Build your MLB model
this weekend
7,070 games, Pinnacle pre-match odds, 45 ML features, 65K line movements.
Zero feature engineering, zero empty cells.
New: Football datasets — 27,109 matches, 15 leagues, from €9 →
The same data and pipeline feed our live prediction models. This isn't theoretical data — it generates verifiable results every day.
Every pick produced by these pipelines is published then settled publicly, result by result — see the full journal on /results/.
4 datasets, one goal: backtest
Pick your dataset or combine them via the game_pk key — one line of code.
MLB Full-Game Odds — Pinnacle (2023-2025)
Pinnacle Full-Time odds: totals, spreads, moneyline + final scores. The base dataset for any MLB model. 3 seasons (2023-2025).
game_pk,game_date,home,away,total_line,over_price,under_price,... 745234,2024-04-01,NYY,HOU,8.5,1.87,1.95,... 745235,2024-04-01,LAD,SF,7.5,1.91,1.91,...
MLB First-5-Innings Odds — Pinnacle (2023-2025)
First 5 Innings odds: isolate the starting pitcher duel without bullpen noise. The MLB modeller's favourite market. 3 seasons (2023-2025).
game_pk,game_date,home,away,f5_total_line,f5_over_price,... 745234,2024-04-01,NYY,HOU,4.5,1.83,... 745235,2024-04-01,LAD,SF,4.0,1.95,...
MLB Modeling Features (2023-2025)
45 pre-computed features + the outcomes to predict. xERA (expected ERA), velocity, bullpen, weather, park factors. Import into scikit-learn and train. 3 seasons (2023-2025).
game_pk,home_sp_xera,away_sp_xera,home_sp_velo,park_factor,... 745234,3.42,4.15,95.2,1.12,... 745235,2.98,3.67,96.1,0.94,...
MLB Line Movements — Pinnacle (2023-2025)
65,350 timestamped snapshots. Opening line identified. Calculate your CLV (your price vs the closing line), spot sharp line moves, backtest timing strategies. 3 seasons (2023-2025).
game_pk,timestamp,market,line,over_price,is_opening 745234,2024-03-30T14:22Z,totals,8.5,1.91,1 745234,2024-04-01T16:45Z,totals,9.0,1.90,0
ProbWin vs. competitors
The only provider that uses its own data in production — with verifiable results.
| ProbWin | Odds Warehouse | BigDataBall | |
|---|---|---|---|
| Source | Pinnacle | Multiple (aggregated) | N/A |
| FT Totals + prices | ✓ | ✓ | ✗ |
| First-5-Innings | ✓ | ✗ | ✗ |
| Run Lines + prices | ✓ | ✗ | ✗ |
| Line Movements | 65K snapshots | Open/Close only | ✗ |
| ML Features | 45 columns | ✗ | Player stats |
| Advanced pitcher metrics | ✓ (xERA, velocity) | ✗ | Basic stats |
| Format | CSV + Parquet | CSV | CSV |
| Zero NULLs | Guaranteed | Not guaranteed | Not guaranteed |
| Price (3 seasons) | €19 | $79 | $90 |
| Used in production | ✓ | ✗ | ✗ |
Who is this for?
ML modelers
45 pre-computed features + outcomes to predict. Import into scikit-learn, XGBoost, or LightGBM and train a model in 10 minutes.
Quantitative bettors
Backtest your strategy across 3 seasons. Calculate your CLV with 65K snapshots. Identify exploitable lines.
Researchers & students
Clean data for papers, studies, sports statistics courses. Zero cleaning needed, focus on analysis.
Season coverage
Each MLB regular season has 2,430 games. Here's the exact coverage.
| Season | Games | FT Odds | F5 Odds | ML Features | Line Movements |
|---|---|---|---|---|---|
| 2023 | 2,430 | 2,333 (96.0%) | 1,964 (80.8%) | 1,822 (75.0%) | ~18K snapshots |
| 2024 | 2,430 | 2,408 (99.1%) | 2,425 (99.8%) | 2,296 (94.5%) | ~19K snapshots |
| 2025 | 2,430 | 2,329 (95.8%) | 2,346 (96.5%) | 2,105 (86.6%) | ~28K snapshots |
Coming soon
Frequently asked questions
game_pk key. A simple pd.merge(ft, features, on='game_pk') is all you need to combine any datasets.From €14 — One-time purchase, instant download
Pick your dataset and start backtesting in 5 minutes.
Explore datasets