Prediction Market Odds Converter

By Alex Thornton · author of the MIT-licensed teachersbettextbook library · Last reviewed 24 July 2026

Every converter turns 65¢ into −186 and stops. That number is a fine translation and a bad cost estimate: as a taker you are paying −199. Here is the conversion, the fee, and the difference — every step shown.

A prediction market odds converter turns the price of a binary event contract into the four formats a bettor reads — implied probability, decimal, fractional and American odds — and then adds the exchange fee to produce the fee-adjusted break-even probability, which is what the position actually costs.

P \;\longrightarrow\; \underbrace{q = P}_{\text{implied}},\quad \underbrace{d = \tfrac{1}{P}}_{\text{decimal}},\quad \underbrace{b = \tfrac{1-P}{P}}_{\text{fractional / net}},\quad \underbrace{P^{*} = P + 0.07\,P(1-P)}_{\text{fee-adjusted}}

where:

$P$ — the contract price in dollars, strictly between 0 and 1. Kalshi quotes it in whole cents, so a 65¢ contract is $P = 0.65$ . A contract pays $1.00 on YES and $0.00 on NO.
$q$ — the implied probability, dimensionless, between 0 and 1. On a two-sided exchange quote it equals $P$ exactly. No de-vig step.
$d$ — decimal odds: total return per $1 risked, stake included.
$b$ — net decimal odds, in dollars of profit per dollar risked. This is also the fractional quote once reduced: $b = 7/13$ prints as 7/13.
$A$ — American odds: $A = +100\,b$ when $b \geq 1$ (that is, $P \leq 0.50$ ), otherwise $A = -100/b$ .
$P^{*}$ — the fee-adjusted break-even probability, in the same units as $P$ . It is the all-in cost per $1 of payout on the standard Kalshi taker schedule. Every format above can be recomputed from $P^{*}$ instead of $P$ — that is the effective price.
$C$ — number of contracts, a positive integer. The fee is charged on the total and rounded up to the next cent.

Worked example — the numbers in the form below

100 contracts at 65¢, taken from the ask (taker):

q = 65.00\%, \qquad d = \tfrac{1}{0.65} = 1.5385, \qquad b = \tfrac{0.35}{0.65} = \tfrac{7}{13}, \qquad A = -\tfrac{100 \times 0.65}{0.35} = -185.71

\text{fee} = \lceil\, 0.07 \times 100 \times 0.65 \times 0.35 \,\rceil_{\text{cent}} = \lceil \$1.5925 \rceil = \$1.60

\text{all-in} = \$65.00 + \$1.60 = \$66.60 \;\Longrightarrow\; P^{*} = 66.60\%

d^{*} = \tfrac{1}{0.665925} = 1.5017, \qquad A^{*} = -\tfrac{100 \times 0.665925}{0.334075} = -199.33

The screen says −186. The till says −199. Same contract, 13.6 points of American odds apart — and the closed-form break-even (66.5925%) sits a hair under the rounded one (66.60%) because the fee rounds up to a whole cent.

Compute the fee side on your own price

The four format conversions above are fixed arithmetic on $P$ — the reference table below covers them from 5¢ to 95¢. The part that moves with your order size is the fee, so that is what this panel computes live: the fee on the fill, the fee-adjusted break-even, and what a round trip costs against holding to settlement.

Comparing a contract against a sportsbook price on the same game? The prediction market arbitrage calculator nets the fee out of both legs →

Why the price is already the probability

A binary event contract has exactly two payoffs: $1.00 if the stated outcome happens, $0.00 if it does not. A risk-neutral buyer pays the expected payoff, which is

\mathbb{E}[\text{payoff}] = p \times \$1.00 + (1-p) \times \$0.00 = p

so price and probability are the same number wearing different units. That identity is why an exchange quote skips a step a sportsbook line cannot. YES at 65¢ and NO at 35¢ are the same claim seen from each end, and they sum to $1.00 by construction — there is no overround to strip. A sportsbook posting −186 / +150 has two implied probabilities summing past 100%, and until you de-vig them neither one means anything.

Which is also the reason the American conversion is the least useful of the four here. It exists so you can line a contract up against a book on the same screen. It is a translation, not the native language of the market.

