analytics_rules_doc.txt

2019 
NBA Hackathon Application –
 Basketball Analytics Prompt
Task
: Use the attached data and instructions to calculate offensive rating and defensive rating for each 
player in each game
 from the 2018 Playoffs
. 
Offensive Rating is defined as the team points scored per 100 possessions while the player is on the 
court. 
Defensive Rating is defined as the number of points per 100 possessions that the team allows 
while that individual player is on the court. A possession is ended by (1) made field goal attempts, (2) 
made final free throw attempt, (3) missed final free throw attempt that results in a defensive rebound, 
(4) missed field goal attempt that results in a defensive rebound, (5) turnover, or (6) end of time period.
Note: When a player is substituted before or during a set of free throws but was on the court at the time 
of the foul that caused the free throw, he is considered to be on the court for the free throws for the 
purposes of offensive and defensive rat
ing. A player substituted in before a free throw but after a foul is 
not considered to be on the court until after the conclusion of the free throws. 
Additionally, when there 
is a technical foul, the players marked on the court at the time of the technical foul are credited with any 
points resulting from the FTs.  This is similar to a normal foul but can be a bit confusing as technical fouls 
often happ
en in the midst of other foul events/FTs/substitutions.
Ex: In a 
Bucks
 vs. 
Nets
 game, Kyle Lowry commits a shooting foul on Kristaps Porzingis. Porzingis shoots 
and scores the first free throw, and then Jonas Valanciunas is substituted in for Serge Ibaka. Porzingis 
makes the second free throw. For the purposes of offensive and defensive ratings
, Ibaka (because he was 
on the court at the time of
 the foul) receives a value of 2 
toward his defensive rating
 for those free 
throws 
while
 Valanciunas is not considered on the court until after the free throws are completed. 
Porzingis receives a value of 2 toward his offensive rating. 
This folder includes three data sets: 
E
vent_Codes.txt, Play_by_Play.txt, 
and 
Game_Lineup.txt
. Please note that each question is permitted a maximum of two file attachments. 
Please submit your answer in a 
.csv
file and save your code, spreadsheets and all other work in a 
.zip
file.
Please submit a 
.csv
file 
titled “
Your_Team_Name_Q1_BBALL.csv
” substituting in the name of 
your team for "
Your_Team_Name
"
.  Please save as a 
.csv
. The final product should have 4 
columns. Column 1: 
Game_ID
, Column 2: 
Player_ID
, Column 3: 
OffRtg
, Column 4: 
DefRtg
.
Provided Data:
-
Event_Codes.txt 
o
This dataset provides look up values for the event message types and action types found 
in the play by play dataset. Each code is converted to an English language description of 
the event.
-
Game_Lineup.txt 
o
This 
dataset provides the start of period player availability.

Game_id
– a unique game code for each game

Period
(Quarter) 
– the associated period of the line up (overtime periods are 
indicated by values greater than 4)

Person_id
– a unique identifier for each 
player

Team_id
– a unique identifier for each team

Status
– a variable indicating whether the player is active (A) or inactive (I)
-
Play_by_Play.txt 
o
This dataset provides play by play information on the event level for each game.
o
To properly sort the events in a game, use the following sequence
 of sorted columns: 
Period
(ascending), 
PC_Time
(descending), 
WC_Time
(ascending), 
Event_Num
(ascending)

Event_Num
– an ordered counter for each event in a game. Note, this number 
may not be perfectly sequential so please use the sorting methodology outlined 
above

Event_Msg_Type, Action_Type
– coded descriptions of what happened 
during the event (see the 
Event_Codes.txt
dataset to see the codes)

WC_Time
– the in-arena time of the event in Unix format. It is coded as tenths 
of a second

PC_Time
– the time on the game clock in tenths of a second (e.g. 7200 
corresponds to 720 seconds/12 minutes remaining in the quarter)

Option1
– on a shot attempt, this column will tell you the point value of the 
shot
•
On free throw attempts, if the value in this column is 1, it means it was a 
made free throw, otherwise, it was missed 

Person1, Person2
– the 
Person_id’s
of the players who are
 directly 
associated with the event (e.g. if the event is an assisted made basket, 
Person1
is the shot maker and 
Person2
is the player who assisted)
•
In the case of a substitution, the 
Event_Msg_Type
will be 8, 
Person1
will be the 
Person_id
for the player leaving the game, and 
Person2
will be the 
Person_id
for the player entering the game

Team_id
– in most scenarios, this is the 
Team_id
associated with the 
Person1
column. However, there are instances when this is not the case. To 
accurately and consistently identify a player’s team, we suggest merging with 
the 
Game_Lineup
dataset on the 
Person1
and 
Person2
columns.