Why Prediction Market Odds Can Add Up to More Than 100%

You open a prediction market for an upcoming MMA fight. Oliveira is at 34%. Holloway is at 94%. That is 128% combined. In a binary market where exactly one outcome can win, the probabilities should add up to 100%. So either the platform is broken, or something else is going on.

This is not a bug. It is a natural consequence of how prediction markets work under the hood. The prices you see are not consensus probabilities — they are the cheapest available offers on two separate order books. Understanding the difference between a displayed price and a true probability is one of the most important concepts in prediction market trading.

This phenomenon has a name: overround (also called vigorish or the book margin). It shows up in prediction markets, sports betting, and any market where each outcome has its own order book. The size of the overround tells you something valuable about the market — specifically, how much it costs to trade and how much disagreement exists among participants.

On Polymarket, every binary market has two tokens: Yes and No. Each token has its own order book with its own bids and asks. The Yes order book is where people buy and sell Yes shares. The No order book is where people buy and sell No shares. These are distinct pools of liquidity with different participants, different depth, and often different levels of activity.

The price displayed for each outcome is typically the lowest ask — the cheapest price at which you can buy that outcome right now. When you see Holloway at 94 cents, that means the cheapest available Yes share for Holloway costs $0.94. When you see Oliveira at 34 cents, the cheapest Oliveira Yes share costs $0.34. These are offers from different sellers on different order books, with no mechanical link forcing them to sum to $1.00.

In a perfectly liquid market with tight spreads, the two lowest asks would hover near $1.00 combined because arbitrageurs would immediately profit from any deviation. But in thin markets with wide spreads, the two ask prices can drift apart. Each order book is being set by a small number of participants who may not be monitoring the other side in real time.

Every order book has two sides: bids (buy orders) and asks (sell orders). The bid is the highest price someone is willing to pay. The ask is the lowest price someone is willing to sell. The gap between them is the spread. When a platform displays a price, it usually shows either the last traded price or the best ask — neither of which is a midpoint probability estimate.

Consider Holloway's order book. The highest bid might be $0.82 and the lowest ask might be $0.94. The spread is $0.12 — enormous by liquid market standards. If you want to buy Holloway Yes shares right now, you pay $0.94. If you want to sell them right now, you get $0.82. The displayed price of $0.94 reflects the cost of immediately buying in, not the market's consensus view of the probability.

The midpoint of the bid-ask spread is a better (though still imperfect) estimate of the market's true implied probability. For Holloway, that midpoint is ($0.82 + $0.94) / 2 = $0.88. For Oliveira, if the bid is $0.04 and the ask is $0.34, the midpoint is $0.19. Those midpoints sum to $1.07 — still above 100%, but much closer. The remaining overround reflects the cost of providing liquidity in a thin market.

The overround exists because of the spread on each individual order book. In a two-outcome market, if both outcomes have a $0.03 spread, the combined overround is roughly $0.03 to $0.06 (depending on where the midpoints sit). A $0.10 spread on each side can create a 10-20% overround. The wider the spreads, the higher the apparent combined probability.

Three factors drive spread width. First, liquidity: markets with low trading volume have fewer active participants posting orders, which means fewer competitive bids and asks, which means wider gaps. Second, uncertainty: when the outcome is genuinely hard to predict, market makers widen their spreads to compensate for the risk of being wrong. Third, market maker activity: professional market makers tighten spreads by posting competitive bids and asks on both sides. Markets without active market makers tend to have structurally wide spreads.

In the MMA example, the fight market likely has low daily volume and few active market makers. The handful of participants each set their own prices with wide margins for safety. The result is a 128% combined probability — not because anyone thinks both fighters have a strong chance, but because the cheapest available offers reflect the cost of trading in a thin market.

A highly liquid market like a US presidential election might show 52% and 49% on its two outcomes — a combined 101%. The 1% overround reflects extremely tight spreads of roughly half a cent per side. Trading costs are minimal, and the displayed prices closely approximate true implied probabilities. Dozens of professional market makers compete on every price level.

A moderately liquid market — say a popular NBA game — might show 62% and 45%, totaling 107%. The 7% overround means spreads of 3-4 cents per side. Still reasonable for most traders, but the displayed prices are not precise probability estimates. The true implied probabilities are closer to 58% and 42%.

An illiquid niche market — an obscure MMA undercard fight, a local election, or a long-tail sports prop — can show 80% and 80%, totaling 160%. This does not mean the market is broken. It means the few participants posting orders are demanding wide margins, and nobody has stepped in to tighten the spread. The true implied probabilities might be 55% and 45%, buried under layers of illiquidity premium.

In theory, overround creates a risk-free profit opportunity. If you can buy Yes shares on all outcomes for less than $1.00 combined, you lock in a guaranteed profit because exactly one outcome will pay $1.00 at resolution. When the combined ask prices exceed $1.00 (as in our 128% example), there is no direct arbitrage on the buy side — you would pay $1.28 for something worth $1.00.

But the opposite trade might work. If combined bid prices for all outcomes are less than $1.00, you can sell (short) every outcome and pocket the difference. In practice, this requires holding No shares on every outcome or having the capital to cover the short positions until resolution. The mechanics are more complex than simply buying low and selling high, which is one reason the overround persists in thin markets.

Market makers are the natural arbitrageurs here. They post competitive bids and asks on both sides of each outcome, earning the spread as profit while naturally pulling combined prices toward 100%. But market making in thin markets is risky — if you are the only liquidity provider and an informed trader comes along, you can lose more on a single adverse fill than you earned from weeks of spread capture. This risk-reward calculation is why many markets sit with wide spreads and high overround: it is simply not profitable enough for market makers to fix.

When you notice odds summing well above 100%, do not trust the displayed prices as probability estimates. Instead, normalize them. Take each displayed price, divide it by the total of all displayed prices, and that gives you the implied probability adjusted for overround. In the 128% example: Holloway's adjusted probability is 94 / 128 = 73.4%. Oliveira's is 34 / 128 = 26.6%. Those are closer to what the market actually implies.

Better yet, look at the midpoints of each order book's bid-ask spread rather than the ask prices alone. This strips out the liquidity premium embedded in the best available offer and gives you a cleaner signal. On 0xinsider.com/terminal, the signal scoring system already accounts for market liquidity when evaluating trades — a large trade in a thin market with wide spreads is treated differently than the same-sized trade in a deep, tight market.

Before entering any position in a high-overround market, check the spread on both sides. If the spread is 10 cents or more, you are paying a significant premium just to get in. Consider whether your edge is large enough to overcome that cost. A 5% edge in a market with a 15% overround means you are likely losing money on every trade. The overround is your cost of entry — and in illiquid markets, it can eat your edge entirely.

The overround is effectively a hidden tax on every trade in an illiquid market. When you buy Yes shares at the ask price in a market with a 20% overround, you are paying 10% more than the fair midpoint value. To break even, the true probability needs to be 10% higher than what you paid. This cost is invisible if you only look at the displayed price, which is why so many beginners get burned in thin markets — they think they are getting fair odds when they are actually paying a steep premium.

Experienced traders gravitate toward liquid markets for exactly this reason. A 1-2% overround means your entry price is very close to the fair value, and even a small informational edge translates into profit. Markets with 15%+ overround require a proportionally larger edge to be profitable, which is why most professional traders avoid them unless they have high-conviction views.

The bottom line: prediction market prices are not probabilities. They are the cheapest available offers on separate order books. When those offers are close together and competitive, the prices approximate probabilities well. When the market is thin and spreads are wide, the prices can diverge wildly from true probabilities — and the overround is the clearest signal that this is happening.