Prediction Market Odds Explained: Price & Edge 2026

Key takeaways

1. A contract price IS the implied probability. 65 cents means the market is pricing a 65% chance, and it pays exactly $1.00 if it resolves yes.
2. Fees follow a parabola, not a flat percentage. Polymarket and Kalshi both charge rate x contracts x price x (1 - price), so the most expensive place to trade is a 50/50 market and the cheapest is a heavy favourite or longshot.
3. At a 50 cent price you need to be right 51.00% of the time to break even on a Polymarket politics market, 51.75% on crypto, 51.75% on Kalshi, and 52.38% at a sportsbook pricing both sides at -110.
4. Polymarket US moved to a new schedule at 10am ET on 1 October 2026: a 0.0695 taker coefficient, a 0.0125 maker rebate paid to you, and taker rebates of 10% to 50% above $250,000 of monthly volume.
5. Edge is not the gap between your number and the price. It is that gap minus the fee, and on a crypto market at 50 cents a 1.5 point edge is already a losing trade.

On this page

1. A price is a probability, and that is the whole trick
2. Convert any price to implied, decimal and American odds
3. What a winning trade actually pays you
4. The fee formula that nobody explains
5. Polymarket US changed its fee schedule on 1 October 2026
6. Your real breakeven: the probability you have to beat
7. Edge and expected value: the maths that decides a trade
8. Why both sides can add up to more than 100%
9. Prediction market odds vs a sportsbook, quantified
10. Frequently asked questions

This is a plain list rather than a set of jump links, because guide pages on this site render section text as plain paragraphs. Scroll to whichever heading you need.

A price is a probability, and that is the whole trick

On Polymarket and Kalshi every market is a yes/no question, and every contract settles at either $1.00 or $0.00. There is no line to decode and no plus or minus number to reverse engineer. If yes is trading at 65 cents, the market thinks there is a 65% chance of yes, and the matching no contract will sit near 35 cents because the two have to add up to roughly a dollar.

That single fact is why prediction markets read so differently from a bookmaker's board. A sportsbook quotes you a payout and hides the probability inside it. A prediction market quotes you the probability and lets you work out the payout. Once you internalise that, the only questions left are what the trade costs and whether your own estimate is better than the crowd's.

Prices move because people trade, not because an oddsmaker adjusts a number. When a contract drifts from 65 cents to 72 cents, somebody with money decided 65% was too low. That is also why prediction-market prices tend to track outcomes reasonably well over large samples: being wrong is expensive.

Convert any price to implied, decimal and American odds

1. 5 cents: 5% implied, 20.00 decimal, +1900 American
2. 10 cents: 10% implied, 10.00 decimal, +900 American
3. 25 cents: 25% implied, 4.00 decimal, +300 American
4. 40 cents: 40% implied, 2.50 decimal, +150 American
5. 50 cents: 50% implied, 2.00 decimal, +100 American
6. 60 cents: 60% implied, 1.67 decimal, -150 American
7. 65 cents: 65% implied, 1.54 decimal, -186 American
8. 75 cents: 75% implied, 1.33 decimal, -300 American
9. 90 cents: 90% implied, 1.11 decimal, -900 American
10. 95 cents: 95% implied, 1.05 decimal, -1900 American

The three formulas are short enough to keep in your head. Implied probability is just the price: 65 cents is 65%. Decimal odds are 1 divided by the price, so 0.65 gives 1.54. American odds split at even money: below 50 cents it is 100 x (1 - price) / price as a positive number, and above 50 cents it is -100 x price / (1 - price).

Worth knowing why this matters beyond curiosity. If you also bet at a sportsbook, converting lets you compare the two venues on the same scale, and the comparison is usually unflattering for the sportsbook. A -186 price on a prediction market and a -186 line at a book look identical until you add the costs, which is the next section's job.

One warning on the conversion. The price you convert should be the price you can actually fill at, not the midpoint shown on a chart. On a thin market the gap between the best bid and the best offer can be several cents wide, and a 3 cent spread on a 50 cent contract is worth far more than any fee discussed below.

What a winning trade actually pays you

Each contract resolves at $1.00 if you are right and $0.00 if you are wrong, so your profit per contract is a dollar minus what you paid. Buy 100 contracts of yes at 60 cents and you have spent $60.00. If it resolves yes you receive $100.00, a profit of $40.00, or about 67% on the money you put up. If it resolves no you lose the whole $60.00.

The asymmetry is the part people misread. A 90 cent favourite only pays 11 cents on the dollar, so it has to win nine times out of ten just to stand still. A 10 cent longshot pays nine times your stake but loses nine times out of ten. Neither is better or worse. They are the same bet priced at different probabilities, and the only thing that separates a good trade from a bad one is whether the price is wrong.

