Market line, defined

It is built in three steps. Prices are collected from two or more sources. Each source is de-vigged independently to remove its margin. The median of the resulting fair probabilities is scaled to total 100%. That becomes the published line.

Two properties are deliberate. The line is published only when at least two distinct sources contribute, so it never reflects a single opinion — though two is a floor, not a promise of depth. Some markets genuinely rest on only a couple of sources, and the data is honest about that rather than implying broad coverage. And no individual source is ever named, and no per-book price is ever served — only the derived aggregate.

What comes back is the market-implied win probability, delivered flat, with one row for each outcome, each tagged source: eo_market, not a nested structure. A book_count field ships on every row, but it is data for your own use, not a marketing claim.

The honest framing of the line stays "a de-vigged aggregate combined from multiple bookmakers and exchanges," never "odds from N books."

The implied probabilities (1 ÷ price) of a market's outcomes at one line add up to 100%, so no margin is left to strip out. That is a property of the method, not proof that the probabilities are right. Compare the line against your own model; don't read it as a recommendation. For a comparison of ways to turn betting odds into probability forecasts, see Štrumbelj, 2014.

Our own model line, eo_model, is coming soon, once it clears validation.

Read the methodology, or see reading the CS2 market odds line via the API.

The same arithmetic on your own numbers: the free no-vig calculator.