The expected value formula
A Kalshi contract on the winning side is worth $1 at settlement. A contract on the losing side is worth nothing. Let p be your probability that the contract wins. Let c be the price you pay and f the fee per contract, both in dollars.
The long form shows where that comes from. If the contract wins, you collect $1 and your profit is 1 - c - f. If it loses, you are out c + f. Weight each result by its chance.
Multiply it out and the p × (c + f) terms cancel. That leaves p - c - f. Gross EV is the same number before the fee: p - c.
On Kalshi the fee depends on the price. Under the fee schedule updated July 7, 2026, the standard taker fee is 0.07 × C × P × (1 - P), rounded up. C is the number of contracts and P is the price. Orders that rest on the book and fill later pay nothing on most series. Some series also charge makers, usually at 0.0175 in place of 0.07. A few series charge half the standard fee, and a few charge nothing. The Kalshi fees guide covers which series differ.
A worked YES example
Suppose you estimate a 62% chance that a market settles YES. The best YES ask is 55c, and you buy as a taker on a standard series. The fee is 0.07 × 0.55 × 0.45, which comes to 1.7325c per contract.
For an order of 100 contracts the numbers work out like this.
| 100 YES contracts at 55c | Amount |
|---|---|
| Cost | $55.00 |
| Taker fee | $1.73 |
| Profit if YES | $43.27 |
| Loss if NO | $56.73 |
| Expected value at 62% | $5.27 |
This one order either makes $43.27 or loses $56.73. The $5.27 is the average across many trades at the same true chance and price. Gross EV was 7c, so the fee took about a quarter of it.
Your role changes the fee. Say you buy at 55c with a resting order, and a taker fills it later. A standard series charges you nothing as the maker, so EV is the full 7c. On a series with maker fees the charge is about 0.43c, and EV is 6.57c. A resting order may never fill, so the extra EV comes with that risk. Kalshi limit orders explains how resting orders queue and fill.
The NO side of the same market
YES and NO trade in one book. Kalshi's API returns only bids for each side, because a bid on one side is an offer on the other. A NO bid at 45c is the same order as a YES ask at 55c. So the best YES ask equals $1 minus the best NO bid, and the best NO ask equals $1 minus the best YES bid. How the Kalshi order book works goes through this in detail.
Say the best YES bid in our market is 53c. The NO ask is then 47c. Your 62% for YES is a 38% estimate for NO. The taker fee at 47c is 0.07 × 0.47 × 0.53, about 1.74c.
The two asks add up to 102c. The 2c above $1 is the bid-ask spread. The spread plus the fee on each side creates a band of estimates where neither side has positive EV for a taker.
| Your YES estimate | Positive EV as a taker |
|---|---|
| Above 56.73% | YES at 55c |
| 51.26% to 56.73% | Neither side |
| Below 51.26% | NO at 47c |
The band is about 5.5 points wide: the 2c spread plus the fee on each side. A wider spread makes it wider. So does a price near 50c, where the fee is highest.
Reading a Kalshi price as a probability
Divide a price in cents by 100 and you have the market's implied probability. A YES contract at 55c implies 55%. Kalshi's help center describes its prices as tied to traders' view of how likely the event is.
- The ask is what you would pay to buy now. It is the price to use for the EV of a trade you take now.
- The midpoint of the best bid and ask summarizes the market's view. In our example it is 54c.
- The last trade can be hours old and can sit outside the current bid and ask.
Events with several outcomes add a step. Kalshi flags some events as mutually exclusive. That means at most one market in the event can settle YES. When the listed outcomes also cover every possibility, exactly one settles YES and the fair YES prices add up to $1. Each ask sits above its fair value by part of the spread, so the asks usually add up to more than $1. The excess is called the overround. Take an event with four outcomes, one of which must win.
| Outcome | YES ask | Implied | Scaled to 100% |
|---|---|---|---|
| A | 48c | 48% | 45.3% |
| B | 30c | 30% | 28.3% |
| C | 18c | 18% | 17.0% |
| D | 10c | 10% | 9.4% |
| Total | $1.06 | 106% | 100% |
Dividing each price by the total is the simplest way to strip out the overround. It assumes the excess is spread in proportion to price. The research covered below suggests cheap contracts on Kalshi have tended to carry more than their share. So outcome D's true chance may sit below 9.4%. The Kalshi odds converter turns cents into percentages and sportsbook odds. Kalshi odds explained covers the conversion in full.