You do not have to wait for resolution either. Because contracts trade continuously, you can sell at any time at the going price. If you bought at 60 cents and the market moves to 72 cents, selling locks in 12 cents per contract without ever learning the answer, which is how most active traders actually make money.

The fee formula that nobody explains

1. Politics, finance, tech and mentions, rate 0.04: $1.00 per 100 shares at 50 cents, $0.75 at 25 or 75 cents, $0.36 at 10 or 90 cents, $0.19 at 5 or 95 cents
2. Sports, economics, culture, weather and other, rate 0.05: $1.25 at 50 cents, $0.94 at 25 or 75 cents, $0.45 at 10 or 90 cents, $0.24 at 5 or 95 cents
3. Crypto, rate 0.07: $1.75 at 50 cents, $1.31 at 25 or 75 cents, $0.63 at 10 or 90 cents, $0.33 at 5 or 95 cents
4. Geopolitics, rate 0: no taker fee at any price
5. Kalshi takers, coefficient 0.07: $1.75 at 50 cents, $1.31 at 25 or 75 cents, rounded up to the nearest cent
6. Kalshi makers, coefficient 0.0175: $0.44 at 50 cents, about a quarter of the taker charge

Here is the detail that most explainers skip entirely. Polymarket's published fee is not a flat percentage of your stake. It is calculated as shares x rate x price x (1 - price), and Kalshi uses the same shape with a 0.07 coefficient. That price x (1 - price) term is a parabola with its peak at 50 cents, so the fee is largest exactly where the market is most uncertain and falls away towards both ends of the range.

Run the numbers and the effect is big. On a crypto market, 100 shares at 50 cents costs $1.75 in fees, while the same 100 shares at 95 cents costs 33 cents. As a share of your outlay it is starker still: 3.5% of a $50 position against 0.35% of a $95 one. Trading coin flips is roughly ten times more expensive, per dollar risked, than trading near certainties.

Two practical consequences. First, if you are hunting small edges, hunt them away from 50 cents, because a 1 point edge disappears into the fee at the peak of the curve. Second, makers are treated very differently from takers. On Polymarket makers are never charged and instead earn a share of the taker fees collected in the market, between 15% and 25% depending on category, so posting a resting order rather than crossing the spread changes your cost from negative to positive.

Polymarket US changed its fee schedule on 1 October 2026

1. Taker fee coefficient 0.0695, which peaks at $1.74 per 100 contracts at a 50 cent price
2. Maker rebate coefficient 0.0125, paid to you, worth $0.31 per 100 contracts at 50 cents
3. Monthly taker volume of $250,000 to $999,999: 10% of your taker fees rebated
4. Monthly taker volume of $1,000,000 to $9,999,999: 25% rebated
5. Monthly taker volume of $10,000,000 and above: 50% rebated
6. Rebates pay out weekly and are set by your previous calendar month's volume

If you trade on the regulated US venue rather than the global one, the schedule above took effect at 10am ET on Thursday 1 October 2026, and it is worth reading before you assume the old numbers still hold. The structure is the same parabola, with a taker coefficient of 0.0695 and a maker rebate of 0.0125, which means a maker is paid rather than charged.

Combination trades sit on their own curve. The published formula adds a second term for them, contracts x price x [0.0695 x (1 - price) + 0.06 x (1 - price) to the fourth power], which loads extra cost onto cheap legs. If you build parlay-style combos out of longshots, that fourth-power term is the thing to model, because it bites hardest exactly where combo traders like to shop.

One honest caveat about all of these numbers. Fee schedules on both venues have moved more than once during 2026, and the category rates on the global platform were only introduced during the year. Treat every figure here as what the operators published at the start of October 2026 and check the live schedule before you size a serious position.

Your real breakeven: the probability you have to beat

1. Geopolitics on Polymarket, no fee: 50.00%
2. Politics, finance, tech and mentions on Polymarket, rate 0.04: 51.00%
3. Sports, economics, culture and weather on Polymarket, rate 0.05: 51.25%
4. Crypto on Polymarket, rate 0.07: 51.75%
5. Polymarket US, taker coefficient 0.0695: 51.74%
6. Kalshi as a taker, coefficient 0.07: 51.75%
7. A sportsbook pricing both sides at -110: 52.38%

The list above answers the only question that matters before you place a trade: how often do I have to be right just to stand still? Those figures are all for a 50 cent contract, where fees peak, and they come from a single rule. Your breakeven true probability is the price plus the fee per share, or price + rate x price x (1 - price).

The same rule works at any price. At 65 cents on a politics market the fee per share is 0.04 x 0.65 x 0.35, which is 0.91 cents, so you need the real probability to be at least 65.91% rather than 65%. At 90 cents the fee drops to 0.36 cents and your breakeven is 90.36%. The further from even money, the smaller the hurdle the fee puts in front of you.

