Cancelled Events and Scalar Settlement: Why Not Every Market Pays $1 or $0

Most explainers tell you a binary event contract pays $1 if you are right and $0 if you are wrong. That is the base case, and it is correct for the large majority of markets. Kalshi documents the standard flow plainly: trading stops at close, the result is verified against the market's named source, and the contract settles.

But a strategy that assumes every market resolves to a clean $1 or $0 will misprice its own backtest. Markets get cancelled. Some settle to a value strictly between the two bounds. A few resolve late, or not at all for a while. If your research model hardcodes a two-point payout, your P&L curve is fiction on exactly the paths that hurt the most. This post separates the cases, each with its own rule, and none of them a universal payout claim.

Case 1: The ordinary binary resolution

Start with what is settled. For a standard binary market, a winning contract credits $1.00 and a losing one credits $0.00, per Kalshi's own help documentation. There is no sliding scale on these and no separate settlement fee reducing the $1.00.

Two nuances worth keeping even here. First, a market staying open after the event appears decided is not a signal that payout criteria failed. Kalshi often has to wait for an official source agency to publish finalized data before it determines a market. Second, do not generalize the $1/$0 payout to scalar, protected, or exceptional contracts. The rest of this post is why.

Case 2: Cancellation (funds returned, not $0 or $1)

Cancellation is its own outcome, distinct from losing. When a market cannot be resolved as written, positions are typically voided and funds returned rather than settled to $0 or $1. Kalshi's emergency powers, housed in Rule 2.8 of the Rulebook, include among their listed reasonable actions 'Cancellation of a Contract and the return of any funds paid to enter Trades on the Contract.'

Kalshi has also self-certified contracts whose whole premise is a cancellation decision, for example a template asking whether an entity will cancel plans for a project, listed with the CFTC in July 2025. The lesson for a model: a cancelled market returns cost basis. It is not a loss and it is not a win, and treating it as either skews your hit rate and your fee accounting.

Case 3: Scalar settlement (a value between the bounds)

Some Kalshi markets finalize at a fractional value instead of a clean 0 or 1. In public market payloads this shows up as a scalar result with an explicit settlement value, and the long/YES side receives that settlement while the short/NO side receives one minus it. This is not an error. It is the designed outcome for certain contracts, formalized in Kalshi's February 2025 CFTC rulebook amendment for scalar contracts.

The most common driver is sports. Kalshi does not void a player prop when the player sits. If a player is ruled out or is active but never takes a snap, the contract settles at a fair price the Exchange sets, not at $0, $1, or your entry price. Once a player participates (in NFL markets, a single snap, even on a play wiped out by penalty), the contract settles on the stats actually recorded, regardless of when the player leaves. That means a DNP can be a gain or a loss depending on where you entered relative to the fair price.

If you come from a sportsbook background where 'did not play' means void, this is the single biggest trap. At least one trader publicly documented a significant loss after assuming a player prop would void on a DNP, when Kalshi instead settled at the exchange-determined fair price.

Case 4: Protected-player and ladder markets

A narrower carveout exists for protected markets. A protected NFL player market settles at its fair price if the player plays the first half but not the second half or overtime and has not reached the target. This is different from the plain DNP rule and different again from the one-snap rule. The point is that 'player markets' are not one rule. They are several, and the contract's own terms decide which applies.

Scalar values also ripple into combos. A combo leg that settles between $0 and $1 (Kalshi marks it with a blue arrow) does not void the combo. The payout is the product of every leg's final value. Per Kalshi's own help documentation: legs settling at $1.00, $1.00, and $0.70 pay $0.70 per contract. A model that treats a combo leg as all-or-nothing will overstate wins and misread the refunds that never happened.

Case 5: Unresolved, delayed, and discretionary outcomes

When an outcome cannot be cleanly determined, Kalshi's Rule 6.3(c) gives it discretion to interpret the terms and, where an expiration value cannot be determined, to use the last traded price to set the payout. Under this mechanic, longs receive the last traded price per contract and shorts receive one minus it. This is a hypothetical illustration of the mechanism; the actual settlement figure in any market depends on the specific circumstances at resolution time.

The headline stress test was the February 2026 market on whether Khamenei would leave as Iran's Supreme Leader. The market drew roughly $54 million in trading volume. Kalshi settled it at the last traded price before his death under a death carveout provision rather than paying YES $1, per contemporaneous reporting and Kalshi's own public statements. The carveout was disclosed in the CFTC-filed rules but less clearly on the user-facing page, and Kalshi subsequently refunded trading fees on the settlement and covered net losses out of pocket. Kalshi has since formally codified a dedicated death-settlement rule (Rule 6.3(e)) with the CFTC. The settlement rules themselves have changed more than once. Treat 'last traded price' as a possible outcome in your model, not an exotic footnote.

What this means for a research model

• Cancellation returns cost basis. Do not score it as a win or a loss, and account for fees accordingly.
• Scalar settlement pays a value in between. For the long side, use the settlement value; the short side gets one minus it.
• DNP in sports usually means fair-price settlement, not void. One snap flips it to a stats-based resolution.
• Combo legs multiply. A scalar leg shrinks the payout; it does not void the whole ticket.
• Unresolved markets can settle at last traded price under discretion. Preserve that branch.
• Rules differ by contract and change over time. Read the specific market's terms rather than a generic payout table.

Where a paper sandbox fits

If you are prototyping against current market data, this is exactly the kind of assumption worth isolating before you trust a curve. Banger is a paper-research beta for prediction-market strategies that runs code-first strategies against current Polymarket and Kalshi data without venue keys, with simulated fills and inspectable logs. Its PyPI package page states that live execution is not shipped in the public paper beta.

A paper result is evidence under a particular data, fill, fee, and valuation model. It is not proof of real-venue execution or profit. The useful move is to inspect how your own result model treats cancellation, scalar values, and last-traded-price settlement, rather than assuming a two-point payout and discovering the gap live. Distinguish marked value, a settlement claim, credited cash, and hypothetical proceeds. The edge cases above are where those four quietly diverge.