Prediction Market Position Sizing: How to Size Every Contract

Why Prediction Market Position Sizing Matters More Than Picking Winners

Finding +EV contracts is the first step. Prediction market position sizing is what determines whether that edge compounds into profit or implodes into a blown bankroll. Two traders can hold the same five contracts with the same probability estimates. The one who sizes correctly makes money. The other goes broke.

The core problem: prediction markets are not sportsbooks. You are not placing isolated bets that settle in a few hours. You are building a portfolio of binary contracts that may sit open for weeks or months, many of which share underlying risk drivers. Position sizing in this context requires tools that go beyond the basic Kelly Criterion. You need fee adjustments, correlation discounts, and liquidity awareness.

This article covers the complete prediction market position sizing framework. Every formula includes a worked example with real numbers.

Kelly Criterion Adapted for Binary Contracts

The Kelly Criterion is the starting point. For a prediction market contract priced at c (the cost of a Yes contract) where you estimate the true probability at p:

Kelly % = p - c / (1 - c) x (1 - p)

This simplifies to the standard Kelly formula when you convert the contract price to decimal odds. The decimal odds of a binary contract are 1 / c. So buying a $0.55 contract is equivalent to betting at decimal odds of 1.818.

Worked example: sizing a 55-cent contract

You find a Kalshi contract priced at $0.55 (Yes). Your model estimates the true probability at 62%.

Step 1: Convert to decimal odds. Decimal odds = 1 / 0.55 = 1.818

Step 2: Run Kelly. Kelly % = (0.62 x 1.818 - 1) / (1.818 - 1) = (1.127 - 1) / 0.818 = 0.127 / 0.818 = 15.5%

On a $10,000 bankroll, full Kelly says allocate $1,550 to this position. That means buying 2,818 contracts at $0.55 each.

Step 3: Apply fractional Kelly. Full Kelly is reckless. Half Kelly captures roughly 75% of the growth rate with dramatically less variance. Half Kelly: $1,550 x 0.50 = $775.

That is 1,409 contracts. This is your starting position size before fee and correlation adjustments. Run any contract through the PM EV calculator to get Kelly sizing automatically.

Fee-Adjusted Kelly: How Platform Fees Change Optimal Size

Kalshi charges approximately 7% on winner profits. Polymarket has no explicit fee but charges through the bid-ask spread. Both reduce your effective edge, which reduces your optimal position size.

Ignoring fees means you systematically oversize every position. Over hundreds of trades, that cumulative oversizing compounds into real capital destruction.

Fee adjustment for Kalshi

For a contract at price c with Kalshi's 7% winner fee, your net profit on a win drops from (1 - c) to (1 - c) x 0.93. The effective decimal odds become:

Fee-adjusted odds = ((1 - c) x 0.93 + c) / c

For the $0.55 contract:

• Pre-fee profit per contract: $0.45
• Post-fee profit per contract: $0.45 x 0.93 = $0.4185
• Effective decimal odds: ($0.4185 + $0.55) / $0.55 = 1.761

Fee-adjusted Kelly: = (0.62 x 1.761 - 1) / (1.761 - 1) = (1.092 - 1) / 0.761 = 0.092 / 0.761 = 12.1%

Full Kelly dropped from 15.5% to 12.1%. Half Kelly drops from $775 to $605. The 7% fee shaved 22% off your optimal position size. On thinner edges, the reduction is even steeper.

Read the full breakdown of how prediction market fees eat your edge. The fee calculator computes the net impact for any contract price on either platform.

Fee adjustment for Polymarket

Polymarket has no winner fee, but the bid-ask spread acts as an entry cost. If the best ask is $0.56 when the mid-price is $0.55, you effectively pay a 1-cent premium. Adjust your entry price to $0.56 for the Kelly calculation. On liquid markets this is negligible. On thin markets with 3-5 cent spreads, it changes the math substantially.

Step 1Estimate true probability

→

Step 2Convert contract price to decimal odds

→

Step 3Subtract platform fees from payout

→

Step 4Run Kelly on fee-adjusted odds

→

Step 5Apply half Kelly (0.50 multiplier)

→

Step 6Check correlation and liquidity limits

Correlated Position Risk: The Portfolio Killer

Here is where most prediction market traders get destroyed. You hold five contracts. Each one is individually +EV and individually sized at 4% of bankroll using half Kelly. Total exposure: 20%. Feels diversified.

But three of those five contracts are political. "Democrats win the presidency," "Democrats win the Senate," "Democrat wins Arizona." These are not three independent bets. They are one bet on Democratic voter turnout wearing three disguises. If turnout disappoints, all three lose simultaneously. Your actual risk is not 12% (3 x 4%). It is closer to the risk of a single 10-12% position because the correlated positions move together.

The correlation discount formula

For two positions with estimated correlation r above 0.5, multiply each Kelly-recommended size by:

Adjustment factor = 1 - r/2

At r = 0.70 (typical for same-party political contracts): factor = 0.65. A $600 half-Kelly position becomes $390.

At r = 0.40 (moderate, like two contracts in the same economic sector): factor = 0.80. A $600 position becomes $480.

