This dataset requires a paid license. Please purchase access at the link below and provide your Hugging Face username. Once purchased, submit an access request on this page. Access is granted manually.

NBA Back-to-Back Player Trends Dataset for Prop Research

Question this dataset helps answer: Which NBA players historically hold or lose minutes, points, rebounds, assists, and PRA output when they enter back-to-backs, third games in four nights, and other low-rest schedule spots?

This dataset helps sports researchers, fantasy analysts, and player-prop model builders compare low-rest NBA player spots using pregame rest buckets, compressed-schedule flags, rolling player form, home-away context, and archived spread-total fields where available.

Use the public sample to check:

Which players in the sample are entering true back-to-back games versus one-day-rest or longer-rest spots
Which columns are pregame schedule or rolling-form features versus postgame stat outcomes
Which teams, players, and seasons are represented before buying the full multi-season file

Preview vs full dataset:

Preview: 10000 rows and 98 columns
Full: 243065 rows and 98 columns

Full dataset: https://thearticulated.gumroad.com/l/nba-back-to-back-player-trends-prop-research

This product is built around a specific recurring workflow problem. Plenty of public NBA box scores exist, but the buyer question is not "where can I find points and rebounds?" The real problem is that analysts who care about player props, fantasy research, schedule fatigue, or workload effects usually have to rebuild the same joins over and over again. They need the player game log, the team's prior schedule spacing, the opponent's rest situation, a way to identify back-to-backs and compressed stretches, enough historical depth to compute rolling form, and a clean line between features that were known before tip-off and outcomes that only existed after the game finished. This dataset packages that structure directly.

Each row represents one player in one completed NBA regular-season or playoff game. The coverage spans the 2017-18 through 2025-26 seasons and includes 243065 player-game rows across 98 columns. That row volume matters. It is large enough for dashboard analysis, historical filtering, rolling-stat research, feature engineering, and exploratory modeling without reducing the product to a tiny case-study sample. Buyers can group by player, team, season, rest bucket, home-away context, or rolling-form state and still retain meaningful sample sizes for comparison.

The pregame side of the table is built around schedule stress and recent form. Key feature groups include days_since_prev_game, rest_bucket, is_back_to_back, games_in_last_4_days_including_today, is_third_game_in_four_nights, games_in_last_6_days_including_today, is_fourth_game_in_six_nights, opponent_days_since_prev_game, rest_advantage_days, is_rest_advantage, starter_flag, and multiple rolling averages such as rolling_minutes_last_5, rolling_points_last_5, rolling_rebounds_last_5, rolling_assists_last_5, rolling_pra_last_5, and rolling_fantasy_points_last_5. These fields are designed to support the exact questions that come up in low-rest research: whether a player's role holds, whether pace and workload shift, and whether recent usage or production behaves differently when rest is compressed.

The postgame side of the table holds the actual outcomes. Those outcome columns include minutes, points, rebounds, assists, steals, blocks, turnovers, plus-minus, team score, opponent score, team margin, PRA combinations, and a simple fantasy-style scoring proxy. That separation matters. A buyer can use the pregame fields as features and keep the postgame fields as targets, which makes the dataset usable for responsible historical modeling or descriptive analysis. It also makes the sample immediately interpretable for dashboard work because the logic of "known before the game" versus "observed after the game" is visible in the schema.

Archived closing spread and total context are also included where the public source archive supports them. Those columns are intentionally useful but not over-promised. The line archive covers an earlier portion of the multi-season window rather than the entire file, so the dataset includes a has_archived_closing_line_context flag. Buyers who want purely schedule-and-role analysis can ignore the line fields. Buyers who want to compare low-rest player output against game-level spread and total environments can filter into the covered subset. That is a better product design than pretending every row has equal betting-line depth when the public archive does not support that claim.

