A Quarter at a Time, Part I

Jun 16, 2015; Cleveland, OH, USA; Golden State Warriors head coach Steve Kerr reacts prior to game six of the NBA Finals against the Cleveland Cavaliers at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports
Jun 16, 2015; Cleveland, OH, USA; Golden State Warriors head coach Steve Kerr reacts prior to game six of the NBA Finals against the Cleveland Cavaliers at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports /
facebooktwitterreddit
Jun 16, 2015; Cleveland, OH, USA; Golden State Warriors head coach Steve Kerr reacts prior to game six of the NBA Finals against the Cleveland Cavaliers at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports
Jun 16, 2015; Cleveland, OH, USA; Golden State Warriors head coach Steve Kerr reacts prior to game six of the NBA Finals against the Cleveland Cavaliers at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports /

Things started out harmlessly. I wanted to do something completely different and asked Darryl (aka. Scrapemaster General) for game-by-game scoring and possession data. But of course, Scrapemaster General Darryl does not only have game-by-game scoring and possession data, he has the whole dataset parsed by quarter. So, I started to look at this information about points and number of possessions looking for patterns. And patterns there are… Part I will cover the most obvious, for which I did not need to further analyze the data. Days of rest for example is surely an interesting topic, but would force me to do additional calculations. For now, I have data for the last two seasons, which implies that the patterns I see are repeatable

1. The Game “Slows Down”

Pace differs significantly between the first and last quarter. On average, first quarters have at least 0.6 possessions per team more than fourth quarters. And the number of possessions decreases very continuously, the second quarter being slower than the first and so on. It is interesting to see that fourth quarters have a slightly higher variance or standard deviation than the three other quarters. The most likely explanation is that close games tend to have much more possessions during the last minutes, with teams that are behind stopping the clock and rushing shots, while games that are out of hand are probably slower, because the leading team is milking the shot clock and the loosing team admits defeat. (Note: this is one of those assumptions that I’ll need to proof…)

possession14-15
possession14-15 /
possession13-14
possession13-14 /

2. Higher Pace Means More Sluggishness

The general finding for pace is that teams with a fast offensive pace tend to score more points per possession. Also, shots that happen earlier in the shot clock tend to yield more points. The problem is that you usually achieve shots early in the shot clock through opponent live ball turnovers or misses. So, as a sum fast possessions lead to lower points per possessions (PPP), as one team remains empty handed. This can be seen when we compare PPP by quarter with the respective number of possessions.

'Sluggishness' influences the away team (blue) more than the home team (red). Lines are linear fits, with indicated confidence intervals. Only one fifth of all quarters are plotted and dots are jittered to give at least an impression of the two distributions
‘Sluggishness’ influences the away team (blue) more than the home team (red). Lines are linear fits, with indicated confidence intervals. Only one fifth of all quarters are plotted and dots are jittered to give at least an impression of the two distributions /
'Sluggishness' influences the away team (blue) more than the home team (red). Lines are linear fits, with indicated confidence intervals. Only one fifth of all quarters are plotted and dots are jittered to give at least an impression of the two distributions
‘Sluggishness’ influences the away team (blue) more than the home team (red). Lines are linear fits, with indicated confidence intervals. Only one fifth of all quarters are plotted and dots are jittered to give at least an impression of the two distributions /

My first impression was this: “The interesting thing is, that this effect has a much higher influence for the away team than for the home team. While both teams score on average at a similar rate during slow quarters with few possessions, the away team tends to score worse during presumably more hectic passages of the game.” But…

3. A Frenzied Start Leads to an Uphill Battle for Away Teams?

… the results come with a caveat. Previous research shows that it is common for the trailing team to have a bit of a comeback. So, this effect is of course hidden somewhere in this data. Looking only at the first quarter and the score tied, we find that the advantage of the home team is regardless of the  pace.

Focus on the 1st quarter shows that home team advantage is not actually related to pace
Focus on the 1st quarter shows that home team advantage is not actually related to pace /
Focus on the 1st quarter shows that home team advantage is not actually related to pace
Focus on the 1st quarter shows that home team advantage is not actually related to pace /

In the following figures, you can see the average score for home and away team per quarters. While the away team looses the first quarter on average by a point, it only loses the second and third quarter by half a point and is almost even for the last quarter. Of course by then the rules of garbage time tend to change a lot of the mechanics.

Points14-15
Points14-15 /
Points13-14
Points13-14 /