At r = 0.0 (independent): no adjustment needed.

Worked example: portfolio of 5 political contracts, 3 correlated

You have a $20,000 bankroll and five open positions:

Without correlation adjustment: total exposure = $5,000 (25.0% of bankroll). Looks reasonable.

With correlation adjustment: the three Democratic contracts have pairwise correlations of approximately 0.65-0.75. Using r = 0.70 as the average, the adjustment factor is 0.65.

• Dems win presidency: $1,200 x 0.65 = $780
• Dems win Senate: $900 x 0.65 = $585
• Dems win Arizona: $1,000 x 0.65 = $650
• Fed cuts by June: $800 (uncorrelated with others, no adjustment)
• Bitcoin above 100K: $1,100 (uncorrelated with others, no adjustment)

Adjusted total exposure: $3,915 (19.6% of bankroll).

The three Democratic positions dropped from $3,100 combined to $2,015. That is a 35% reduction. The effective number of independent positions went from 5 to approximately 3.4 after accounting for the shared political driver.

This is not conservative. It is correct. Run your full portfolio through the correlation calculator to see your true effective exposure.

Sizing Across Multi-Outcome Markets

Multi-outcome markets add a structural complication. In a primary with 6 candidates, buying multiple contracts means your positions are inherently correlated: if Candidate A wins, Candidates B through F all lose. You cannot win on more than one.

The naive approach is to size each candidate contract independently with Kelly. This ignores the negative correlation between contracts in the same market and leads to overallocation.

Worked example: 5-candidate primary

You identify three underpriced candidates after normalizing the overround:

Total Kelly allocation: 11.1% of bankroll. But these are mutually exclusive outcomes. At most one wins. The expected loss when your chosen candidate loses is higher than the independent model assumes because all three positions lose together when a candidate outside your set wins.

The practical rule: cap total allocation to a single multi-outcome market at 1.5x the Kelly size of your largest individual position. Here, the largest is 5.2%, so the market cap is 7.8%. Scale all three positions proportionally downward:

• Candidate A: 5.2% x (7.8 / 11.1) = 3.7%
• Candidate C: 3.1% x (7.8 / 11.1) = 2.2%
• Candidate D: 2.8% x (7.8 / 11.1) = 2.0%

Use the multi-outcome calculator to identify the mispriced contracts and size across the full field.

Liquidity Constraints: Can You Actually Fill the Order?

Kelly tells you the optimal size. The order book tells you what is possible. A $5,000 position on a contract with $800 of liquidity at your target price means you fill at progressively worse prices. Slippage eats your edge.

Slippage math

If the best ask is $0.55 with 500 contracts available, the next level is $0.56 with 300 contracts, and the next is $0.57 with 200 contracts, a 1,000-contract order fills at a weighted average price of:

(500 x $0.55 + 300 x $0.56 + 200 x $0.57) / 1,000 = $0.558

That 0.8-cent slippage on a $0.55 contract reduces your edge by 1.5 percentage points. If your estimated edge was 4%, slippage just cut it to 2.5%. Your Kelly size should be recalculated on the slippage-adjusted entry price, not the displayed best ask.

The liquidity cap rule

Never allocate more than 10% of the available daily volume to a single position. If a contract trades $15,000 per day, your maximum position is $1,500 regardless of what Kelly recommends. Exceeding this moves the market against you and signals your intent to other participants.

The liquidity calculator simulates slippage across the order book before you commit capital.

The Equal-Weighting Mistake

The most common position sizing error: allocating the same dollar amount to every contract. $500 on this one, $500 on that one. Five positions, $2,500 total. Clean and simple.

It is also wrong. Equal weighting ignores three things that should drive allocation:

1. Edge magnitude. A contract where you estimate 8% edge deserves a larger allocation than one with 2% edge. Kelly scales with edge for a reason. Equal weighting overallocates to weak edges and underallocates to strong ones.

2. Odds structure. A $0.10 contract and a $0.80 contract have completely different risk profiles. $500 on a $0.10 contract is 5,000 contracts with a potential $4,500 payout. $500 on a $0.80 contract is 625 contracts with a potential $125 payout. Same dollar amount, vastly different risk-reward.

3. Correlation. Equal weighting treats all positions as independent. They are not. Three equally weighted political contracts create a concentrated directional bet that an edge-weighted, correlation-adjusted portfolio would have flagged and reduced.

Kelly with fee and correlation adjustments solves all three problems. The formula is doing real work. Use it.

Practical Position Limits

Theory gives you a formula. Practice gives you guardrails. These rules prevent the formula from producing dangerous outputs when inputs are uncertain.

Single position cap: 2-5% of bankroll. If half Kelly recommends more than 5%, your probability estimate is either very confident or very wrong. Verify the estimate before sizing above 5%. Most professional prediction market traders cap at 3%.

Total portfolio exposure cap: 25% of bankroll. After all correlation adjustments, your open positions should not exceed 25% of bankroll. The remaining 75% is dry powder for new opportunities and a buffer against estimation error.

Correlation group cap: 10% of bankroll. No single risk driver should account for more than 10% of your total exposure. If "Democratic turnout" drives 15% of your portfolio, you are running a political fund, not a diversified book.