Responsible use note

This dataset is for research, analysis, dashboard building, educational use, and exploratory modeling. It does not provide betting advice, picks, guarantees, or predictions of future outcomes. It should not be marketed as a shortcut to guaranteed profit, and it should not be interpreted as proof that any single schedule angle automatically creates an edge.

Modeling notes

Each row represents one player in one completed NBA regular-season or playoff game.
Pregame features include rest spacing, compressed-schedule flags, opponent rest context, home-away status, starter flag, and rolling player form.
Postgame outcomes include minutes, points, rebounds, assists, steals, blocks, turnovers, PRA combinations, and game-result fields.
The coverage window runs from the 2017-18 season through the 2025-26 season.
The full dataset contains 243065 rows, which is large enough for train-test splits by season, by calendar cutoff, or by player history depth.
A sensible modeling workflow would split by season or by date rather than shuffling randomly, because the product is inherently time-ordered.
The line archive is partial rather than universal, so any workflow that depends on closing spread or total fields should filter on has_archived_closing_line_context.
Starter status is included because it is analytically useful, but buyers should remember it comes from the recorded game log rather than a separate pre-tip lineup feed.
The dataset is descriptive infrastructure, not a finished predictive system.

Why the free sample matters

The public sample is not filler. It proves that the core buyer task is real. A buyer can open the sample and immediately test whether the file contains the right player IDs, team context, rest-bucket logic, rolling features, and outcome columns. They can verify which seasons are covered, which players appear, how the schema separates pregame context from postgame results, and whether the low-rest workflow fits their notebook or dashboard plan. That is exactly what a public discovery asset should do: prove usefulness before the full purchase.

Source and build notes

The build uses public SportsDataverse NBA player box score and schedule files, then engineers team rest context and rolling player features locally. The process is reproducible, versionable, and avoids paid odds feeds. That keeps the dataset commercially useful without depending on a brittle or private source stack. The output is a single analysis-ready table rather than a collection of raw files that still need substantial cleaning and joins.

