Kelly Criterion for Polymarket: Never Blow Up Your Bankroll Again

Target Keywords: "polymarket kelly criterion", "polymarket position sizing", "prediction market bankroll sizing", "kelly criterion prediction markets"

Introduction

Most Polymarket traders don't blow up because they're bad at predicting outcomes. They blow up because they bet too much on any single market.

The Kelly Criterion is a mathematical formula that tells you exactly how much of your bankroll to risk based on your perceived edge. It was developed by mathematician John Kelly in 1956 and has been used by professional gamblers, traders, and investors ever since. Applied to Polymarket, it's the most principled answer to the question every trader faces: "How much should I put on this market?"

This guide explains the Kelly formula, how to apply it to Polymarket's binary markets, and why you should almost always bet less than Kelly tells you to.

For the broader context of prediction market bankroll management, see the complete bankroll management guide for prediction markets. For tracking your results across markets, read the Polymarket deposits and withdrawals guide.

Why Position Sizing Matters More Than Win Rate

Consider two traders, both of whom correctly predict 60% of Polymarket outcomes:

Trader A bets 30% of their bankroll each time. After a string of three losses (a reasonable occurrence with 40% loss probability), they've lost 65% of their capital. Recovery requires nearly doubling what's left.

Trader B bets 5% of their bankroll each time. After three consecutive losses, they've lost 14% of their capital. Their edge can recover that easily.

Same prediction accuracy. Completely different outcomes. This is why position sizing isn't a secondary concern — it's the primary determinant of whether a trader survives long enough for their edge to pay off.

The Kelly Formula

The Kelly Criterion formula:

f = (bp - q) / b

Where:

• f = fraction of bankroll to wager
• b = net odds received (what you win per dollar risked)
• p = your estimated probability of winning
• q = probability of losing (1 - p)

Adapting Kelly for Prediction Market Shares

On Polymarket, you're not betting at fixed odds — you're buying shares priced between $0.01 and $0.99 that pay $1.00 if correct. Here's how to translate share prices into Kelly inputs:

If you buy YES shares at a price of $0.40:

• You pay $0.40 per share
• If correct, you receive $1.00 per share — a profit of $0.60
• Net odds: b = $0.60 / $0.40 = 1.5

If you buy YES shares at a price of $0.70:

• You pay $0.70 per share
• If correct, you receive $1.00 per share — a profit of $0.30
• Net odds: b = $0.30 / $0.70 ≈ 0.43

Example Calculation

You believe a market priced at $0.40 (40% implied probability) has a true probability of 65%.

• b = 0.60 / 0.40 = 1.5
• p = 0.65 (your estimate)
• q = 0.35 (1 - 0.65)

f = (1.5 × 0.65 - 0.35) / 1.5
f = (0.975 - 0.35) / 1.5
f = 0.625 / 1.5
f = 0.417

Full Kelly says bet 41.7% of your bankroll. Do not do this.

Why You Should Use Fractional Kelly

Full Kelly is mathematically optimal only if:

1. Your probability estimate is perfectly accurate
2. You're playing an infinite number of independent bets
3. You experience no variance in the short term

None of these conditions hold in prediction market trading.

Your probability estimates are uncertain. Your edge might be smaller than you think. Short-term variance can destroy a bankroll before the long-run math plays out.

This is why experienced traders use Half Kelly or Quarter Kelly:

In the example above (Full Kelly = 41.7%):

• Half Kelly: 20.8% — still aggressive
• Quarter Kelly: 10.4% — more reasonable for most traders
• Tenth Kelly: 4.2% — conservative, similar to the flat 5% percentage rule

The Overbetting Asymmetry

Kelly has an important asymmetry: overbetting is much more dangerous than underbetting.

If your true optimal Kelly bet is 10% and you bet 20%, your long-term growth rate is significantly damaged even though you're only 2x overbetting. If you bet 5% (half of optimal), your growth rate is reduced modestly but you're protected from variance.

The practical conclusion: when uncertain about your edge, bet less than Kelly suggests.

Estimating Your True Probability

The Kelly formula is only as good as your probability estimate. This is where most traders go wrong — they don't have a rigorous method for estimating true probabilities.

Anchoring to Market Price vs. Own Research

Before calculating Kelly, ask: Why do I believe the true probability differs from the market price?

If your answer is "I just feel strongly about this" — that's not an edge. Kelly assumes you have information or analysis the market doesn't fully reflect.