The three conversions, in full

Decimal. Total return per dollar risked, stake included. A $1.00 payoff for a price of $P$ returns $1/P$ per dollar:

d = \dfrac{1}{P} \qquad P = 0.65 \Rightarrow d = 1.5385

Fractional. Profit per unit staked, as a reduced fraction. For a contract priced at $c$ whole cents this is $(100-c)/c$ in lowest terms:

b = \dfrac{1-P}{P} = \dfrac{100-c}{c} \qquad c = 65 \Rightarrow \dfrac{35}{65} = \dfrac{7}{13}

American. The sign flips at 50¢, because 50¢ is even money:

A = \begin{cases} +\,\dfrac{100\,(1-P)}{P} & P \leq 0.50 \\[8pt] -\,\dfrac{100\,P}{1-P} & P > 0.50 \end{cases}

At $P = 0.50$ both branches give 100, which is why an even-money contract is quoted +100 rather than ±100. Above 50¢ you are laying odds; below it you are taking them.

Reference table: contract price in every format

Fee columns use the standard Kalshi taker rate, 0.07, and the closed-form per-contract fee $0.07\,P(1-P)$ — the cent rounding is applied to the order total, not per contract, so it does not appear here. Rates are the exchange’s published standard schedule and can vary by market.

Price	Implied	Decimal	Fractional	American	Taker fee	Fee-adj. break-even	Fee-adj. American
5¢	5.00%	20.0000	19/1	+1900	0.333¢	5.3325%	+1775
10¢	10.00%	10.0000	9/1	+900	0.630¢	10.6300%	+841
20¢	20.00%	5.0000	4/1	+400	1.120¢	21.1200%	+373
25¢	25.00%	4.0000	3/1	+300	1.313¢	26.3125%	+280
33¢	33.00%	3.0303	67/33	+203	1.548¢	34.5477%	+189
50¢	50.00%	2.0000	1/1	+100	1.750¢	51.7500%	−107
65¢	65.00%	1.5385	7/13	−186	1.593¢	66.5925%	−199
75¢	75.00%	1.3333	1/3	−300	1.313¢	76.3125%	−322
90¢	90.00%	1.1111	1/9	−900	0.630¢	90.6300%	−967
95¢	95.00%	1.0526	1/19	−1900	0.333¢	95.3325%	−2042

Read the last two columns against the fourth. At 50¢ the fee moves an even-money contract to −107 — it is no longer even money. At 5¢ it moves +1900 to +1775, which looks small until you notice it is 6.65% of everything you put in.

Cheap in cents, expensive in percent

The fee formula peaks at a coin flip and shrinks toward the tails, so a 5¢ contract costs a third of a cent to trade. The fraction of your money it consumes runs the other way. Divide the fee by the outlay and the price cancels out of the numerator:

\dfrac{0.07\,P(1-P)}{P} = 0.07\,(1-P)

That is a clean, monotone result: the fee eats 6.65% of every dollar at a 5¢ price and 0.35% at 95¢. The cheapest contracts carry the heaviest proportional cost, which is the exact inverse of how they look on a fee schedule. It is also the opposite of sportsbook vig, which is heaviest on longshots in cents and in percent.

The same 0.333¢, two very different trades

P = 0.05:\quad \text{fee} = 0.3325\text{ cents} \;\Rightarrow\; \dfrac{0.3325}{5.00} = 6.65\%\ \text{of outlay}

P = 0.95:\quad \text{fee} = 0.3325\text{ cents} \;\Rightarrow\; \dfrac{0.3325}{95.00} = 0.35\%\ \text{of outlay}

Identical in cents, nineteen times apart in percent. A fee schedule quoted per contract hides that entirely.

Fair probability and break-even are two different numbers

This is the distinction the whole page exists for, so it gets stated flatly. $P^{*}$ is arithmetic on a posted price: it is what the contract costs, all in, per $1 of payout. Your fair probability $p$ is an estimate of how often the event happens. They are computed from different inputs and they mean different things. Expected value is the gap:

\mathrm{EV}_{\text{per contract}} = p \times \$1.00 - \bigl(P + 0.07\,P(1-P)\bigr) = p - P^{*}

On the worked example, all-in cost is $66.60 for 100 contracts that pay $100:

p = 70\%:\quad \mathrm{EV} = 0.70 \times \$100 - \$66.60 = +\$3.40

p = 66.60\%:\quad \mathrm{EV} = \$66.60 - \$66.60 = \$0

That second line is the honest zero. When your own estimate lands exactly on the fee-adjusted break-even, the answer is $0, not “marginal value” and not a rounding opportunity. Tools that print a single number labelled “fair odds” have quietly merged $p$ and $P^{*}$ , and once merged the edge is unrecoverable.

Maker, taker, and the flat 0.25% case

Two schedules, then an exception that matters only at the bottom of the price range.

