Cameronkelly92/nba-closing-lines-ats-2025-26

NBA 2025-26 closing lines and ATS results

Every game of the 2025-26 NBA season — 1224 regular-season, 6 play-in and 85 play-off games (the NBA Cup knockout games are listed by the source as a separate competition and are not included) — with the betting market's consensus opening and closing lines (spread, total, moneyline win probability) and every bet already graded: against the spread (ATS), over/under and moneyline. One row per game, ready for pandas, R, SQL or a spreadsheet.

Built for backtesting, closing line value (CLV) research and sports-analytics teaching.

Quick look

Games	1315
Home teams covering the closing spread	51.9% (pushes: 0)
Overs at the closing total	49.5% (pushes: 0)
Closing favourites winning outright	68.7%
Closing home-win probability (average) vs home wins	55.2% vs 55.6%
Bookmakers per game (median)	23

import pandas as pd
df = pd.read_csv("hf://datasets/minmaxxxer/nba-closing-lines-ats-2025-26/nba_2025_26_closing_lines_ats.csv")
df.groupby("home_team").ats_result.value_counts().unstack()     # ATS record of every team at home

Columns

Column	Meaning
game_id	Stable game ID (Flashscore)
date_et, start_time_utc	Game day in US Eastern time; start in UTC
season_type, stage, round	regular / postseason; Regular season, Play-In or Playoffs; play-off round (First Round … NBA Finals)
home_team, away_team, home_team_id, away_team_id	Teams and their stable IDs
home_score, away_score, overtime	Final score, overtime included
bookmakers	Bookmaker brands in the moneyline consensus
open_home_win_prob, close_home_win_prob	No-vig home win probability at the opening and at the close (overtime included)
close_fair_home_odds, close_fair_away_odds	Fair (no-vig) decimal moneyline at the close
open_spread_home, close_spread_home	The market's main spread, home handicap (-5.5 = home favoured by 5.5)
close_home_cover_prob	No-vig probability that the home team covers the closing spread
open_total, close_total, close_over_prob	Main total at the opening and the close; no-vig over probability at the close
moneyline_winner, favorite, favorite_won	Winner (home / away; push = a tie), closing favourite and whether it won
ats_result, ats_margin, ats_result_open	Side that covered the closing spread (home/away/push), home margin + spread, same on the opening spread
total_result, total_margin, total_result_open	over/under/push on the closing total, points minus total, same on the opening total

How it was made

Opening and latest (= closing, once the game started) odds of every bookmaker that priced the game were collected; each bookmaker's margin was removed proportionally and the median across bookmaker brands taken (a brand listed under several licences counts once). The main spread and total are the lines most bookmakers quoted closest to 50/50. Bets are graded the way US books settle them: moneyline, spread and total include overtime.

Want more?

This file is a free sample. The NBA, NFL, MLB & NHL Betting Odds Actor produces the same data — plus the opening and closing price of each of the 25 bookmakers that priced this season (DraftKings, FanDuel, BetMGM, bet365, Betway …), every alternate line, team statistics and scores by quarter — for any season back to 2009, any date range, today's games, and the NFL, MLB, NHL, WNBA and college leagues. Python and API access, CSV / Excel / JSON export, pay per game.

Prefer a plain REST API? The Historical Sports Odds API on RapidAPI serves the same games with every bookmaker's opening and closing odds and graded bets — NBA, NFL, MLB and NHL since 2009, Europe's top soccer leagues since 2005 — with a free plan to try it.

Data compiled from publicly available game and odds information; not affiliated with the NBA, Flashscore or any bookmaker. For research and analysis.