Edge confidence scaling. When your edge estimate comes from a backtested model with 500+ observations, use half Kelly. When it comes from gut feel or limited data, use quarter Kelly. When you are genuinely unsure, do not trade. No position is also a position.

These limits interact with each other. A single position might pass the 5% cap but push a correlation group above 10%. Always check portfolio-level metrics after sizing each new position. The correlation calculator and Kelly Criterion tool handle this math. Read the Kelly Criterion guide for the foundational formula and the correlated positions guide for portfolio-level adjustments.

For the broader question of how much total capital to deploy on prediction markets, see prediction market bankroll management. That article covers total bankroll sizing. This article covers how to divide that bankroll across individual positions.

For a complete framework on combining position sizing with market selection and timing, read the prediction market strategy guide.

Correlation-Adjusted Sizing: Two Worked Examples

The correlation discount formula above covers the theory. Here are two practical scenarios that show how correlation adjustment changes real portfolio decisions.

Example 1: Economic cluster with mixed correlations

You hold three macro contracts on a $25,000 bankroll:

CPI and Fed rates have a strong correlation (r = 0.60). High inflation makes rate cuts less likely. Adjustment factor: 1 - 0.60/2 = 0.70.

Gold correlates moderately with both (r = 0.35), but below the 0.5 threshold for the simple discount. No adjustment needed for gold on its own. However, your total macro cluster exposure matters.

Adjusted positions:

• CPI above 3.5%: $1,250 x 0.70 = $875 (3.5%)
• Fed holds rates: $900 x 0.70 = $630 (2.5%)
• Gold above $2,500: $700 (2.8%, no adjustment)

Pre-adjustment cluster exposure: $2,850 (11.4%). Post-adjustment: $2,205 (8.8%). The cluster stays under the 10% cap without further reduction.

Example 2: Sports event with hidden correlation

Two NBA playoff contracts:

These look independent. One is about winning the championship. The other is about a series going long. But if the Celtics lose in 4 or 5 games, both contracts lose. A quick Celtics exit kills both positions. The correlation is conditional on the Celtics losing early, estimated at r = 0.45.

At r = 0.45, the adjustment factor is 1 - 0.45/2 = 0.775. Borderline, but worth applying:

• Celtics win Finals: $1,100 x 0.775 = $853
• Series Over 5.5: $600 x 0.775 = $465

The $383 reduction seems small, but across 20-30 open positions with various hidden correlations, these adjustments compound into meaningful risk reduction. Run any portfolio combination through the correlation calculator to surface dependencies you might miss.

The Position Sizing Pipeline

Every contract you trade should pass through this sequence:

1. Estimate the true probability. Use expected value analysis to confirm the contract is +EV.
2. Subtract platform fees. Use the fee calculator to get fee-adjusted odds.
3. Run Kelly on the fee-adjusted numbers. Use the Kelly Criterion calculator.
4. Apply half Kelly (or quarter Kelly if edge confidence is low).
5. Apply the correlation discount for any positions sharing a risk driver. Use the correlation calculator.
6. Check liquidity. Use the liquidity calculator to verify you can fill without excessive slippage.
7. Verify portfolio-level caps: single position under 5%, correlation group under 10%, total exposure under 25%.

Skip any step and you are guessing. The math is not complicated. The discipline to run it every time is what separates profitable traders from the 90% who lose money.

If you are also trading sportsbooks alongside prediction markets, cross-platform arbitrage can generate positions that require their own sizing framework. And understanding how prediction markets work is the prerequisite for all of this.

Frequently asked questions

How much should I risk on a single prediction market contract?

Use half Kelly as a starting point, then adjust for fees and correlation. Most professional traders cap individual positions at 2-5% of bankroll. If Kelly recommends more than 5% even at half Kelly, verify your probability estimate carefully. Oversized recommendations usually indicate overestimated edge.

How do Kalshi fees affect position sizing?

Kalshi's 7% winner fee reduces your effective payout and therefore your optimal position size. On a $0.55 contract, the fee drops full Kelly from 15.5% to 12.1% of bankroll, a 22% reduction. Always calculate Kelly on fee-adjusted odds, not raw contract prices. The fee calculator computes this automatically.

What is the correlation discount for position sizing?

When two positions share a risk driver with correlation r above 0.5, multiply each Kelly-recommended size by (1 - r/2). At r = 0.70, each position is reduced to 65% of its standalone size. This prevents overexposure to a single risk factor disguised as multiple independent bets.

Should I equal-weight all my prediction market positions?

No. Equal weighting ignores edge magnitude, odds structure, and correlation. A contract with 8% edge deserves a larger allocation than one with 2% edge. Kelly naturally scales position size to edge strength. Equal weighting systematically overallocates to weak trades and underallocates to strong ones.

How does liquidity affect position sizing in prediction markets?

Never allocate more than 10% of a contract's daily volume to a single position. Large orders cause slippage that erodes your edge. A 0.8-cent slippage on a $0.55 contract cuts a 4% edge to 2.5%. Always simulate order book impact before committing capital.