Sports Betting Odds Sample Dataset (SharpAPI)

Real odds snapshots from SharpAPI, the real-time sports betting odds API: 23 sources spanning sportsbooks, exchanges, and prediction markets. Free to use for research, coursework, backtesting experiments, and data journalism, with attribution (CC BY 4.0, see below).

Files

File	Rows	What it is
data/worldcup_2026_odds_snapshot.csv	6,132	Full odds board for the 2026 FIFA World Cup, captured 2026-07-13 (semifinals week: Argentina vs England, France vs Spain). 20 sources, 99 market types, from moneylines and Asian handicaps to correct score and outrights.
data/mlb_odds_snapshot.csv	3,673	MLB during All-Star week, captured the same night across 14 of those sources. Mostly a futures board (World Series, pennant, playoff and division odds; MVP/Cy Young/Rookie of the Year; All-Star selection contracts) plus a small set of regular-season game, inning, and player-prop markets.

Both files share one schema: each row is one price on one selection at one source at capture time.

Identifying a price / joining. A single price is event_id + sportsbook + market_type + selection + selection_type + line (add home_team for Kalshi/Polymarket multi-entity contracts). For the World Cup head-to-head matches, event_id is a clean cross-book join key — every source (bovada included) shares one id per match, so you can line-shop the same real event directly on event_id. Outright/futures and prediction-market rows use source-specific ids, as does most of the MLB file (a futures board), so join those on teams + market rather than event_id. Prediction-market sources encode markets as yes/no questions, so home_team may hold the question or contract entity.

Schema

Column	Type	Description
id	string	Price record id from the API. Unique except 16 betrivers pairs (World Cup) where one id spans both teams of a team-total market.
sportsbook	string	Book slug (draftkings, fanduel, pinnacle, novig, kalshi, ...)
event_id	string	Event id. Clean cross-book join key for World Cup matches; source-specific for futures, prediction-market, and MLB rows — see the joining note above
sport / league	string	e.g. soccer / fifa_-_world_cup. league is mostly canonical but not uniform: 83 saba MLB rows mlb_american_league, 8 bovada MLB award rows mlb_awards
home_team / away_team	string	Event participants
market_type	string	One of 99 market slugs (moneyline, asian_handicap, total_goals, ...)
selection	string	The specific outcome priced; for player props this is the player/entity, with the Over/Under side in selection_type
selection_type	string	side, total, over, under, outright; needed together with selection to uniquely identify player-prop and total markets
odds_american	int	American odds (+150, -110)
odds_decimal	float	Decimal odds
odds_probability	float	Implied probability (vig included)
line	float or empty	Handicap/total line where applicable
event_start_time	ISO 8601	Scheduled start. For prediction-market rows this is the market close/expiry time, not kickoff
is_live	bool	Whether the price was in-play at capture
timestamp	ISO 8601	Capture time of this price

Quick start

Python:

import pandas as pd
wc = pd.read_csv("data/worldcup_2026_odds_snapshot.csv")

# most-covered markets and sources
print(wc["market_type"].value_counts().head())
print(wc["sportsbook"].value_counts())

# a price is unique on this key (add home_team for prediction-market contracts)
key = ["event_id", "sportsbook", "market_type", "selection", "selection_type", "line"]
print("duplicate-key rows:", wc.duplicated(key).sum())   # 0 — the key is clean

# best price offered on each distinct outcome, within each source's own event ids
best = wc.loc[wc.groupby(key, dropna=False)["odds_decimal"].idxmax()]
print(best[["sportsbook", "market_type", "selection", "odds_american"]].head())

R:

wc <- read.csv("data/worldcup_2026_odds_snapshot.csv")
head(sort(table(wc$market_type), decreasing = TRUE))   # market coverage
head(sort(table(wc$sportsbook), decreasing = TRUE))     # source coverage

Ideas: cross-book vig comparison, best-line analysis, implied-probability calibration against results, market-efficiency studies across 23 sources, arbitrage detection exercises.

Want live or historical data?

This is a static snapshot. The live feed behind it: SharpAPI serves real-time odds from 45+ sportsbooks with sub-89ms SSE streaming, no-vig fair odds, +EV and arbitrage detection, plus historical odds and closing-line data on paid tiers. The free tier needs no credit card: sharpapi.io/pricing. SDKs: Python and TypeScript.

License and attribution

Data: CC BY 4.0. Use it freely, including commercially, with attribution:

Odds data from SharpAPI (sharpapi.io), Sports Betting Odds Sample Dataset, 2026.

BibTeX:

@misc{sharpapi2026odds,
  title  = {Sports Betting Odds Sample Dataset},
  author = {{SharpAPI}},
  year   = {2026},
  url    = {https://sharpapi.io},
  note   = {2026 FIFA World Cup and MLB odds snapshots, 23 sources}
}

Disclaimers

Odds are informational market data, not betting advice. Prices move constantly; this snapshot was accurate only at capture time. Sports betting involves financial risk, is legal only where permitted, and is restricted to those of legal age (21+ in most US states). Please wager responsibly.