By The Closing Line

Updated October 25, 2026 with Jonah Reed · Fictional AI writer for The Closing Line

A positive expected-value calculation says something conditional: if the probability and cost inputs describe the trade accurately, its modeled average payoff exceeds its cost. It does not establish that your probability estimate is correct. A sportsbook comparison, a scanner result, and an executable Kalshi quote each answer a different question.

This guide keeps those questions separate, then changes one assumption to see whether the result survives. The worked numbers are synthetic, not a live opportunity or a recommended position size.

What +EV actually means

Assume an ordinary binary YES contract paying $1 if Yes resolves and $0 otherwise, held through settlement with no special payout adjustment. For a filled quantity n, estimated Yes probability p, entry price P in dollars, and total entry fee F:

Expected net value = n × p − n × P − F

The first term is expected payout; the other terms are assumed entry costs. This model excludes early selling, software, funding costs, taxes, and the time money remains committed. Change the model when those matter. It is not a formula for perpetual futures or nonstandard settlement terms.

The result is a probability-weighted average across the stated outcomes, not the cash you receive in either outcome. Buying more contracts on the same result does not produce more independent observations of whether your forecast was right.

Step 1: State the probability estimate and its limits

Write down the exact event, the estimate, its source, and when it was made. Separate a model forecast from a market-derived reference. Then record what new information would make you revise it.

Match the proposition before comparing numbers. Kalshi says each market has its own rules and outcome-verification source. A forecast about a game winner may not answer a contract about a scoring threshold, a particular observation window, or an exceptional result. Read the market rules; use our close-versus-settlement guide when the timing is unclear.

Step 2: De-vig the market you're comparing to

Decimal odds describe total return including stake, as Pinnacle explains. Taking their reciprocal gives a raw implied probability. Here is a hypothetical complete two-outcome market, with matched rules and both quotes recorded at the same time:

  • Outcome A at decimal 1.75: 1 ÷ 1.75 ≈ 57.14%.
  • Outcome B at decimal 2.25: 1 ÷ 2.25 ≈ 44.44%.
  • Together: approximately 101.59%, rather than 100%.

One proportional normalization divides each unrounded reciprocal by their sum. For A, that is (1 ÷ 1.75) ÷ [(1 ÷ 1.75) + (1 ÷ 2.25)] = 56.25%. B becomes 43.75%.

The arithmetic removes the excess total under that proportional allocation. It does not discover the event's true probability. Keep 56.25% as a comparison reference, with its assumptions attached. A missing outcome or mismatched contract makes this two-outcome example unsuitable.

Step 3: Check the price and quantity you can actually get

Kalshi's order book displays resting quantities at their prices. An ask is the seller's offered price; it is not the same thing as the last recorded trade. Our bid, ask, and last-trade guide explains the distinction.

Record the side, quantity available, and timestamp beside the quote. A spreadsheet using 50¢ for every contract has assumed that the full example quantity fills there. If it does not, recalculate using the actual fills and fees. A scanner can help organize comparisons; a displayed disagreement alone cannot validate the probability input.

Step 4: Subtract the fees — every time

Check the applicable market and order treatment rather than applying one fee to every contract. Kalshi's fee guidance notes market-specific differences and maker fees in some markets. The published schedule, checked October 25, gives the general taker calculation using a multiplier, quantity, and price. Our maker-versus-taker guide covers the execution tradeoff.

For the synthetic example, assume a single fill of 100 contracts at $0.50, all taking liquidity under the general schedule with multiplier 1. The model fee is 1 × 0.07 × 100 × 0.50 × 0.50 = $1.75. That exact amount needs no additional rounding here. Actual fill accounting can involve rounding adjustments and rebates; see Kalshi's rounding documentation. Use the recorded charge for a completed trade.

A hypothetical example

Suppose you use the 56.25% reference as the probability input. This is an assumption for the calculation, not a claim that the reference is accurate. Entry costs $50 plus the $1.75 fee, or $51.75.

Expected net value = $56.25 − $50 − $1.75 = +$4.50

But the two realized outcomes in this simplified example are +$48.25 if Yes pays $100, or −$51.75 if Yes pays nothing. Neither is +$4.50. The expected value weights those outcomes by the assumed probabilities; it does not predict which will occur.

Step 5: Change the least certain input

Keep the fill price, quantity, and fee unchanged, but lower the estimate to 51%. Expected net value becomes $51 − $50 − $1.75 = −$0.75. The break-even estimate is 51.75% under these assumptions.

That comparison identifies what needs defending: the probability input. It does not establish a confidence interval or tell you how much to buy. A stake formula cannot verify an estimate, and repeating this arithmetic does not turn a mistaken forecast into an advantage.

Common ways people fool themselves

  • Calling an adjusted market quote the true probability.
  • Comparing different rules, observation windows, or stale quotes.
  • Treating a displayed price as a full-quantity fill.
  • Reporting expected value as an observed profit.
  • Leaving fees or excluded costs outside the decision.

The repeatable process

Keep a short worksheet:

  • Contract: exact question, rules, and comparison-market match.
  • Probability: source, timestamp, comparison quotes, and lower-estimate scenario.
  • Execution: side, price, quantity, and fee basis.
  • Limits: excluded costs and evidence that would invalidate the estimate.

An unresolved field is a reason to keep investigating.

Commercial disclosure: The Closing Line has a commercial interest in referrals to Bot for Kalshi. If you already have an explicit rule, its documented builder workflow describes reviewing supported conditions and limit-order actions, then using Paper before permitting live orders. Paper models activity without submitting exchange orders. That can help inspect instructions; it does not validate your forecast or establish profitable live results.

Update note: This revision replaces unsupported “true probability,” growth, and automatic stake-sizing claims with a conditional calculation and sensitivity check. It removes the earlier calculator references and the previous illustration, which presented assumed value as established worth. The original publication date and URL are retained.

About the update editor

Jonah Reed is a fictional AI editorial persona created by The Closing Line. The name represents a consistent data and probability writing voice, not a real person or independent financial professional. The original article remains credited to The Closing Line. The Closing Line is responsible for this content and its corrections. Questions or corrections.