Which leagues do the football datasets cover?

15 European leagues: Premier League and Championship (England), LaLiga and LaLiga2 (Spain), Bundesliga and 2. Bundesliga (Germany), Serie A and Serie B (Italy), Ligue 1 and Ligue 2 (France), Eredivisie (Netherlands), Liga Portugal, Jupiler Pro League (Belgium), Süper Lig (Türkiye) and Super League (Greece) — each present in every season from 2021-22 through 2026-27, with no gaps.

What are the three levels of closing odds?

Each match carries up to three closing odds, each in its own column: the Pinnacle closing line (the sharp bettors' benchmark), the market-average closing line, and the best closing price available among the tracked bookmakers. They are never mixed.

Does Pinnacle closing cover every season?

It is near-complete from 2021-22 through 2024-25 (99-100%). The historical source stopped publishing it on 17 January 2026: season 2025-26 is therefore covered in Pinnacle closing up to 16 January, while the average and best closing lines cover the entire season.

Where does the xG in the datasets come from?

xG (expected goals) measures the quality of chances created. Every row is labelled: xg_source is real when the xG comes from the post-match provider, or pseudo when it is computed by the ProbWin model from shots on and off target — the exact formula is published in the README.

Data Shop › Football

15 leagues · 6 seasons · no gaps

Football Datasets

27,109 matches from 2021-22 through 2026-27 (in progress), with results, xG and up to four odds families per match — including three levels of closing lines. Every file matches its product page exactly — detailed README included, coverage quantified column by column.

27,109Matches

15Leagues

6Seasons

3Closing levels

The 15 leagues covered each present in all 6 seasons

Premier League Angleterre

Championship Espagne

LaLiga Espagne

LaLiga2 Allemagne

Bundesliga Allemagne

2. Bundesliga Italie

Serie A Italie

Serie B France

Ligue 1 France

Ligue 2 Pays-Bas

Eredivisie

Liga Portugal Belgique

Jupiler Pro League Turquie

Süper Lig Grèce

Super League

Choose your dataset

1X2 & Results

The match-winner market: full-time and half-time score, result, xG, Bet365 and Pinnacle pre-match odds, and three closing lines (Pinnacle · average · best).

€9 per season · €29 for all 6 seasons
See the 6 products →

Over/Under 2.5 goals

The total-goals market: total and OVER/UNDER outcome, xG, and the same odds families on the 2.5 line — pre-match and closing.

€9 per season · €29 for all 6 seasons
See the 6 products →

Machine Learning dataset — 96 pre-match stats

To train your own models: 96 statistics computed before each kick-off (form, xG, rest days, standings…) plus the actual result to predict, across all 27,190 matches. Includes OAxG, our opponent-adjusted xG — a ProbWin exclusive. Verified leak-free.

€39 for all 6 seasons
See the product page →

2026-27 season in progress — weekly updates

The season being played, in all 3 formats (1X2, Over/Under, 96 ML stats): fresh edition every Monday, unlimited downloads until 30 June 2027. Buy once, re-download all season.

€12 for the whole season
See the product page →

The Complete Football Pack — all 3 datasets, −29%

1X2 & Results + Over/Under 2.5 + Machine Learning (96 stats including exclusive OAxG), all joinable on match_id. Bought separately: €97.

€97 €69 View details Buy the pack

Which dataset for which job?

You want to…	Get	Price
Backtest a strategy on the match-winner market	1X2 & Results	from €9
Backtest a strategy on goal totals	Over/Under 2.5	from €9
Train a prediction model (machine learning)	Complete Football Pack (ML + both odds datasets)	€69 instead of €97
Study a single season before committing	Any single season	€9
Follow the season being played, updated weekly	2026-27 season in progress	€12

All datasets share the match_id key: they join in one line of pandas.

# join odds + ML stats in 3 lines odds = pd.read_parquet("football_1x2_all_seasons.parquet") feats = pd.read_parquet("football_features_all_seasons.parquet") df = odds.merge(feats, on="match_id") # 96 stats + 4 odds families per match

What you are buying — no surprises

Three closing levels, each in its own column: Pinnacle, market average, best available.
Limit disclosed: Pinnacle closing stops on 16 Jan 2026 (source stopped publishing) — average and best closing cover the entire 2025-26 season.
Labelled xG row by row (real provider / pseudo ProbWin model, published formula).
Zero missing values on core columns; per-season coverage quantified in each README.
Single-user licence (your models, commercial included, are yours — resale of the files prohibited) · CSV + Parquet · 72 h link, 5 downloads.

Secure Stripe paymentOne-time price, no subscription.

Instant downloadLink sent as soon as payment clears.

Quantified README includedEvery column documented, coverage per season.

Personal support[email protected]

Frequently asked questions

Which leagues exactly?

Premier League, Championship, LaLiga, LaLiga2, Bundesliga, 2. Bundesliga, Serie A, Serie B, Ligue 1, Ligue 2, Eredivisie, Liga Portugal, Jupiler Pro League, Süper Lig and the Greek Super League — first AND second divisions of the big five. Each league is present in every one of the 6 seasons, no gaps.

Free odds data exists — why pay?

Because you are not buying raw odds: you are buying the work it would take weeks to redo. Multiple sources cross-checked and deduplicated, xG on every match (labelled, formula supplied), exclusive OAxG, 96 leak-free pre-match stats, a match_id key that joins everything, clean CSV + Parquet, and a README that quantifies each column's coverage — instead of leaving you to guess it.

Why three closing odds?

Serious backtesting is measured against the close (the last odds before kick-off). Pinnacle is the professionals' benchmark; the market average gives the consensus; the best closing price is what a line-shopping bettor could actually take. Three different jobs — you get all three.

What is xG?

xG (expected goals) measures the quality of the chances each team created: an xG of 2.3 means the match's chances were "worth" 2.3 goals on average. Every row states where its xG comes from: official provider (real) or ProbWin's shot-based calculation (pseudo, formula supplied).

Can I join the datasets together?

Yes: they all share the match_id key. 1X2 + Over/Under + Machine Learning join in one line of pandas.

What about the current season?

The bundle includes season 2026-27 as a season-to-date version (cut-off stated), with average and best closing lines (~70% to date, public archives complete them with a few days' delay). The file is frozen at the cut-off date stated on the product page.

Multi-bookmaker odds aggregated from public archives; "pseudo" xG, pre-match stats, deduplication and join keys computed by ProbWin. Secure Stripe payment. [email protected] · MLB datasets →