Compare the bottom two rows and you can see why prediction markets took share from sportsbooks. A book that prices both sides of a coin flip at -110 is implying 52.38% on each, which sums to 104.76% and leaves it a hold of about 4.55%. The most expensive prediction-market category asks you to beat 51.75%. That difference, a little over half a point on every trade, is the entire argument.

Edge and expected value: the maths that decides a trade

Edge is where every guide I read on this topic stops short, and it is the only calculation that tells you whether to click buy. Your expected value per contract is your own probability estimate minus the price you pay, minus the fee. If that number is not comfortably positive, the trade is noise.

A worked example. Suppose a politics market has yes at 60 cents and you have done the work and believe the real chance is 68%. For 100 shares you pay $60.00 plus a taker fee of 0.04 x 100 x 0.60 x 0.40, which is 96 cents, for a total outlay of $60.96. Your expected payout is 0.68 x $100, or $68.00, so your expected profit is $7.04. That is a return of about 11.6% on the money at risk, from an 8 point edge that the fee trimmed to 7.04 points.

Now the example that should change your behaviour. Take a crypto market with yes at 50 cents where you think the true probability is 52%. Your 100 shares cost $50.00 and the fee is 0.07 x 100 x 0.25, or $1.75, so you are in for $51.75 against an expected payout of $52.00. Your whole edge is worth 25 cents. Shade your estimate down to 51.5% and the same trade loses money. A 2 point read on a crypto coin flip is not an opportunity, and a 1.5 point read is a slow leak.

The lesson generalises. Small edges only survive where fees are small, which means away from 50 cents, in the cheaper categories, or as a maker rather than a taker. If your process produces 1 to 2 point reads, the fee curve decides whether you have a strategy or a hobby.

Why both sides can add up to more than 100%

In a clean two-sided market, yes at 65 and no at 35 sum to exactly 100 and there is nothing to exploit. In practice the two best offers often sum to a little more than a dollar, and in multi-outcome markets the gap can be wide. Each candidate or team trades as its own yes contract, and when demand is one-sided or liquidity is thin nothing forces the set to total 100%.

Say a four-way race shows 40, 30, 20 and 15 cents. Those sum to 105 cents, so buying one of each would cost $1.05 to guarantee a $1.00 payout, a 5 cent overround you are paying for the convenience of trading a thin book. Read that sum before you treat any single leg's price as the market's true probability, because an inflated total means every leg is slightly rich.

The reverse happens too, and it is the closest thing to a free lunch these venues offer. When the outcomes sum to less than 100 cents, buying the full set locks in a profit at resolution. It is rare, it is usually small, and fees plus the spread normally eat it, which is precisely why it is rare.

Prediction market odds vs a sportsbook, quantified

Every comparison of these two things that I have read asserts that prediction markets are cheaper and then declines to say by how much. Here is the arithmetic. A standard -110 line on both sides of a proposition implies 52.38% for each outcome. Those add to 104.76%, and the book's hold is the 4.76 points of overround divided by the total, about 4.55% of everything staked.

A prediction market at the same 50 cent price charges you the fee and nothing else. On a Polymarket politics contract that is 1 cent per share, so 2% of a 50 cent position, and your breakeven moves from 50% to 51.00% instead of 52.38%. On a geopolitics market the fee is zero and your breakeven is exactly 50%. That is the whole difference, and it compounds across every trade you make.

Two caveats so this stays honest. The sportsbook figure assumes a -110 two-way market, and books price plenty of things far worse than that, particularly parlays and player props. And a prediction market's quoted price is only available in the size that is actually resting on the book, so on an illiquid market the spread you cross can cost you more than any sportsbook's margin. Cheap fees and a thin order book are not the same thing as a cheap trade.

Sources and when we last checked

1. Polymarket US fee schedule, docs.polymarket.us/fees, effective 10am ET on 1 October 2026: the 0.0695 taker coefficient, the 0.0125 maker rebate and the volume rebate tiers
2. Polymarket global trading fees, help.polymarket.com article 13364478: the shares x rate x price x (1 - price) formula and the per-category rates
3. Kalshi fee schedule, help.kalshi.com: the 0.07 taker and 0.0175 maker coefficients, rounded up to the nearest cent
4. CFTC register of Designated Contract Markets, cftc.gov, for checking the regulated status of each venue yourself

Every figure in this guide was checked against those four documents on 1 October 2026. Fee schedules on both venues moved more than once during 2026, and the per-category rates on the global Polymarket platform were only introduced during the year, so treat these numbers as a snapshot. If you are sizing a real position rather than reading for background, open the live schedule and confirm the rate for your category first.

Frequently asked questions

Related guides