The Kelly Criterion is a mathematical formula that calculates the exact percentage of your bankroll to risk on any given trade in order to maximize long-run portfolio growth. Originally developed by Bell Labs researcher John L. Kelly Jr. in 1956, it remains the gold standard for position sizing in prediction markets, sports betting, and quantitative trading — because it is one of the only sizing methods with a proven theoretical optimum.
If you've ever wondered why some traders seem to compound their bankrolls steadily while others blow up chasing big positions, the answer is almost always Kelly discipline — or the lack of it. This guide covers everything: the formula, the intuition, common mistakes, fractional Kelly, and exactly how to apply it on platforms like Kalshi and Polymarket.
What Is the Kelly Criterion? (Quick Answer)
Quick Answer: The Kelly Criterion is a bet-sizing formula that tells you what fraction of your bankroll to wager on a trade given your estimated edge and the market's odds. It maximizes the expected logarithm of wealth, which translates to the fastest possible long-run compounding without risking ruin.
The core formula is deceptively simple:
f* = (bp – q) / b
- f* = fraction of bankroll to wager
- b = net odds received on the bet (i.e., profit per dollar risked)
- p = your estimated probability that the outcome is YES
- q = probability the outcome is NO (q = 1 – p)
Let's make this concrete with a Kalshi example. Suppose there's a contract asking "Will the Fed cut rates in May 2026?" trading at 40¢ (implying a 40% market probability). You've done your research and believe the true probability is 60%. Here's your Kelly calculation:
- b = 1.50 (buying at 40¢ pays out 60¢ profit per dollar risked, so b = 60/40 = 1.5)
- p = 0.60 (your estimated probability)
- q = 0.40
- f* = (1.5 × 0.60 – 0.40) / 1.5 = (0.90 – 0.40) / 1.5 = 0.50 / 1.5 = 33.3%
Kelly says to wager 33.3% of your bankroll. On a $1,000 account, that's $333.
Why Kelly Beats Fixed Bet Sizing
Quick Answer: Fixed bet sizing (e.g., always risking $50 per trade) ignores both your edge and your odds, causing either chronic under-betting when edge is large or over-betting that risks ruin. Kelly dynamically scales your position to the strength of your edge — the bigger the mispricing, the bigger the stake.
Consider two traders with identical win rates on identical Kalshi markets:
- Trader A uses flat $50 bets regardless of edge
- Trader B uses Kelly sizing
Over 100 trades, Trader B's bankroll compounds geometrically while Trader A's grows arithmetically at best. The difference is not luck — it's mathematics. Kelly's original 1956 paper in the Bell System Technical Journal proved that no system can outperform Kelly in the long run without accepting additional risk of ruin.
The Most Common Kelly Mistake: Overestimating Your Edge
Quick Answer: Kelly's output is only as good as your probability estimate. Overconfident probability estimates produce over-sized positions — and over-betting is more dangerous than under-betting because the downside is asymmetric. A bad estimate that says 70% when true odds are 55% can produce a Kelly recommendation that destroys your bankroll.
This is the real reason most professional gamblers and quant traders use fractional Kelly — typically betting 25–50% of the full Kelly recommendation. Here's what fractional Kelly does in practice:
- Full Kelly (f* = 33%): Maximum theoretical growth rate, but extreme volatility in portfolio value
- Half Kelly (f* = 16.5%): Achieves ~75% of full Kelly's growth rate with roughly half the drawdown
- Quarter Kelly (f* = 8.3%): Very conservative, ideal for new traders or high-uncertainty markets
For most prediction market traders, Half Kelly is the practical sweet spot. You capture most of the compounding benefit while protecting yourself against estimation error — which is inevitable, especially in fast-moving political and economic markets.
Applying Kelly to Prediction Markets: Step-by-Step
Quick Answer: To apply Kelly to a prediction market trade, you need three inputs: the current contract price (which gives you the implied market probability), your independent probability estimate, and your bankroll. The formula then outputs your optimal position size as a percentage of capital.
Step 1: Establish Your Independent Probability Estimate
This is where the real work happens. Do not anchor to the market price. Use polling data, historical base rates, modeling, or news analysis to form a view before looking at the contract price. On Polymarket, Kalshi, and other platforms, the market price is just another trader's opinion — often a well-informed one, but not gospel.
Step 2: Calculate Your Edge
Edge = Your probability – Market implied probability. If a Kalshi contract is priced at 55¢ and you estimate 65% probability of YES, your edge is 10 percentage points. No edge means no trade — Kelly will correctly output 0 (or negative, signaling the other side).
Step 3: Run the Kelly Formula
Plug your numbers in: f* = (bp – q) / b. Use your independent probability estimate for p, and derive b from the market price (b = (1 – price) / price for a binary YES position).
Step 4: Apply Your Fractional Multiplier
Multiply the Kelly output by your chosen fraction (0.5 for half Kelly is recommended for most traders). This is your position size as a percent of bankroll.
