Prediction Market Position Sizing: Complete 2026 Guide

Prediction market position sizing is the process of calculating how much of your bankroll to allocate to a single trade based on your estimated edge, the market's implied odds, and your risk tolerance — it is the most direct mechanism for converting accurate predictions into long-term, compounding profit.

Most traders spend 90% of their time finding good predictions and 10% on how much to bet. That ratio should probably be reversed. You can have a genuine edge on every trade and still lose your bankroll if your sizing is wrong. Conversely, disciplined position sizing turns even a modest win rate into compounding, sustainable returns.

Key Takeaway: Position sizing is not a risk management afterthought — it is the core mechanism that converts predictive accuracy into actual profit. The three frameworks covered in this guide (full Kelly, fractional Kelly, and tiered conviction sizing) each serve different risk profiles, and knowing when to use each is what separates recreational traders from consistently profitable ones.

Why Does Position Sizing Matter More Than Picking Winners?

Imagine two traders. Trader A correctly predicts 60% of outcomes but bets 50% of their bankroll each time. Trader B correctly predicts only 55% of outcomes but never risks more than 5% per trade. After 100 trades, Trader B is almost certainly wealthier — and almost certainly still in the game.

This is not a hypothetical. A landmark study on professional sports bettors published in the Journal of Gambling Studies found that even bettors with genuine edges went broke at statistically significant rates due to overbetting alone — not bad predictions. The same dynamic plays out in prediction markets every day.

Prediction markets have a specific structural wrinkle that makes this even more critical: prices move. A contract you buy at 40 cents might trade at 20 cents before it resolves at $1.00. Your sizing has to account for the possibility of that drawdown without forcing you to exit a winning position at a loss. The CFTC's guidance on event contract risk specifically flags position concentration as a primary source of retail trader loss in regulated prediction market products.

What Are the Main Position Sizing Methods for Prediction Markets?

Method 1: What Is the Full Kelly Criterion and How Do You Calculate It?

The Kelly Criterion calculates the mathematically optimal bet size to maximize long-run bankroll growth. The formula is: f* = (bp − q) / b, where b is the net odds received, p is your estimated probability of winning, and q is the probability of losing (1 − p).

Example: You believe a Federal Reserve rate cut contract on Kalshi has a true 65% probability of resolving YES, but the market is pricing it at 55 cents (implying 55%). Your edge is 10 percentage points.

• b = (1 − 0.55) / 0.55 = 0.818 (the net odds)
• p = 0.65, q = 0.35
• f* = (0.818 × 0.65 − 0.35) / 0.818 = ~17.8% of bankroll

Full Kelly maximizes growth but produces extreme volatility. Even professional quant funds rarely run full Kelly. It is useful as a ceiling — a reference point for your maximum rational bet — not a number to actually deploy.

Method 2: What Is Fractional Kelly and Why Is It the Practical Standard?

Half Kelly (f* × 0.5) or Quarter Kelly (f* × 0.25) dramatically reduces variance while capturing most of the long-run growth benefit. Research by Thorp and MacLean in the Journal of Financial and Quantitative Analysis demonstrated that Half Kelly achieves approximately 75% of full Kelly's growth rate while cutting variance by 50%. For prediction markets, where your probability estimates are never perfectly calibrated, fractional Kelly provides a crucial buffer against overconfidence in your own edge calculations.

Using the Fed rate cut example above, Half Kelly would produce a position of approximately 8.9% of bankroll — still a meaningful stake, but one that survives a string of correlated losing trades or a mid-trade price dip without forcing an early exit.

Quarter Kelly (approximately 4.5% in the same example) is appropriate when:

• Your probability estimate carries high uncertainty (thin market, limited data)
• The contract has a long time horizon with significant information risk
• You are running a large portfolio of open positions simultaneously
• The market is relatively illiquid and exit options are limited

Method 3: What Is Tiered Conviction Sizing and When Should You Use It?

Tiered conviction sizing bypasses the Kelly formula entirely and assigns fixed bankroll percentages to predefined confidence bands. A common framework used by systematic prediction market traders looks like this:

• Low conviction (edge under 5%): 1% of bankroll
• Medium conviction (edge 5–10%): 3% of bankroll
• High conviction (edge over 10%): 5% of bankroll
• Maximum position cap: 5–8% regardless of estimated edge

This approach is particularly valuable for traders who find Kelly calculations difficult to execute consistently in fast-moving markets, or who are working across many simultaneous open positions. The tradeoff is that it caps upside on very high-edge opportunities, but the discipline of a hard maximum position cap is often worth that cost.

How Does Bankroll Management Interact With Position Sizing?