If your answer is "I have domain expertise, primary source data, or a systematic analysis that shows the market is mispriced" — that's the foundation for a valid Kelly calculation.

Calibrating Over Time

The best traders track their estimated probability vs. outcome over time to test their calibration. If you consistently estimate 70% probability for events that resolve correctly 55% of the time, your Kelly inputs are inflated and your bets should be smaller.

Manual tracking of every trade — including your confidence estimate at entry — is how you measure and improve your calibration. Record in your bankroll journal:

• Market name and entry price
• Your estimated probability
• Outcome (correct or incorrect)
• Implied edge vs. actual result

After 50+ trades, you'll have data on whether your estimates are accurate.

Kelly With Multiple Simultaneous Positions

One complication: Polymarket traders often hold multiple open positions at the same time. The basic Kelly formula assumes a single sequence of independent bets, not a portfolio of simultaneous positions.

When you have multiple positions, each individual Kelly percentage must be reduced to account for correlation and total exposure. Simple rule:

Never let total active exposure exceed 25-30% of your bankroll regardless of what individual Kelly calculations suggest.

This is because:

1. Prediction market positions are often correlated (political markets move together)
2. Capital tied in unresolved positions isn't available for new opportunities
3. A series of simultaneous losses can occur when correlated events move against you

For the full framework on managing exposure across multiple markets, see the bankroll management guide.

When Kelly Says "Don't Trade"

An important output of the Kelly formula that traders often ignore: Kelly equals zero or below when you have no edge.

If a market is priced at $0.65 and you believe the true probability is 60%, Kelly calculation:

• b = 0.35 / 0.65 ≈ 0.54
• p = 0.60, q = 0.40

f = (0.54 × 0.60 - 0.40) / 0.54
f = (0.324 - 0.40) / 0.54
f = -0.076 / 0.54
f = -0.14

Negative Kelly means the market is priced in your favor less than you think. The correct position size is zero — don't trade this market.

This is one of the most valuable outputs of the Kelly framework. It forces you to explicitly calculate whether you have a positive expected value before entering a market, rather than trading on vague confidence.

Practical Kelly Implementation

Here's a simple process for applying Kelly to each Polymarket trade:

Step 1: Record the current share price (this is the market's implied probability).

Step 2: Write down your estimated true probability and the specific reason you believe it differs from the market price.

Step 3: Calculate net odds b = (1 - price) / price.

Step 4: Calculate Kelly fraction f = (b × p - q) / b.

Step 5: Apply your fractional Kelly multiplier (typically 0.25 for most traders).

Step 6: Compare to your maximum percentage rule (e.g., never exceed 5% of bankroll per position). Use the more conservative of the two numbers.

Step 7: Record everything in your bankroll journal before entering the trade.

The act of writing it down before you trade creates accountability. If you can't articulate your probability estimate and the reasoning behind it, you're not ready to trade the market.

Use the compounding calculator to model how consistent Kelly-based position sizing affects your bankroll growth over time compared to flat betting.

Tracking Kelly Performance

After each market resolves, add to your journal:

• Was your edge estimate correct in direction? (You thought >50% when price was below 50%, and it resolved YES)
• Was your edge magnitude accurate? (You estimated 70%, it implied 40%, did your 70% feel right in retrospect?)
• What would Full Kelly have been? What did you actually bet?
• What was your profit or loss?

Monthly, calculate your average edge capture: how often your direction was correct vs. how often it should have been based on implied probabilities. Traders with real edges have positive expected value that compounds over hundreds of trades — and the data to prove it.

Responsible Trading

Kelly Criterion is a tool for managing risk, not eliminating it. Even with optimal sizing, prediction market trading involves losing positions. The goal is to stay in the game long enough for your edge to express itself over a large sample of trades.

Never risk money you can't afford to lose. If a losing streak causes financial stress, your bankroll is too large — scale down regardless of what Kelly says.

Disclaimer

This article is for educational and informational purposes only. It does not constitute financial advice or a recommendation to trade on Polymarket, Kalshi, or any other prediction market. Prediction markets involve significant risk including total loss of capital. The Kelly Criterion is a theoretical framework for risk management and does not guarantee profitable outcomes. Your probability estimates may be incorrect. Past performance does not guarantee future results. Manage Bankroll is a personal finance tracking tool for manually recording numbers. It does not connect to, integrate with, or access any prediction market platform or external financial service.