Prediction Market Overround: Why the Odds Don't Add Up to 100%

TL;DR

• Overround is the amount by which a market's outcome prices exceed 100% when added at face value. A three-candidate market pricing at 45, 35, and 25 cents sums to 105 cents, a 5-point overround.
• It isn't sportsbook vig. Sportsbook vig is a fixed house markup that doesn't shrink with volume. Prediction market overround emerges from order-book competition between traders and narrows as liquidity deepens.
• To find a market's true implied probability, divide each outcome's price by the sum of all outcomes' prices. The three example prices above become 42.9%, 33.3%, and 23.8%, which sum to exactly 100%.
• The overround rarely reaches zero, even on the most liquid markets. What changes with liquidity and time to resolution is the size of the gap, not whether it exists.

Ever added up the prices on a multi-candidate prediction market and landed on 104 or 105 cents instead of a clean dollar? You didn't miscount.

That gap between the sum of every outcome's price and 100% is the overround, also called the vig. It shows up because platforms build in a small edge for the traders and market makers absorbing risk on the other side of every position. Until you correct for it, the raw price you're looking at isn't the market's true probability estimate, just its face value.

What Is the Prediction Market Overround?

Say a market has three candidates for an election. Candidate A trades at 45 cents, Candidate B at 35 cents, Candidate C at 25 cents. Add those up and you get 105 cents, or 105%. That extra 5 points is the overround.

In a perfectly efficient two-sided market, Yes and No on a single binary contract sum to roughly $1.00 before fees. But once you're looking across multiple competing outcomes in the same prediction market, resolved by which one wins, the sum tends to drift above 100%.

The gap isn't a bug. It compensates market makers and the platform for taking on the other side of every position, and it widens further when a market is thin.

Is Prediction Market Vig the Same as Sportsbook Vig?

No, and the difference matters for how you read the number.

Sportsbook vig is a fixed markup a bookmaker sets on both sides of a bet before anyone places a position. A sportsbook prices a coin-flip game at -110/-110 specifically so it profits regardless of outcome. The vig is baked into the line by the house, and it stays there no matter how much money comes in.

Prediction markets don't work that way. There's no house setting the price. Traders on Polymarket, Kalshi, Limitless, Predict.Fun, and Opinion set prices against each other through direct order matching.

The overround emerges from the mechanics of multi-outcome order books, not from a bookmaker's markup. Thin liquidity, wide bid-ask spreads, and a market maker's need to be compensated for risk all push the sum above 100%.

As more traders enter and liquidity deepens, the overround tends to compress, because competition narrows the spread. A sportsbook's vig doesn't compress with volume. It's fixed by design, and that's the core distinction: sportsbook vig tells you what the house is charging, while prediction market overround tells you how efficiently that specific market is pricing risk right now.

How Do You Calculate the True Implied Probability?

If you want the market's actual implied probability, not the raw price, you have to strip the overround out. The standard method is proportional normalization: divide each outcome's price by the sum of all outcomes' prices.

Take the three-candidate example above: 45, 35, and 25 cents, summing to 105. Divide each by 1.05. Candidate A becomes 42.9%, Candidate B becomes 33.3%, Candidate C becomes 23.8%. Those three now sum to exactly 100%, and that's the number worth comparing against a poll, a model, or the same race priced on a different platform.

This matters most when you're checking odds across Polymarket, Kalshi, Limitless, Predict.Fun, and Opinion side by side. Each platform's order book carries a different overround depending on its liquidity and how many market makers are active on that specific event.

A candidate priced at 45 cents on one platform and 43 cents on another isn't necessarily a real disagreement. Normalize both first. Sometimes what looks like platforms disagreeing closes entirely once you've backed out each platform's respective overround, and the gap was just two different vig levels.

Why Does the Overround Change Over Time?

The size of the overround moves with liquidity and time to resolution. A newly listed market with few participants can show an overround of 10% or more, since only a handful of market makers are pricing risk and each demands a wider cushion.

As trading volume builds and more traders take positions on both sides, that cushion typically narrows. Markets also tend to tighten as they approach resolution, when uncertainty about the outcome shrinks and the remaining risk to a market maker is lower.

This is also why a stale overround reading is close to useless. A number pulled hours ago on a fast-moving event tells you what liquidity looked like then, not now. PredictionHero currently tracks 342,146 markets across 32,747 events and matches 3,061 of them across all five platforms, which is enough live coverage that pulling a current order book instead of a cached one is always the better move before normalizing a price.

How Do the Five Platforms Handle Multi-Outcome Pricing?

Regulatory structure doesn't determine the size of a given market's overround. Liquidity does.

But regulatory structure does affect who's providing that liquidity and how, which is why the same event can price differently across venues even after normalization.

Where Can You Check Normalized Odds Across Platforms?

Comparing a raw price against a poll or a model only works once the overround is stripped out, and doing that by hand for five platforms on every event gets old fast. Check current order books and cross-platform consensus pricing on PredictionHero.

Sources

• CFTC. "Understanding Prediction Markets and Event Contracts." cftc.gov

PredictionHero aggregates publicly available prediction market data for informational purposes only. This is not financial advice. Prediction markets may not be available in all jurisdictions.