Dataset Preview

record_id	game_id	season_end_year	season_label	season_type_label	game_date	player_id	player_name	player_short_name	player_position	player_position_abbreviation	team_id	team_abbreviation	team_display_name	opponent_team_id	opponent_team_abbreviation	opponent_team_display_name	home_away	is_home	starter_flag	is_playoff_game	venue_full_name	venue_address_city	venue_address_state	venue_indoor	days_since_prev_game	rest_bucket	is_back_to_back	games_in_last_4_days_including_today	is_third_game_in_four_nights	games_in_last_6_days_including_today	is_fourth_game_in_six_nights	opponent_days_since_prev_game	opponent_is_back_to_back	rest_advantage_days	is_rest_advantage	is_rest_disadvantage	has_archived_closing_line_context	closing_spread_line	closing_total_line	team_favorite_flag	pickem_flag	team_implied_total	opponent_implied_total	minutes	field_goals_made	field_goals_attempted	three_point_field_goals_made	three_point_field_goals_attempted	free_throws_made	free_throws_attempted	defensive_rebounds	rebounds	assists	fouls	plus_minus	points	team_score	opponent_team_score	team_margin	won_game	total_points	points_rebounds_assists	points_rebounds	points_assists	rebounds_assists	simple_fantasy_points	points_per_minute	rebounds_per_minute	assists_per_minute	three_point_attempt_rate	free_throw_attempt_rate	prior_games_played	season_to_date_games	rolling_minutes_last_3	rolling_minutes_last_5	rolling_points_last_3	rolling_points_last_5	rolling_rebounds_last_5	rolling_assists_last_5	rolling_pra_last_5	rolling_three_pa_last_5	rolling_fantasy_points_last_5	season_to_date_minutes_avg	season_to_date_points_avg	season_to_date_rebounds_avg	season_to_date_assists_avg	home_points_avg_before_game	away_points_avg_before_game	same_rest_bucket_points_avg_before_game	source_url	source_domain	last_collected_at
400974838_1006	400974838	2018	2017-18	regular_season	2017-10-29 00:00:00	1006	Richard Jefferson	R. Jefferson	Forward	F	7	DEN	Denver Nuggets	17	BKN	Brooklyn Nets	away	False	False	False	Barclays Center	Brooklyn	NY	False	2	one_day_rest	False	2	False	3	False	2	False	0	False	False	True	-5.5	222.5	True	False	114	108.5	2	0	0	0	0	0	0	1	1	0	0	-2	0	124	111	13	True	235	1	1	0	1	1.2	0	0.5	0	0	0	0	0																	https://raw.githubusercontent.com/sportsdataverse/hoopR-nba-raw/main/nba/json/final/400974838.json	github.com, raw.githubusercontent.com	2026-06-11 17:31:07
400974865_1006	400974865	2018	2017-18	regular_season	2017-11-01 00:00:00	1006	Richard Jefferson	R. Jefferson	Forward	F	7	DEN	Denver Nuggets	28	TOR	Toronto Raptors	home	True	False	False	Ball Arena	Denver	CO	True	2	one_day_rest	False	3	True	4	True	2	False	0	False	False	True	-2	217	True	False	109.5	107.5	5	0	0	0	0	1	2	0	0	0	0	-14	1	129	111	18	True	240	1	1	1	0	1	0.2	0	0	0	0.4	1	1	2.0	2.0	0.0	0.0	1.0	0.0	1.0	0.0	1.2	2.0	0.0	1.0	0.0			0.0	https://raw.githubusercontent.com/sportsdataverse/hoopR-nba-raw/main/nba/json/final/400974865.json	github.com, raw.githubusercontent.com	2026-06-11 17:31:07
400974886_1006	400974886	2018	2017-18	regular_season	2017-11-04 00:00:00	1006	Richard Jefferson	R. Jefferson	Forward	F	7	DEN	Denver Nuggets	9	GS	Golden State Warriors	home	True	False	False	Ball Arena	Denver	CO	True	1	back_to_back	True	3	True	4	True	2	False	-1	False	True	True	9	232.5	False	False	111.75	120.75	6	2	3	1	1	1	2	1	1	0	1	2	6	108	127	-19	False	235	7	7	6	1	7.2	1	0.166667	0	0.166667	0.333333	2	2	3.5	3.5	0.5	0.5	0.5	0.0	1.0	0.0	1.1	3.5	0.5	0.5	0.0	1.0			https://raw.githubusercontent.com/sportsdataverse/hoopR-nba-raw/main/nba/json/final/400974886.json	github.com, raw.githubusercontent.com	2026-06-11 17:31:07
400974949_1006	400974949	2018	2017-18	regular_season	2017-11-13 00:00:00	1006	Richard Jefferson	R. Jefferson	Forward	F	7	DEN	Denver Nuggets	22	POR	Portland Trail Blazers	away	False	False	False	Moda Center	Portland	OR	False	2	one_day_rest	False	2	False	3	False	3	False	-1	False	True	True	-1	212.5	True	False	106.75	105.75	10	0	1	0	1	0	0	0	0	1	0	-4	0	82	99	-17	False	181	1	0	1	1	1.5	0	0	0.1	0.1	0	3	3	4.333333	4.333333	2.333333	2.333333	0.666667	0.0	3.0	0.333333	3.133333	4.333333	2.333333	0.666667	0.0		0.0	0.5	https://raw.githubusercontent.com/sportsdataverse/hoopR-nba-raw/main/nba/json/final/400974949.json	github.com, raw.githubusercontent.com	2026-06-11 17:31:07
400974991_1006	400974991	2018	2017-18	regular_season	2017-11-19 00:00:00	1006	Richard Jefferson	R. Jefferson	Forward	F	7	DEN	Denver Nuggets	13	LAL	Los Angeles Lakers	away	False	False	False	crypto.com Arena	Los Angeles	CA	True	2	one_day_rest	False	2	False	2	False	2	False	0	False	False	True	-5	222.5	True	False	113.75	108.75	5	0	1	0	0	0	0	0	0	1	0	-1	0	109	127	-18	False	236	1	0	1	1	1.5	0	0	0.2	0	0	4	4	7.0	5.75	2.333333	1.75	0.5	0.25	2.5	0.5	2.725	5.75	1.75	0.5	0.25		0.0	0.333333	https://raw.githubusercontent.com/sportsdataverse/hoopR-nba-raw/main/nba/json/final/400974991.json	github.com, raw.githubusercontent.com	2026-06-11 17:31:07

