Skip to content
Sports Index
PRO
Research 5 min read

Kalshi Fees and Expected Value: An AI Research Worksheet

For a simple Yes contract held to a $1-or-$0 settlement, expected profit is expected payout minus purchase cost and fees. AI can check the arithmetic, but the probability estimate and fee inputs still need evidence.

A model saying “this team has a 54% chance” is not enough to assess a 50¢ contract. You need to know what you can buy, how much it costs, what the contract pays, and how the model arrived at 54%. Each of those can change the conclusion.

This worksheet is for an ordinary binary event contract purchased and held through settlement. It does not model leverage, perpetual futures, partial early payouts, unusual settlement values or an active sequence of entries and exits. If your contract differs, start with its actual cash flows. The rules checklist helps identify those cases.

Which Kalshi fee schedule should you use?

As checked October 27, 2026, Kalshi’s fee PDF is effective July 7, 2026. Its general taker model is M × 0.07 × C × P × (1 − P); its maker model uses 0.0175 instead of 0.07. C is quantity, P is price in dollars, and M is the applicable multiplier. The schedule includes exceptions. Do not assume M or maker treatment from the market’s headline. Kalshi fee schedule, effective July 7, 2026.

Rounding also needs care. Kalshi’s API documentation describes different target balance precision for direct and non-direct members, plus per-fill trade fees, rounding adjustments and rebates. A blanket instruction to round every trade fee up to the nearest cent is not a reliable implementation of the current rules. Kalshi API: Fee Rounding.

For a practical research worksheet, record the fee shown for the specific proposed order and reconcile it with the applicable schedule. Keep its date and the market identifier. If the inputs disagree, stop and resolve the discrepancy. Do not ask AI to silently choose whichever fee makes the trade look attractive.

Define the inputs before asking for a calculation

SymbolMeaningCommon mistake
CNumber of contracts actually assumed filledUsing requested quantity when only part can fill
pAverage paid price in dollarsUsing midpoint, last trade or cents without converting
qEstimated probability of the Yes outcomeTreating model confidence as a calibrated forecast
FTotal entry fees in dollarsConfusing a total fee with a per-contract fee
YSettlement payout per contractAssuming every settlement is exactly $0 or $1

The most general version is expected profit = C × expected payout per contract − total purchase cost − applicable costs. With only a $1 Yes payout and a $0 No payout, expected payout per contract is q. If all contracts fill at p, expected profit becomes C × (q − p) − F. Break-even probability is p + F/C under these assumptions.

A complete hypothetical example

Assume 100 Yes contracts fill at $0.50, the general taker model applies with M = 1, the entry fee is $1.75, there are no other charges, and the contracts are held to an ordinary $1-or-$0 settlement. This is an invented calculation, not a market quote or a recommendation.

CalculationResult
Purchase cost100 × $0.50 = $50.00
Assumed entry fee1 × 0.07 × 100 × 0.50 × 0.50 = $1.75
Total entry outlay$51.75
Profit if Yes pays $1$100 − $51.75 = $48.25
Loss if Yes pays $0$51.75
Break-even Yes probability0.50 + 1.75/100 = 51.75%
Assumed probability qExpected payoutExpected profit
50%$50.00−$1.75
52%$52.00+$0.25
54%$54.00+$2.25

The 52% case leaves only 25¢ of modeled expected profit across 100 contracts. A small probability error or a worse fill can erase that margin. None of these rows predicts the next result. A single event can lose even when the assumed expected value is positive.

The equation also explains why “I win more often than I lose” is incomplete. If contracts are expensive enough, a high win rate can still lose money. Conversely, an inexpensive contract can have positive modeled expected value while losing more often than it wins. The price and payout determine what win rate would be sufficient.

How should you handle spread and limited depth?

Imagine a separate hypothetical book with a 46¢ best bid and a 50¢ best ask. The 48¢ midpoint is a summary, not a guaranteed purchase price. If you use the 50¢ ask as the entry assumption, do not then subtract the entire 4¢ spread as an additional entry charge. That would count the entry disadvantage twice.

Now assume only 30 contracts are available at 50¢ and the next 70 at 53¢. Buying 100 across those levels costs $15 + $37.10 = $52.10 before fees, an average of 52.1¢. At a hypothetical 54% probability, only $1.90 remains before entry fees. That is smaller than the apparent $4 gap calculated from 100 contracts at the top 50¢ price.

Kalshi’s Quick Order documentation explains that orders can span price levels. A buy limit specifies a maximum purchase price, but it does not create liquidity or guarantee a complete fill. A limit that immediately crosses available offers is not automatically a fee-free maker order. Kalshi: Quick Orders; Kalshi: Limit Orders; Kalshi: Fees.

If you plan to sell before settlement, build a different worksheet: actual or assumed sale proceeds minus purchase cost, entry fees and exit fees. A rising midpoint does not establish the proceeds available for your quantity. Do not apply the held-to-settlement q − p formula to an early-exit strategy without modeling the sale.

Copy this prompt to audit a worksheet

Audit this hypothetical event-contract worksheet using deterministic arithmetic. Inputs: contract and payoff rules [text]; quantity [C]; price levels and quantities [table]; fee source and total fee [details]; probability estimate and range [q]; holding assumption [settlement or early sale]. Check units, fill quantity, weighted purchase price, total cost, break-even and sensitivity. Do not invent a fee, probability, or executable price. If an input is missing, show the unresolved field and stop the affected calculation. Avoid double-counting spread.

A successful answer shows the arithmetic line by line and names the assumptions. An unsuccessful one gives a colorful “edge score” while hiding fees or replacing your supplied price with a remembered price. Check the key calculations with a calculator or spreadsheet you control.

How do you account for uncertainty in q?

Use a sensitivity table before trying to optimize anything. If your estimate is 54% but plausible values range from 49% to 56%, calculate all three cases. In the 100-contract, $51.75-outlay example, expected profit ranges from −$2.75 at 49% to +$4.25 at 56%. Calling only the middle case “the edge” would hide the central uncertainty.

Do not interpret an informal range as a statistical confidence interval. Say how it was constructed: alternative assumptions, different model outputs, or a properly estimated uncertainty interval. Those methods answer different questions. To evaluate the probability process over time, use the model-testing guide.

Common questions

01Are maker orders always free?

No. Some markets charge maker fees when resting orders execute. Check the market’s current treatment and order details. Canceling an unfilled resting order is described separately in Kalshi’s fee guidance. Kalshi: Fees.

02Can I use a fee from an old blog post?

Use it only as historical evidence for the period it describes. For a new analysis, open the current official schedule and record the applicable inputs. A search snippet can remain visible after the underlying document changes.

03What should I do if the calculation is positive?

Check whether the probability estimate has earned that precision and whether the price can be obtained for the assumed size. Positive arithmetic is a conditional result, not an instruction to trade. Return to the full AI research workflow and document the strongest reason the estimate could be wrong.

Sources and freshness

Reviewed October 27, 2026. This is educational research, not a trade recommendation. All numerical trade examples are hypothetical. Recheck the exact contract, source dates, prices and fees before relying on an analysis. No model performance is claimed.

Official sources: Kalshi fee schedule, effective July 7, 2026; Kalshi API: Fee Rounding; Kalshi: Fees; Kalshi: Quick Orders; Kalshi: Limit Orders.