\text{taker} = \lceil\, 0.07\,C\,P(1-P) \,\rceil_{\text{cent}} \qquad \text{maker} = \lceil\, 0.0175\,C\,P(1-P) \,\rceil_{\text{cent}}

A resting order that gets filled pays roughly a quarter of what crossing the spread costs — patience is discounted, in exchange for the risk of never being filled. Settlement is free, so the exit costs nothing if the contract is held to resolution, and a round trip pays the fee twice.

The exception: some major events replace the maker formula with a flat 0.25% of notional. Notional is $1.00 per contract, so that is a flat 0.25¢ per contract at any price — and a flat charge on a variable price is a wildly different animal at 3¢ than at 50¢. Setting the two maker schedules equal,

0.0175\,P(1-P) = 0.0025 \;\Longrightarrow\; P(1-P) = \tfrac{1}{7} \;\Longrightarrow\; P = \tfrac{1 \pm \sqrt{3/7}}{2}

gives crossover prices of 17.27¢ and 82.73¢. Between them the flat schedule is cheaper; outside them it is more expensive, and it gets ugly fast at the bottom.

Price	Flat 0.25% fee	Share of the price	Break-even, flat	Break-even, standard maker	Break-even, standard taker
1¢	0.250¢	25.00%	1.2500%	1.0173%	1.0693%
2¢	0.250¢	12.50%	2.2500%	2.0343%	2.1372%
3¢	0.250¢	8.33%	3.2500%	3.0509%	3.2037%
5¢	0.250¢	5.00%	5.2500%	5.0831%	5.3325%
10¢	0.250¢	2.50%	10.2500%	10.1575%	10.6300%
50¢	0.250¢	0.50%	50.2500%	50.4375%	51.7500%

The row worth staring at is 3¢. Under the flat schedule a maker — the discounted role, the one that supplies liquidity and waits — breaks even at 3.25%, while a standard-schedule taker breaks even at 3.204%. The discount has inverted. Solving $0.07\,P(1-P) = 0.0025$ puts that inversion below 3.71¢. Which schedule applies is stated on the contract, not inferable from the price, so it has to be read there.

The cent rounding is a real cost on small orders

The fee is computed on the order total and then rounded up to the next whole cent. On a hundred contracts that is noise. On one contract it is not:

C = 1,\ P = 0.50:\quad 0.07 \times 1 \times 0.25 = \$0.0175 \;\longrightarrow\; \lceil\, \cdot \,\rceil_{\text{cent}} = \$0.02

That is a 14.3% surcharge on the stated fee, paid purely for being small. The effect vanishes as size grows — at 100 contracts the same trade costs exactly $1.75 with nothing rounded away — which means the fee schedule is mildly regressive at the bottom of the size range. The live panel above uses the rounded figure, because that is the number that leaves your account.

Selling, and the NO side

Buying NO at $1-P$ and selling YES at $P$ are the same position, and they carry the same fee, because $P(1-P)$ is symmetric:

0.07\,P\,(1-P) \;=\; 0.07\,(1-P)\,\bigl(1-(1-P)\bigr)

So the converter runs identically from either end. If YES is 65¢, NO is 35¢, and NO converts to +186 with a fee-adjusted +173. What is not symmetric is which side you actually pay: on a book quoting 64¢ bid / 65¢ ask, a taker buys YES at 65¢ and buys NO at 100 − 64 = 36¢. The spread is charged to whoever is in a hurry, on both sides. Bid, ask, depth and the microprice covers how to read that properly — and why the midpoint is the lazy answer.

What the converted number is not

It is not a prediction, and the exchange is not an oracle. A 2026 study of more than 300,000 settled Kalshi contracts found prices informative and sharpening as expiry approached — and low-priced contracts still winning less often than the fee-adjusted break-even required.^[2] That is the favorite–longshot bias, first measured at racetracks in 1949^[3] and found in essentially every betting pool since. Our own fair-line machinery fits a shading exponent against settled markets for exactly that reason; the tails get pulled in rather than taken at face value.

Two more things the conversion cannot see. Staleness: a tidy quote on a market nobody has touched in three hours converts just as cleanly as a live one, and means far less — which is why every number on this site carries both a compute time and a venue-update time. Depth: the ask is a price for the contracts resting at that ask. Convert 65¢ all you like; if only 40 contracts sit there, a 500-contract order fills at a worse average.