Access Requirements (Paid Dataset)

This dataset is behind manual gated access.

To obtain access:

Purchase the dataset here:
here
Provide your Hugging Face username at checkout.
Return to this Hugging Face page and click:
"Request Access"
Your access will be approved within 1-12 hours.

Once approved, you can use the Python snippet at the bottom of this README to load the dataset.

Dataset Structure

Total rows: 243,065

Total columns: 98

Splits

data: 243,065 rows

Data Files

data: data/data.parquet

Data Dictionary

The table below describes the columns included in this dataset.

column	pandas_dtype	dataset_type	description
record_id	object	string	Text column. Stable unique identifier for one player in one completed NBA game.
game_id	object	integer-like string	Stored as text, but values appear to represent whole numbers. SportsDataverse and ESPN-derived game identifier.
season_end_year	Int64	integer	Whole-number numeric column. Ending year of the NBA season, such as 2026 for the 2025-26 season.
season_label	object	string	Text column. Human-readable season label such as 2025-26.
season_type_label	object	string	Text column. Regular-season or playoff label for the game.
game_date	datetime64[ns]	datetime	Date or timestamp column. Calendar date of the completed game.
player_id	object	integer-like string	Stored as text, but values appear to represent whole numbers. Player identifier from the public source feed stored as text.
player_name	object	string	Text column. Player full display name.
player_short_name	object	string	Text column. Short display name from the source feed.
player_position	object	string	Text column. Position label from the source feed.
player_position_abbreviation	object	string	Text column. Position abbreviation from the source feed.
team_id	Int64	integer	Whole-number numeric column. Team identifier for the player row.
team_abbreviation	object	string	Text column. Team abbreviation for the player row.
team_display_name	object	string	Text column. Team display name for the player row.
opponent_team_id	Int64	integer	Whole-number numeric column. Opponent team identifier.
opponent_team_abbreviation	object	string	Text column. Opponent team abbreviation.
opponent_team_display_name	object	string	Text column. Opponent team display name.
home_away	object	string	Text column. Home or away label for the player row.
is_home	bool	boolean	True/false column. Boolean flag showing whether the player was on the home team.
starter_flag	bool	boolean	True/false column. Boolean flag showing whether the player started the game.
is_playoff_game	bool	boolean	True/false column. Boolean flag for playoff rows.
venue_full_name	object	string	Text column. Venue name from the public schedule feed.
venue_address_city	object	string	Text column. Venue city from the public schedule feed.
venue_address_state	object	string	Text column. Venue state or province from the public schedule feed.
venue_indoor	bool	boolean	True/false column. Boolean flag for indoor venue status.
days_since_prev_game	Int64	integer	Whole-number numeric column. Calendar-day gap between the team's previous completed game and this game.
rest_bucket	object	string	Text column. Bucketed rest label derived from days_since_prev_game.
is_back_to_back	bool	boolean	True/false column. Boolean flag for the second night of a back-to-back or zero-rest spot.
games_in_last_4_days_including_today	Int64	integer	Whole-number numeric column. Count of team games within the current 4-day window including this game.
is_third_game_in_four_nights	bool	boolean	True/false column. Boolean flag showing whether the team reached a three-games-in-four-nights spot.
games_in_last_6_days_including_today	Int64	integer	Whole-number numeric column. Count of team games within the current 6-day window including this game.
is_fourth_game_in_six_nights	bool	boolean	True/false column. Boolean flag showing whether the team reached a four-games-in-six-nights spot.
opponent_days_since_prev_game	Int64	integer	Whole-number numeric column. Calendar-day gap for the opponent team entering the same game.