Position sizing frameworks only work if your total bankroll is treated as a discrete, protected pool of capital — not as a running balance you top up casually. Three bankroll management principles directly affect how you should apply any sizing formula:

1. Never risk more than 10–15% of total bankroll in open positions simultaneously. Even perfectly sized individual trades can produce correlated losses during macro uncertainty events, when multiple political or economic contracts move together.
2. Recalculate position sizes based on current bankroll, not original bankroll. A Kelly bet sized against a $10,000 bankroll after a 30% drawdown leaves you at $7,000 — the correct Kelly bet is now calculated from $7,000, not $10,000.
3. Maintain a liquidity reserve of 20–30%. Prediction markets frequently present high-edge opportunities on short notice (breaking news, policy announcements). A full-deployed bankroll cannot respond; a partially deployed one can.

What Common Position Sizing Mistakes Do Prediction Market Traders Make?

Even traders who understand Kelly and fractional sizing fall into consistent behavioral traps that erode returns over time:

• Overestimating edge: Calibration research consistently shows that people overestimate their probability accuracy by 5–15 percentage points on average. This turns a Half Kelly bet into something closer to full Kelly in practice — with full Kelly's volatility but without its theoretical justification.
• Sizing against current price, not true probability: A contract trading at 15 cents looks cheap. But if the true probability is 10%, the edge is negative and no bet is correct, regardless of how small it seems in dollar terms.
• Ignoring correlation between open positions: Holding five different contracts that all resolve YES if the Federal Reserve cuts rates is not five independent bets — it is one large correlated bet. Position sizes should be summed across correlated positions when assessing total risk exposure.
• Fixed sizing on variable-liquidity markets: A 3% position on a market with $500,000 in volume behaves very differently from a 3% position on a market with $5,000 in volume. Thin markets amplify slippage and limit your ability to exit before resolution.

How Should You Adjust Position Sizing as a Market Approaches Resolution?

Most position sizing frameworks assume static conditions, but prediction market contracts are dynamic. As a contract approaches its resolution date, the calculus changes in two important ways:

First, time-adjusted edge grows. A contract you bought at 40 cents that is now trading at 70 cents — with resolution in 48 hours — has a very different risk profile than the same contract with 30 days remaining. The expected value of adding to or holding the position is higher, and the window for adverse price movement is smaller.

Second, liquidity often contracts. As resolution approaches, market makers reduce their activity, spreads widen, and your ability to size in or out efficiently decreases. This argues for scaling into positions earlier rather than later, even at a slightly higher price, to preserve execution quality.

A practical adjustment rule: treat any position within 72 hours of resolution as fully sized — do not add to it regardless of apparent edge, because the exit window is too narrow to justify the additional exposure.

What Is the Right Position Sizing Framework for Beginners in Prediction Markets?

For traders new to prediction markets, the answer is simple: start with flat 1–2% position sizing on every trade for the first 50–100 trades. This is not optimal — it leaves growth on the table and does not account for edge variation across trades. But it accomplishes something more important: it keeps you in the game long enough to build a calibration dataset on your own probability estimates.

After 50–100 resolved trades, you can calculate your actual win rate against your predicted win rate (your Brier score, essentially) and determine whether your edge estimates are accurate enough to justify fractional Kelly sizing. Most beginners discover they were overconfident by 8–12 percentage points — which means their Kelly bets would have been approximately twice as large as was rational.

The goal of the first 100 trades is not profit maximization. It is calibration. Once you know how accurate your predictions actually are — not how accurate you feel they are — you can apply fractional Kelly with confidence.

---

Key Takeaways

• Position sizing is the primary lever of long-run profitability: A 55% accurate trader using disciplined 3–5% position sizing will outperform a 65% accurate trader using 25–50% sizing over any statistically meaningful sample of trades.
• Full Kelly is a ceiling, not a target: The Kelly Criterion calculates the mathematical maximum rational bet size; fractional Kelly (50% or 25% of full Kelly) is the appropriate practical standard because real-world probability estimates are always imperfect.
• Half Kelly delivers ~75% of Kelly growth at 50% of the variance: This tradeoff, documented by Thorp and MacLean, makes Half Kelly the dominant position sizing method among systematic prediction market traders.
• Correlated positions must be sized as a single aggregate exposure: Multiple contracts tied to the same underlying event or macro variable should be treated as one position when calculating total bankroll risk, not summed individually.
• Beginners should use flat 1–2% sizing for the first 50–100 trades to build calibration data before applying Kelly-based frameworks — most new traders overestimate their edge by 8–12 percentage points.
• A 20–30% liquidity reserve is a structural advantage: Keeping a portion of bankroll undeployed allows traders to respond to high-edge opportunities created by breaking news or sudden market mispricings that fully-deployed traders cannot access.