Break-even probability after the fee
Set EV to zero and solve for p. The break-even probability is the price plus the fee per contract.
The fee shifts the line down by the same amount at every probability, so break-even moves from 55% to 56.73%.
| Price | Taker fee per contract | Break-even as taker | Break-even as maker on a maker-fee series |
|---|---|---|---|
| 10c | 0.63c | 10.63% | 10.16% |
| 25c | 1.31c | 26.31% | 25.33% |
| 50c | 1.75c | 51.75% | 50.44% |
| 75c | 1.31c | 76.31% | 75.33% |
| 90c | 0.63c | 90.63% | 90.16% |
A resting order on a standard series pays no fee, so its break-even is the price itself. In cents the fee peaks at 50c. As a share of the price it is largest on cheap contracts. At 10c the fee is 6.3% of the price, and at 90c it is 0.7%.
These figures use the exact formula. Kalshi rounds each fill's cost plus fee up to the next $0.0001 for members who trade directly. For members who clear through a futures commission merchant (FCM), it rounds up to the next cent. If rounding makes you overpay on one fill, the extra carries across the other fills of the same order. Kalshi refunds it as it adds up. One contract at 50c pays $0.0175 direct and $0.02 through an FCM. So on very small orders, break-even sits slightly higher. The Kalshi fee calculator gives the fee and break-even for any price and size.
Positive EV can sit on either side
Nothing in the formula ties positive EV to the favorite. Suppose you estimate 30% for an outcome priced at 25c. Gross EV is 5c. After a 1.31c taker fee it is 3.69c per contract. That is positive EV on the side you expect to lose.
Edges on cheap contracts are fragile for two reasons. The first is error per dollar. One point of estimate error moves EV by 1c a contract at any price. At 25c that 1c is 3.8% of the 26.31c you commit. At 55c it is 1.8% of 56.73c.
The second is the evidence from Kalshi itself. A January 2026 working paper from University College Dublin studied Kalshi prices from 2021 through April 2025. It covered 46,282 contracts across 12,403 events. Each contract had at least $1,000 in volume. The paper found a favorite-longshot bias. On average, contracts priced at 10c or less lost buyers over 60% of their money after fees. Contracts priced above 50c earned small positive returns. The authors' model reproduces the pattern. It uses modest disagreement among traders and a slight tendency to overstate small probabilities. An estimate that finds value in a 25c contract is betting against that pattern.
Four ways a positive EV turns out wrong
- The probability is off. At a fixed price, every point of error moves EV by 1c per contract. The table below shows the 55c YES trade at a range of true probabilities.
- The book is thin. The quoted EV holds only for the contracts resting at the ask. If 40 contracts rest at 55c and the next level is 58c, contract 41 costs 58c plus a 1.71c fee. At 62% its EV falls to 2.29c.
- The fee is left out. A 2c gross edge at 50c becomes 0.25c after the 1.75c taker fee.
- The rules define a different event. Your 62% has to be the chance of YES as the market's rules define it. The rules name the source and the time that decide it. A forecast of the headline event can differ from that. How Kalshi settles a market walks through a rules block line by line.
| True probability | EV per contract for a taker at 55c |
|---|---|
| 55% | -1.73c |
| 56% | -0.73c |
| 57% | 0.27c |
| 58% | 1.27c |
| 60% | 3.27c |
| 62% | 5.27c |
| 65% | 8.27c |
If the true chance is 57% and you estimated 62%, the 5.27c edge you saw is really 0.27c. Many traders want a margin wider than their typical estimate error before they act.
EV in dollars and per day of capital
Cents per contract leave out size. Multiply by the contracts you can fill. A 20c edge on 5 resting contracts is worth $1. A 3c edge on 2,000 contracts is worth $60.
