The Kelly Criterion is a mathematical formula — f* = (bp − q) / b — that calculates the exact fraction of your bankroll to wager on any bet in order to maximize the long-run geometric growth rate of your capital without risking ruin.
If you've ever wondered why some prediction market traders consistently grow their accounts while others blow up on a single high-confidence trade, the answer almost always comes down to position sizing. The Kelly Criterion is the gold standard framework for solving that problem — and it's more accessible than most people think.
This guide starts from the very beginning and goes all the way to advanced multi-market applications. Whether you're new to prediction markets or already trading on Kalshi and Polymarket, you'll leave with a concrete, numbers-driven system you can apply immediately.
What Is the Kelly Criterion and Why Does It Matter for Prediction Markets?
The Kelly Criterion was developed by Bell Labs researcher John L. Kelly Jr. in 1956 and published in the Bell System Technical Journal. Originally designed for signal transmission problems, it was quickly adopted by gamblers and investors because it solves a universal problem: how much of your capital should you risk on a positive expected value opportunity?
In prediction markets — where you're buying contracts priced between $0 and $1 based on the probability of real-world events — Kelly is especially powerful. Every trade has an implied market probability, and every informed trader has their own estimated probability. The gap between those two numbers is your edge. Kelly tells you exactly how hard to push that edge.
Without a framework like Kelly, even traders with genuine edge consistently underperform because they bet too much on good trades (destroying their bankroll on variance) or too little on great trades (leaving growth on the table). If you're newer to the space and want a foundation first, check out our complete beginner's guide to trading prediction markets.
How Does the Kelly Formula Actually Work?
The core Kelly formula is:
f* = (bp − q) / b
- f* = the fraction of your bankroll to wager
- b = the net odds received on the bet (profit per $1 risked)
- p = your estimated probability the bet wins
- q = your estimated probability the bet loses (1 − p)
Let's run a real example. Suppose Kalshi is pricing a Fed rate cut contract at 40 cents (implying 40% probability). After your analysis, you believe the true probability is 55%. Here's how Kelly sizes that position:
- b = 0.60 / 0.40 = 1.50 (you risk $0.40 to win $0.60)
- p = 0.55
- q = 0.45
- f* = (1.50 × 0.55 − 0.45) / 1.50 = (0.825 − 0.45) / 1.50 = 0.375 / 1.50 = 25%
Full Kelly says to wager 25% of your bankroll. In practice, most professional traders use Fractional Kelly — more on that below.
Why Do Most Traders Use Fractional Kelly Instead of Full Kelly?
Full Kelly maximizes long-run growth but produces brutal short-term volatility. Research by economist Edward Thorp, who applied Kelly successfully in both blackjack and financial markets, consistently found that half-Kelly (wagering 50% of the Kelly recommendation) captures roughly 75% of the maximum growth rate while cutting variance by 75%. That's an exceptional risk-adjusted trade-off.
The three most common fractional Kelly approaches used in prediction markets:
- Half Kelly (0.5f*): The most popular choice. Dramatically smoother equity curve, strong long-run growth. Recommended for most traders.
- Quarter Kelly (0.25f*): Conservative. Ideal when you have genuine uncertainty about the accuracy of your probability estimates.
- Full Kelly (1.0f*): Only appropriate when your edge is extremely well-calibrated and you have a large enough bankroll to absorb variance.
In the example above, Half Kelly would recommend wagering 12.5% of your bankroll rather than 25%. For most prediction market traders, that's the right call — your probability estimates are rarely perfect, and overconfidence is the most common Kelly mistake.
What Happens When You Overbet or Underbet?
Kelly's mathematical elegance lies in its asymmetry of errors. Overbetting is always more damaging than underbetting by an equivalent amount:
- Betting 2× Kelly produces the same long-run growth rate as betting 0% (i.e., not trading at all) — you're working hard for nothing
- Betting beyond 2× Kelly is a path to guaranteed eventual ruin, regardless of your edge
- Underbetting reduces growth but never threatens your bankroll
This is why professional traders almost universally err toward fractional Kelly. The cost of being slightly too conservative is modest. The cost of being slightly too aggressive compounds catastrophically.
How Do You Apply Kelly Criterion Across Multiple Prediction Market Positions?
This is where the framework gets more nuanced — and more powerful. When you're holding positions across multiple simultaneous markets (say, a Fed contract on Kalshi, a sports outcome on Polymarket, and a political event on another platform), you need to account for correlation between positions.
