 25% ROI in expectation
You won't win every trade. But if your probability estimates are accurate, you'll win enough times (and at big enough margins) to come out ahead over hundreds of trades.
This is why the quality of your probability estimate matters so much. If your model thinks "50% chance" but actually means "45% chance," your edge is 5c instead of 10c, and losing half your trades eats most of that.
The discipline of checking whether your model is calibrated — where we publish ZenHodl's 5,345-game regular-season calibration report — is why calibration matters more than raw accuracy.
How to Think About Risk
Three rules that will save you money:
1. Size positions using Kelly or fractional Kelly
Kelly's formula tells you the optimal bet size given your edge. In practice, full Kelly is too aggressive (you'll go broke from variance). Use quarter-Kelly or half-Kelly for prediction markets.
Rough rule: if your edge is 10%, don't bet more than 2-3% of your bankroll on a single contract.
2. Avoid thin markets
Depth matters. If a market has only $500 on the order book and you're trying to buy $200 worth, you'll move the price against yourself. Stick to markets with at least $5-10k depth unless you're trading very small size.
3. Understand resolution risk
Sometimes markets resolve in disputed ways. The 2024 election markets on Polymarket had public disputes about exact outcomes. Kalshi has a CFTC-regulated resolution process but that comes with its own timelines.
Always read the exact resolution criteria before trading. "Will [Team X] win the championship this year?" is clearer than "Will player X make the All-Star Team?" (what defines All-Star?).
Where ZenHodl Fits
ZenHodl is a calibrated sports prediction API. We give you probability estimates across 10+ sports. If you're trading on Kalshi or Polymarket, our API tells you what we think a game's true probability is — which lets you compare against the market price and find edges.
We're not affiliated with Kalshi or Polymarket. We don't take a cut of your trades. We just publish calibrated numbers that you can integrate into your own decision-making.
If you want to try it, a 7-day free trial gives you full API access with no credit card required.
Where to Learn More
Next-level resources across our network:
Strategy and math: - Can You Actually Win at Sports Betting Long Term? — the 52.4% threshold and the 5 practices that matter - Kelly Criterion for prediction market sizing — position sizing math - Calibrated probabilities in prediction markets — why calibration beats accuracy
Platform comparison: - Best Prediction Market Apps 2026 — Kalshi vs Polymarket vs others - Complete Guide to Prediction Market APIs — Polymarket CLOB, Kalshi REST, Python code
Build your own: - How to build a sports prediction model — Python tutorial - Feature engineering for sports win probability — the 15 features that matter - Calibrating XGBoost with isotonic regression — honest probability calibration
Production examples: - NCAAMB 2025-26 season report — 5,345 games, ECE 4.39% - 2026 March Madness backtest — 67 tournament games, 71.6% accuracy - Super Bowl LX retrospective — 9 of 13 playoff games correct
Summary
Prediction markets are exchanges where the price of a contract equals its probability. They're better than sportsbooks because they have no vig, you can sell before resolution, and public prices aggregate diverse information. In 2026, Kalshi is the US-regulated choice, Polymarket is the international heavyweight, and Manifold is the play-money sandbox for testing.
Trade with small size until you trust your probability estimates. Always check liquidity. Understand resolution risk. Use Kelly sizing. Don't treat it like a casino — treat it like a market.
This beginner's guide is part of our Prediction Markets 101 series. Last updated April 22, 2026.