opponent_is_back_to_back	bool	boolean	True/false column. Boolean flag showing whether the opponent was also on a back-to-back.
rest_advantage_days	Int64	integer	Whole-number numeric column. Team rest gap minus opponent rest gap.
is_rest_advantage	bool	boolean	True/false column. Boolean flag for positive rest_advantage_days.
is_rest_disadvantage	bool	boolean	True/false column. Boolean flag for negative rest_advantage_days.
has_archived_closing_line_context	bool	boolean	True/false column. Boolean flag showing whether archived spread and total context was available for the game.
closing_spread_line	float64	float	Decimal numeric column. Team-side archived closing spread line when available. Negative values indicate the player's team was favored.
closing_total_line	float64	float	Decimal numeric column. Archived closing game total line when available.
team_favorite_flag	bool	boolean	True/false column. Boolean flag for games where the player's team closed as the favorite when archived lines were available.
pickem_flag	bool	boolean	True/false column. Boolean flag for games with an effectively even archived spread line.
team_implied_total	float64	float	Decimal numeric column. Approximate team implied total derived from the archived spread and total lines when available.
opponent_implied_total	float64	float	Decimal numeric column. Approximate opponent implied total derived from the archived spread and total lines when available.
minutes	float64	float	Decimal numeric column. Minutes played by the player in the game.
field_goals_made	Int64	integer	Whole-number numeric column. Field goals made by the player.
field_goals_attempted	Int64	integer	Whole-number numeric column. Field goals attempted by the player.
three_point_field_goals_made	Int64	integer	Whole-number numeric column. Three-point field goals made by the player.
three_point_field_goals_attempted	Int64	integer	Whole-number numeric column. Three-point field goals attempted by the player.
free_throws_made	Int64	integer	Whole-number numeric column. Free throws made by the player.
free_throws_attempted	Int64	integer	Whole-number numeric column. Free throws attempted by the player.
offensive_rebounds	Int64	integer	Whole-number numeric column. Offensive rebounds by the player.
defensive_rebounds	Int64	integer	Whole-number numeric column. Defensive rebounds by the player.
rebounds	Int64	integer	Whole-number numeric column. Total rebounds by the player.
assists	Int64	integer	Whole-number numeric column. Assists by the player.
steals	Int64	integer	Whole-number numeric column. Steals by the player.
blocks	Int64	integer	Whole-number numeric column. Blocks by the player.
turnovers	Int64	integer	Whole-number numeric column. Turnovers by the player.
fouls	Int64	integer	Whole-number numeric column. Personal fouls by the player.
plus_minus	Int64	integer	Whole-number numeric column. Plus-minus from the public box score.
points	Int64	integer	Whole-number numeric column. Points scored by the player.
team_score	Int64	integer	Whole-number numeric column. Final points scored by the player's team.
opponent_team_score	Int64	integer	Whole-number numeric column. Final points scored by the opponent team.
team_margin	Int64	integer	Whole-number numeric column. Final point differential from the player-team perspective.
won_game	bool	boolean	True/false column. Boolean flag showing whether the player's team won.
total_points	Int64	integer	Whole-number numeric column. Combined final points in the game.
points_rebounds_assists	Int64	integer	Whole-number numeric column. Sum of player points, rebounds, and assists.
points_rebounds	Int64	integer	Whole-number numeric column. Sum of player points and rebounds.
points_assists	Int64	integer	Whole-number numeric column. Sum of player points and assists.
rebounds_assists	Int64	integer	Whole-number numeric column. Sum of player rebounds and assists.
stocks	Int64	integer	Whole-number numeric column. Sum of player steals and blocks.