Step 5: Set Position Limits
Even Kelly can produce large single-position recommendations. Most risk-conscious traders cap any single position at 10–20% of bankroll regardless of Kelly's output — especially important if your edge estimate has wide uncertainty bands. For a deeper look at overall risk controls, see our Prediction Market Risk Management: The Complete Guide.
Kelly in Multi-Market Portfolios
Quick Answer: When trading multiple prediction market contracts simultaneously, the simple Kelly formula needs adjustment because positions are not independent — capital committed to one trade is unavailable for others. The solution is to treat your total open exposure as the relevant constraint and size each position relative to remaining available capital.
A practical rule of thumb for multi-market Kelly:
- Run Kelly independently for each open opportunity
- Sum all recommended position sizes — if they exceed 100% of bankroll, scale down proportionally
- Prioritize markets with the highest Kelly recommendations (highest edge-to-odds ratio)
- Reserve 20–30% of bankroll as a liquidity buffer for new opportunities
If you're actively trading across Kalshi and Polymarket simultaneously, this portfolio-level Kelly discipline becomes critical. Our complete beginner's guide to trading prediction markets covers how to structure a multi-platform approach from the ground up.
Real Market Example: March Madness Bracket Markets
During the 2026 NCAA Tournament, Kalshi listed individual game winner contracts. A trader who correctly identified that a #5 seed was undervalued at 30¢ against a vulnerable #4 seed — and estimated the true probability at 48% — would run the following Kelly calculation:
- b = (1 – 0.30) / 0.30 = 2.33
- p = 0.48, q = 0.52
- f* = (2.33 × 0.48 – 0.52) / 2.33 = (1.12 – 0.52) / 2.33 = 0.60 / 2.33 = 25.7%
- Half Kelly: 12.9% of bankroll
On a $2,000 bankroll, that's a $258 position. If the #5 seed wins (as your model predicted), the $258 position returns $602 — a $344 profit. The Kelly framework told you exactly how much confidence to put behind your conviction without overexposing yourself to a single tournament game.
According to the CFTC's 2023 ruling expanding event contract markets, regulated prediction markets like Kalshi operate under federal oversight — meaning position sizing discipline matters not just for profitability, but for operating within a legitimate, licensed framework.
When Kelly Says Don't Trade
One of Kelly's most underappreciated features is that it naturally filters out bad trades. If your estimated probability is lower than the market implied probability, Kelly outputs a negative number — which means the edge is on the other side. If your estimate exactly matches the market, Kelly outputs zero. Only when you have a genuine, quantifiable edge does Kelly recommend a non-zero position.
This makes Kelly an excellent trade filter, not just a sizing tool. Before placing any prediction market trade, run the numbers. If Kelly says 0% or negative, that's a signal to pass — not to rationalize a position.
FAQ: Kelly Criterion in Prediction Markets
What is the Kelly Criterion in simple terms?
The Kelly Criterion is a formula that tells you what percentage of your bankroll to bet on a trade based on your edge and the market odds. It's designed to maximize long-run growth without risking going broke.
Can I use the Kelly Criterion on Kalshi?
Yes. Kalshi's binary contracts (YES/NO at prices between 1¢ and 99¢) map directly onto the Kelly formula. The contract price gives you the market-implied probability, and you supply your own estimate to calculate edge and position size.
What is fractional Kelly and should I use it?
Fractional Kelly means betting a fixed fraction (commonly 50%) of the full Kelly recommendation. It reduces portfolio volatility significantly while retaining most of the compounding benefit. Most experienced prediction market traders use Half Kelly or less.
What happens if I bet more than Kelly recommends?
Over-betting the Kelly amount reduces your long-run growth rate and increases drawdown risk. Betting exactly 2x Kelly produces the same expected growth as not betting at all — with dramatically higher variance. Betting more than 2x Kelly leads to expected ruin over time.
How do I estimate probabilities for the Kelly formula?
Use independent research: historical base rates, polling data, fundamental analysis, or quantitative models. The key is forming a view before looking at the market price to avoid anchoring bias. Track your estimates over time to calibrate your accuracy.
Is the Kelly Criterion the best position sizing method for prediction markets?
For traders with reliable probability estimation skills, Kelly is theoretically optimal. For beginners or those in high-uncertainty markets, fractional Kelly or fixed fractional sizing (e.g., never risking more than 5% per trade) may be more practical until your edge estimation improves.
Putting It All Together
The Kelly Criterion is not magic — it's math applied to discipline. It won't improve your probability estimates, but it will ensure that when your estimates are right, you're positioned to compound those wins efficiently, and when you're wrong, the damage is contained. That asymmetry is the entire point.
Start with the basic formula on paper trades. Build your edge estimation skills. Graduate to Half Kelly on live markets. Add position caps. Track your actual win rates against your estimates and recalibrate. That feedback loop — estimate, size, execute, review — is what separates systematic traders from gamblers in prediction markets.
For traders who want help automating this process — tracking open positions, calculating Kelly sizes across multiple markets, and flagging high-edge opportunities — tools like Prevayo are built specifically for prediction market analytics, bringing quantitative rigor to a market where most participants are still sizing positions by gut feel.