Time matters too. The 55c trade puts up 56.73c per contract to earn an expected 5.27c. That is a 9.3% expected return. If the market settles in 2 days, the cash is free again quickly. If it settles in 9 months, the same 9.3% is spread over about 270 days. Meanwhile the cash cannot back other trades. For eligible US accounts, Kalshi pays variable interest on cash and open positions. That offsets part of the cost. Kalshi says most markets settle within a few hours of the outcome being known.
You can also exit early by selling into the bids on the book. The exit price depends on who is bidding then. A taker sale pays the fee again. Selling a Kalshi contract before settlement covers how that exit is priced.
How often a positive EV still loses
Suppose you make 100 trades like the YES example, each in a different market. Each has a true 62% chance and a 55c price. You expect 62 wins. You finish behind if 56 or fewer win. Under a binomial model, the chance of that is about 13%. Over 500 such trades it falls below 1%, and over 1,000 trades it is about 0.03%.
Those odds assume your 62% is right and the trades are independent. Markets on the same event or the same data release tend to win or lose together. That raises the chance of a losing run.
From expected value to position size
Expected value tells you whether a price is in your favor. The Kelly criterion tells you what share of your bankroll to commit to a $1 contract. Here c is the price plus the fee.
For the YES example, c is 56.73c and full Kelly is 12.2% of bankroll. Many traders bet a fraction of that, because their probabilities carry error. The Kelly criterion guide has the derivation and the fractional versions. The Kelly and EV calculator runs both numbers for any price, side and fee type.
Questions traders ask
Is a positive expected value guaranteed to make money?
No. Positive EV means the average result over many similar trades is positive, provided your probability is right. Single trades lose often. In the 62% at 55c example, a run of 100 independent trades still finishes behind about 13% of the time.
What counts as a good EV on Kalshi?
There is no fixed threshold. Traders usually judge EV after fees, in dollars they can fill, and against how long the cash stays tied up. A 2c net edge on a deep market that settles tomorrow can be worth more than a 15c edge on 10 contracts that settle next year.
How do I turn a Kalshi price into a probability?
Divide the price in cents by 100. A 37c YES price implies 37%. For a trade placed now, the ask is the price that counts. The midpoint of the best bid and ask gives the market's central view. In events where exactly one market settles YES, dividing each price by the total removes the overround.
Does expected value include Kalshi's fee?
Net EV does. It is your probability minus the price minus the fee. The standard taker fee is 0.07 × contracts × price × (1 - price), rounded up. That is 1.75c per contract at 50c. Resting orders pay nothing on most series. On series that charge makers, they usually pay 0.0175 times the same product.
Can NO have positive EV when YES has negative EV?
Yes. NO has positive EV for a taker when your YES probability is below the YES bid minus the fee on NO. With a 53c YES bid and a 55c YES ask, NO at 47c is positive below 51.26% and YES is positive above 56.73%. Between them neither side is positive, and the two are never positive at once.
Sources
- Kalshi, Fee Schedule for July 2026, updated July 7, 2026 (accessed Oct 19, 2026)
- Kalshi API documentation, Get Series, fee_type and fee_multiplier fields (accessed Oct 19, 2026)
- Kalshi API documentation, Fee Rounding (accessed Oct 19, 2026)
- Kalshi API documentation, Orderbook Responses (accessed Oct 19, 2026)
- Kalshi API documentation, Get Event, mutually_exclusive field (accessed Oct 19, 2026)
- Kalshi Help Center, What are prediction markets (accessed Oct 19, 2026)
- Kalshi Help Center, How are prices determined (accessed Oct 19, 2026)
- Kalshi Help Center, Fees (accessed Oct 19, 2026)
- Kalshi Help Center, Market FAQs (accessed Oct 19, 2026)
- Kalshi Help Center, Quick Order Sale (accessed Oct 19, 2026)
- Kalshi Help Center, APY on Kalshi (accessed Oct 19, 2026)
- Constantin Bürgi, Wanying Deng and Karl Whelan, Makers and Takers: The Economics of the Kalshi Prediction Market, University College Dublin, January 2026 (accessed Oct 19, 2026)