simple_fantasy_points	float64	float	Decimal numeric column. Simple fantasy-style scoring proxy derived from points, rebounds, assists, steals, blocks, and turnovers.
points_per_minute	float64	float	Decimal numeric column. Points scored per minute played.
rebounds_per_minute	float64	float	Decimal numeric column. Rebounds per minute played.
assists_per_minute	float64	float	Decimal numeric column. Assists per minute played.
three_point_attempt_rate	float64	float	Decimal numeric column. Three-point attempts per minute played.
free_throw_attempt_rate	float64	float	Decimal numeric column. Free-throw attempts per minute played.
prior_games_played	Int64	integer	Whole-number numeric column. Count of prior recorded games for the same player before this row.
season_to_date_games	Int64	integer	Whole-number numeric column. Count of prior same-season games for the same player before this row.
rolling_minutes_last_3	float64	float	Decimal numeric column. Pregame rolling average of minutes across the previous 3 recorded games.
rolling_minutes_last_5	float64	float	Decimal numeric column. Pregame rolling average of minutes across the previous 5 recorded games.
rolling_points_last_3	float64	float	Decimal numeric column. Pregame rolling average of points across the previous 3 recorded games.
rolling_points_last_5	float64	float	Decimal numeric column. Pregame rolling average of points across the previous 5 recorded games.
rolling_rebounds_last_5	float64	float	Decimal numeric column. Pregame rolling average of rebounds across the previous 5 recorded games.
rolling_assists_last_5	float64	float	Decimal numeric column. Pregame rolling average of assists across the previous 5 recorded games.
rolling_pra_last_5	float64	float	Decimal numeric column. Pregame rolling average of PRA across the previous 5 recorded games.
rolling_three_pa_last_5	float64	float	Decimal numeric column. Pregame rolling average of three-point attempts across the previous 5 recorded games.
rolling_fantasy_points_last_5	float64	float	Decimal numeric column. Pregame rolling average of simple fantasy points across the previous 5 recorded games.
season_to_date_minutes_avg	float64	float	Decimal numeric column. Pregame same-season average minutes before this game.
season_to_date_points_avg	float64	float	Decimal numeric column. Pregame same-season average points before this game.
season_to_date_rebounds_avg	float64	float	Decimal numeric column. Pregame same-season average rebounds before this game.
season_to_date_assists_avg	float64	float	Decimal numeric column. Pregame same-season average assists before this game.
home_points_avg_before_game	float64	float	Decimal numeric column. Pregame average points in prior home games for the same player.
away_points_avg_before_game	float64	float	Decimal numeric column. Pregame average points in prior away games for the same player.
same_rest_bucket_points_avg_before_game	float64	float	Decimal numeric column. Pregame average points in prior games that shared the same rest_bucket for the same player.
source_url	object	string	Text column. Game-level JSON URL from the public SportsDataverse schedule source.
source_domain	object	string	Text column. Source domain summary for this dataset row.
last_collected_at	datetime64[us]	datetime	Date or timestamp column. UTC timestamp when this dataset build collected and transformed the row.

Intended Use

This dataset is intended for research, experimentation, analysis, and model prototyping.

Loading the Dataset


import os
from datasets import load_dataset

HUGGINGFACE_API_KEY_KARMANE = os.environ.get("HUGGINGFACE_API_KEY_KARMANE")

dataset = load_dataset(
    "Karmane/nba-back-to-back-player-trends-prop-research",
    token=HUGGINGFACE_API_KEY_KARMANE,
)

print(dataset)
print(dataset[list(dataset.keys())[0]][0])

# getting the DataFrame itself
# df = dataset[list(dataset.keys())[0]].to_pandas()

Karmane. (2025). NBA Back-to-Back Player Trends Dataset for Prop Research. Hugging Face. /datasets/Karmane/nba-back-to-back-player-trends-prop-research

Downloads last month: 20

Total file size:

20.8 MB