APBRmetrics

This is a question aimed for those of you who derive lineups off of play-by-play data.

Take the simple example where you have a sequence of events that look like this:

5:06 Kobe Bryant fouls Tony Parker on 2 point field goal attempt
5:06 Tony Parker hits 1st free throw
5:06 Tim Duncan subbed in for Kawhi Leonard
5:06 Nick Young subbed in for Kobe Bryant
5:06 Tony Parker hits 2nd free throw

Notice that the game clock is at 5:06 for all of those events. The plus/minus for each player is as follows:

Kobe Bryant -1
Tony Parker +2
Tim Duncan +1
Kawhi Leonard +1
Nick Young -1

However, shouldn't it look like the following (because Kobe Bryant, Kawhi Leonard, and Tony Parker were all on the court when the foul that led to the two free throw attempts was committed)?:

Kobe Bryant -2
Tony Parker +2
Tim Duncan 0
Kawhi Leonard +2
Nick Young 0

Has anyone ever accounted for this (perhaps by putting substitutions as the last events) or is the overall impact marginal?

1andDunn wrote: Has anyone ever accounted for this (perhaps by putting substitutions as the last events) or is the overall impact marginal?

As far as I know everyone is accounting for this.

How would this be done?

It's easy in this scenario to re-order the events so that the substitutions come after the free throws and the plus/minus works out:

5:06 Kobe Bryant fouls Tony Parker on 2 point field goal attempt
5:06 Tony Parker hits 1st free throw
5:06 Tony Parker hits 2nd free throw
5:06 Tim Duncan subbed in for Kawhi Leonard
5:06 Nick Young subbed in for Kobe Bryant

However, I can think of scenarios where re-ordering the events so that the substitutions come last can hurt you. For example, this scenario:

5:06 Timeout Lakers
5:06 Manu Ginobli subbed in for Lamarcus Aldridge
5:06 Kobe Bryant fouls Tony Parker (Spurs in the bonus)
5:06 Tony Parker hits 1st free throw
5:06 Tim Duncan subbed in for Kawhi Leonard
5:06 Nick Young subbed in for Kobe Bryant
5:06 Tony Parker hits 2nd free throw

Is there a simple way to handle both these cases at once?

A very simple hack is to subtract 0.1 seconds from the stated time of all substitutions in the PbP

Most people should be accounting for this. Even the official +/- from the NBA does.

In pbp data, I do it in the script deciding who's on the floor. There are a ton of weird nuances for these situations (for example, players subbed in during the possession but before the foul), but you can code it so the lineups don't change with new subs coming in until after the last free throw.