Common mistakes

Quoting the screen price as the cost. A 65¢ taker fill costs 66.59¢ per $1 of payout. The gap is 13.6 points of American odds — larger than most edges anyone is chasing.
De-vigging an exchange quote. There is nothing to remove. YES + NO already sums to $1.00; running a de-vig over it just introduces error.
Reading a cheap fee as a cheap trade. 0.333¢ is 6.65% of a 5¢ contract. Fee in cents and fee in percent point opposite ways.
Assuming the maker schedule is always the discount. Under the flat 0.25% variant, below 3.71¢ a maker pays more than a standard taker.
Comparing a contract to a book line without adjusting either. The contract needs its fee added; the book price needs its vig removed. Skip either and the comparison is between two different objects.
Forgetting the second fee. Entry and a pre-settlement exit are both charged. Round trip at 50¢ is 3.5¢ per contract, not 1.75¢.
Treating the price as the true probability. It is the market’s estimate, and it leans in the tails. Expected value needs your own $p$ as a separate input.

Check this number yourself

The fee math here is four public functions in the MIT-licensed teachersbettextbook package. Same formulas, same rounding rule, no account required.

pip install teachersbettextbook

>>> from teachersbettextbook import kalshi_fee, kalshi_breakeven, kalshi_fee_of_outlay
>>> kalshi_fee(100, 0.65)          # 100 contracts at 65 cents, taker
1.6
>>> kalshi_breakeven(0.65)         # closed form, before the cent rounding
0.665925
>>> kalshi_fee_of_outlay(0.05)     # the fee as a share of money at risk
0.0665

The odds conversions are one line each: 1/P for decimal, (1-P)/P for net fractional, and the sign rule above for American. More on what is open and what is not: open source.

Sources

Kalshi. “Fee Schedule” (July 2026 revision). — the taker and maker formulas, the round-up-to-the-cent rule, the flat 0.25% maker variant, and the zero settlement fee.
Burgi, C., Deng, W. & Whelan, K. (2026). “Makers and Takers: The Economics of the Kalshi Prediction Market.” GWU Working Paper 2026-001. PDF. — 300k+ settled contracts: informative prices, favorite–longshot bias after fees.
Griffith, R. M. (1949). “Odds Adjustments by American Horse-Race Bettors.” American Journal of Psychology 62(2), 290–294.
Wolfers, J. & Zitzewitz, E. (2004). “Prediction Markets.” Journal of Economic Perspectives 18(2), 107–126. — the standard survey of why a contract price reads as a probability.

Frequently asked questions

How do you convert a prediction market contract price to odds?

A contract price in dollars is already the implied probability, so the rest follows from it. Decimal odds are 1 divided by P, net fractional odds are (1−P)/P reduced, and American odds are +100(1−P)/P when P is at or below 50¢ and −100P/(1−P) above it. A 65¢ contract is 65% implied, 1.5385 decimal, 7/13 fractional and −186 American.

What is the fee-adjusted price of a prediction market contract?

It is the contract price plus the per-contract trading fee, expressed back in odds. On the standard Kalshi taker schedule the fee is 0.07 × P × (1−P) per contract, so a 65¢ contract effectively costs 66.59¢ for a claim that pays $1.00. That is 1.5017 decimal, or −199 American, against the −186 the screen price alone converts to.

Is the contract price the same as the fair probability?

No, and merging them is the most common error on these markets. The price converts to a break-even probability, which is arithmetic on a posted number. The fair probability is a separate estimate of how often the event actually happens. Expected value is the gap between the two, so a page that prints only one of them cannot show you an edge.

Why is a converted prediction market price different from a sportsbook line?

A two-sided exchange quote sums to $1.00 by construction, so YES at 65¢ and NO at 35¢ imply exactly 100% between them and need no de-vigging. A sportsbook posts two prices that sum above 100%, and the excess is its margin. The exchange charges its cost as a stated fee instead of burying it in the price.

How does the flat 0.25% maker fee change cheap contracts?

Some major events replace the standard maker formula with a flat 0.25% of notional, which is 0.25¢ per contract at any price. On a 3¢ contract that is 8.33% of the price and lifts break-even to 3.25%, above even the 3.20% a standard taker pays. The two schedules cross at about 17.3¢ and 82.7¢.

Does a prediction market price tell you the true probability of an event?

It is an estimate, not an oracle. A 2026 study of more than 300,000 settled Kalshi contracts found prices informative and sharpening toward close, while low-priced contracts still won less often than the fee-adjusted break-even required. That favorite–longshot lean is why this site shades tail prices rather than treating an exchange quote as unbiased truth.