The settlement source is the contract

An event contract pays one unit if something happens and nothing if it does not. That is how most people describe it, and it is slightly wrong in a way that matters. The contract pays if its settlement source reports that the thing happened, under the rules written for it, at the time the rules say. Most of the time the two are the same. The trades that go badly tend to sit in the gap between them.

What a contract actually specifies

Every listed event contract carries a rulebook, short or long, and it answers four questions. What is the underlying: a final score, a closing price, a temperature, a count. Which source reports it: a league's official statistics, a named exchange or index, a weather station, a resolution process. When is the source read: the end of regulation time, a fixed minute in a fixed time zone, the first report after the event. And what happens when something goes wrong: a postponed match, a source that is down, a figure that is later revised.

A trader who knows the answers to all four is trading the contract. One who knows only the headline is trading his idea of it.

Where the gap opens

Extra time. "Team A wins" can mean in regulation, including overtime, or including a shootout, and the rulebook picks one. A position that is right about the match and wrong about the definition is simply wrong.

Postponements and abandonments. Contracts handle them differently. Some settle on the eventual result if the event is completed within a window, some resolve to a fixed value, some are voided and return the stake. The price of a contract on a match under a weather warning should reflect which of these applies, and often does not.

Revised data. Official statistics are corrected after the fact: a goal reassigned, a play reviewed, a weather observation quality-controlled the next morning. The rules say whether the contract settles on the first report or the final one. If it settles on the first, the correction changes the record and not the payoff.

Price references. A contract on "the price at 16:00" needs a defined price: a last trade, a midpoint, an average over a window, an index built from several venues. Each behaves differently around the reference time. A single print can be moved by one trade; an average over a minute across several venues is far harder to push.

Station and instrument. A temperature contract settles on one station, not on the city. Stations differ from each other, and from the forecast grid, by more than most people expect, and an official daily maximum can come from a reading at an hour nobody was watching.

Resolution by process. Some markets resolve through a proposal and dispute process instead of a feed: someone asserts the outcome, others may challenge it within a window, and a vote or an arbiter settles disputes. These markets pay on what the process accepts. Clear outcomes resolve cleanly; loose wording invites argument, and the argument has a price.

Why it matters for pricing, not only for risk

The settlement source decides what information is worth having. If a contract settles on an official feed, the fastest faithful copy of that feed is the most valuable input a trader can hold, because everything the market will eventually pay on flows through it. Every other signal, a broadcast, a social post, a spectator's view, is useful only to the extent that it predicts what the source will say, and when it will say it.

The source also shapes the end of a contract's life. Close to resolution the price of a binary contract should run to zero or one. When it sits at 0.97 instead, the remaining three cents are a price on the residual: the chance of an error, a revision, a dispute or a rule nobody read. Sometimes that is an easy three cents. Often it is a fair fee for a real risk, and the trader who collects it many times is paid well until the one time he is not.

Reading a market before it trades

For every contract type we trade, the rulebook is read before the first order: the underlying, the source, the time, and every exception. Then the record is checked against it. Past resolutions are set against the source's own data, to see whether the rules are applied as written and how disputes have gone. Wherever the source can be recorded, we record it, with our own timestamp on every update, so that each settlement can be reproduced from our own data and any difference between our view and the official one shows up before it costs anything.

None of this is glamorous. It is, though, where a good share of the losses that look like bad luck turn out to have come from: a position that was right about the world and wrong about the contract.