The practical approach used by systematic traders:
- Calculate Kelly for each position independently using the formula above
- Sum your total Kelly exposure across all open positions
- Apply a portfolio-level cap — most traders cap total Kelly exposure at 20-30% of bankroll regardless of what individual position calculations suggest
- Reduce correlated positions proportionally — if two positions move together (e.g., two contracts tied to the same Fed decision), treat them as a single combined position for sizing purposes
For a deeper look at how these positions fit into a broader portfolio framework, our prediction market portfolio strategy guide walks through diversification and correlation management in detail.
Common Kelly Criterion Mistakes in Prediction Markets
Even traders who know the formula make these errors repeatedly:
- Overestimating edge: The most common mistake. If you believe a contract is 60% likely but the true probability is 50%, Kelly recommends a large bet on a zero-edge trade. Calibrate your estimates ruthlessly.
- Ignoring transaction costs and spreads: Kalshi and Polymarket both have bid-ask spreads. These reduce your effective edge and should be factored into b before calculating f*.
- Treating Kelly as a maximum rather than an optimum: Kelly isn't a ceiling — it's the mathematical optimum. Betting significantly below Kelly also costs you growth.
- Recalculating Kelly mid-position: Kelly should be calculated at entry. Recalculating as prices move leads to erratic behavior and often results in overbetting into losers.
- Applying Kelly without a win rate baseline: If you don't have a track record, start with Quarter Kelly until you've established actual calibration data across 50+ trades.
A Step-by-Step Kelly Criterion Workflow for Prediction Market Traders
- Identify the market: Find a contract where you have a reasoned view that differs from the market price
- Estimate your true probability (p): Be explicit. Write it down. Base it on data, not intuition alone.
- Calculate the market's implied probability: This is just the contract price (e.g., a 40-cent contract implies 40%)
- Confirm positive expected value: Only proceed if your p exceeds the implied probability (accounting for the spread)
- Apply the Kelly formula: f* = (bp − q) / b
- Apply your fractional multiplier: Multiply f* by 0.5 (or 0.25 for lower confidence)
- Check portfolio exposure: Ensure total exposure across all open positions stays within your cap
- Execute and track: Log your entry probability estimate alongside the market price. Review calibration monthly.
Frequently Asked Questions About Kelly Criterion in Prediction Markets
Q: What is the Kelly Criterion in simple terms?
A: The Kelly Criterion is a formula that tells you what percentage of your bankroll to wager on a positive expected value bet in order to maximize long-run growth. The formula is f* = (bp − q) / b, where p is your win probability, q is your loss probability, and b is your net payout odds.
Q: Is the Kelly Criterion safe to use in prediction markets?
A: Full Kelly carries significant volatility risk. Most prediction market traders use Half Kelly (50% of the Kelly output), which captures most of the growth benefit with dramatically less variance. Never bet more than full Kelly.
Q: What if I don't know my exact edge?
A: Use Quarter Kelly until you have 50+ trades of calibration data. Underbetting reduces growth; overbetting risks ruin. When uncertain, err conservative.
Q: Can I use Kelly Criterion on Kalshi and Polymarket?
A: Yes. Both platforms price contracts between $0 and $1. The contract price directly gives you the market's implied probability. Calculate your own probability estimate, plug into the formula, and size accordingly — accounting for each platform's spread.
Q: How does Kelly handle losing streaks?
A: Kelly naturally handles losing streaks through bet sizing — as your bankroll decreases, your dollar bet sizes decrease proportionally. This prevents ruin. The formula is designed to be applied as a fixed fraction of your current bankroll, not a fixed dollar amount.
Q: What's the relationship between Kelly Criterion and risk management?
A: Kelly IS risk management — it's the mathematical framework that determines how much risk to take on any given position. Think of it as the engine inside a broader risk management system.
Q: How many trades do I need before trusting my Kelly calculations?
A: Most quantitative traders require at least 100 resolved trades before treating their win rate as statistically reliable. Until then, use conservative fractional Kelly and focus on calibrating your probability estimates.
Start Applying Kelly Criterion Today
The Kelly Criterion is the closest thing to a universal answer for the question every trader faces: how much should I bet? It's not complicated once you break it down — and the difference between traders who apply it and those who don't is often the difference between sustained compounding growth and account blowups.
Start simple. Pick one trade. Write down your estimated probability. Run the formula. Apply Half Kelly. Track the result. Do it again 99 more times. That feedback loop, applied consistently, is how edge compounds into a real edge.
Tools like Prevayo are built to help you do exactly this — tracking your positions, monitoring market probabilities across platforms, and surfacing the data you need to make better-calibrated decisions. The math is only as good as the inputs, and better inputs start with better market intelligence.