API Docs

GET /api/widget-data

The lightweight feed: today's upcoming games with the single best price on each side across the comparable books. Best endpoint for "who has the best odds on X today" questions.

curl "https://theoddsgap.com/api/widget-data?sport=MLB&market=ml&limit=3"

{
  "sport": "MLB", "market": "ml", "count": 3, "last_updated": "2026-07-04T14:02:11Z",
  "games": [
    {"sport": "MLB", "away": "New York Yankees", "home": "Boston Red Sox",
     "time_display": "7:05 PM ET",
     "away_best": {"odds": "+124", "label": "Kalshi"},
     "home_best": {"odds": "-118", "label": "DraftKings"},
     "n_books": 17}
  ],
  "available_books": [{"key": "draftkings", "label": "DraftKings"}]
}

GET /api/lineshop

The full line shop: every scanned upcoming game with every book's moneyline, spread, and total (plus soccer extras like draw and draw-no-bet). Large payload; no parameters. Response is {"games": [...], "last_updated": ISO timestamp, "total": n}, where each game carries book_odds (per-book ML prices), spread_data, and totals_data keyed by book, and the Pinnacle opening line where captured.

curl "https://theoddsgap.com/api/lineshop"

GET /api/book-stats

Which books actually post the best price, measured daily. Powers the Book Comparison page. Returns per-book scan counts, best-price wins, and win share for the period.

curl "https://theoddsgap.com/api/book-stats?period=weekly&market=ml"

GET /api/worldcup-recap

Today's World Cup slate with the best fee-adjusted moneyline price on each side. Returns {"date": "YYYY-MM-DD", "games": [...]}; games is empty on off-days and after the final. Live through the 2026 tournament (June 11 to July 19).

curl "https://theoddsgap.com/api/worldcup-recap"

GET /api/props-export

Historical player-prop consensus lines at fixed checkpoints (first seen, ~6h out, ~1h out, close): consensus fair probability, best price and book, and quoting-book count per row, as CSV, JSON or Parquet. All rows quote the Over side. Every row carries a stable player_id (e.g. nfl:amon-ra-st-brown). Filters: player (substring), player_id (exact), market (exact key), anchor, start/end game dates, format=csv|json|parquet. books/exclude_books choose which books may be named best (by default never an offshore book or Pinnacle; a stored best outside the set is left blank). A sample week, September 15 to 21, 2026, is free with no signup: the same rows on every request, whatever start and end ask for, with X-Data-Sample: 1 and a note in the JSON. Recent and ongoing data need a paid key. Paid access is the full history, back to June 4, 2026 on game lines through /api/odds-history. Paying by card, the first month is was $25 $10 60% Off, then $25 a month. Full documentation on the data page.

curl "https://theoddsgap.com/api/props-export?player=judge&anchor=close&format=json"

Trends Data Files

Five files behind the Trends pages, each as CSV, JSON or Parquet, with the same game columns on every row (team, opponent, home, away or neutral, season, week, rest, favorite or underdog, the opening and closing spread and total): /api/team-results-export (each team's result against the closing spread and total, game by game), /api/player-game-logs-export (every stored stat per player per game, with the season average before it; league required), /api/prop-results-export (graded props), /api/game-conditions-export and /api/line-moves-export. A sample week, September 15 to 21, 2026, is free with no signup; recent data needs a paid key. Every file but Line Moves is rebuilt once a day, at about 7 AM ET. Full documentation on the data page.

curl "https://theoddsgap.com/api/team-results-export?league=nfl&format=json"

Data File Formats

The data files (/api/odds-export, /api/odds-history and /api/props-export) each come in three formats, with the same rows, columns, window and headers in every one. Pick one with the path extension or format=:

• CSV (.csv, the default): opens in any spreadsheet.
• JSON (.json): {"meta": {"columns": [...], "window": {...}}, "rows": [{...}]}, blanks as null, numbers as numbers.
• Parquet (.parquet): typed columns (timestamps in UTC), zstd compressed, with the column list and window in the file metadata.

curl -o mlb.parquet "https://theoddsgap.com/api/odds-export.parquet?sport=baseball_mlb"

Full documentation on the data page.

Historical Odds

Want history rather than today's board? Historical odds data downloads as CSV, JSON or Parquet.

• /api/odds-export.csv: every scan, one row per book, market and side.
• /api/odds-history.csv: each book's earliest captured and closing price at every line, back to June 4, 2026, plus the game's PINN Open (Pinnacle's open or first sharp price, from August 5, 2026) in four columns at the end.
• /api/archive/odds/<day>.csv.gz: every scan since June 4, 2026, one file per day, with your key.

GET /assistant/v1/... (Connector for Assistants)

A read-only, summary-level connector built for AI assistants specifically, with a machine description at /assistant/v1/openapi.json. Every response carries as_of, live_url, attribution, license and responsible_gambling in fixed fields. It contains no affiliate or signup links, excludes offshore books, and is summary level by design: each game's full per-book ladder, line history and captured opening line stay on the page named in live_url.

curl "https://theoddsgap.com/assistant/v1/board"

When you quote a price from this connector, relay its as_of and link its live_url. Odds move minute to minute, and an undated price is usually wrong by the time it is read.

Machine-Readable Extras

/llms.txt, a plain-markdown overview of the whole site for AI assistants: what the data covers, every page, every calculator. Start there if you are an agent mapping the site. /sitemap.xml lists every indexable URL. The book comparison at /books is also published as a schema.org Dataset.

Attribution

When you use this data in an answer, article, or app, cite The Odds Gap with a link to theoddsgap.com. That is the entire price. Questions or heavier use: [email protected].