Odds API vs sports data API

The names sit close enough that people search for one meaning the other, and plenty of products need both.

What a sports data API gives you

A sports data API describes the sport itself: fixtures and kick-off times, live scores, results, league tables, line-ups, and player and team statistics. It answers who is playing, what the score is and how a side has performed this season.

None of that is a price. A sports data API can tell you Arsenal play Chelsea on Sunday and, later, that the match finished 2-1. What William Hill offered on Arsenal to win is a different question. Some providers sell both under one contract. They are still two datasets.

What an odds API gives you

An odds API gives you bookmakers' prices for outcomes: the odds each book shows on each selection in a market. For the same fixture it carries the match-result market (h2h in the feed) with a price on Arsenal, the draw and Chelsea from every book that prices it. The basics are in what an odds API is.

How events, markets, outcomes and offers nest is laid out in the anatomy of an odds API response.

The two side by side

Typical products in each category. Individual providers draw the line in different places.

Fixtures are where the two overlap

An odds feed has to say which event a price belongs to, so every event carries its sport, teams and kick-off time. OddsRelay's /v2/events lists the events on its matched boards, one sport at a time, with those fields and the boards that carry each.

That fixture data exists to anchor prices. It carries no scores, results or statistics. Since 2 November 2026 the feed has also carried in-play prices, each event marked in_play, but no match clock, so the score and everything else that happens on the pitch belong to a sports data API.

When you use both, the join between them is yours to build. Team names and kick-off times are the practical keys. OddsRelay's event_id is opaque and never changes once served, so keep the pairing in your own table from that id to the other provider's.

When you need both

Many products do. A previews site shows the fixture, the form table and the current prices side by side. The sports data API supplies the first two and the odds API the third.

Which one carries the weight depends on what breaks when it is wrong. If the product is about the sport, the sports data leads and the prices are decoration. If the product is about prices, as oddsmatchers and comparison sites are, the odds feed has to be right and the stats are context. Pick your main supplier for the half your product cannot run without, and treat the other as an add-on.

Matched betting runs on a pair of prices

Matched betting depends on the relationship between two prices for the same selection: a bookmaker's back price and an exchange's lay price. Scores and statistics cannot produce that pair. It comes only from odds data joined across bookmakers and exchanges.

OddsRelay's matched boards do that join. Each outcome carries back offers from bookmakers and lay offers from Betfair, Smarkets and Matchbook, where available is the money on offer at the lay price. Only pairs with enough lay money behind them make the board. The rating and qualifying loss are yours to compute, as below, and the oddsmatcher widget shows them.

{
  "event_id": "or_evt_917dd44bce05",
  "commence_time": "2026-09-20T14:00:00Z",
  "home_team": "Arsenal",
  "away_team": "Chelsea",
  "markets": [{
    "key": "h2h",
    "outcomes": [{
      "name": "Arsenal",
      "back": [{ "bookmaker": "william_hill", "price": 2.9, "link": null }],
      "lay":  [{ "exchange": "smarkets", "price": 3.0, "available": 175, "link": null }]
    }]
  }]
}

const back = 2.9, lay = 3.0, commission = 0.05, stake = 10;

const rating = (back / lay) * 100;                      // 96.7
const layStake = (stake * back) / (lay - commission);   // 9.83
const qualifyingLoss = stake * (back - 1) - layStake * (lay - 1); // -0.66

If you want prices without the pairing, raw carries each covered bookmaker's prices per market, each market with its own last_update.