We’re Talking About Playoffs

It appears the Sixers are headed for the three seed in the upcoming playoffs. As the seeds become more locked in stone, the hot-takery regarding certain teams or individual play styles not “working” in the postseason increases. And yet...are the common tropes about the change in playoffs vs. regular season true or just oft repeated falsehoods? Let’s take a look and figure it out. I’ll try to answer the following questions for each category.

• Is there a difference between the regular season/playoffs?
• Is the different significant and/or meaningful?
• If there is a difference, does it align with conventional thinking?

All data will be from the 2014 through 2017 seasons, including both regular season and playoffs.

Pace

This one is relatively easy to figure out. We’ll look at overall pace by game number in the season.

Hmm, a little all over the place. Figure 2 will restrict the games to those played between playoff teams during the regular season as well.

The overall trend remains more or less the same, but the average pace is certainly slower. Something to keep in mind is that as we get out to game 100 or so, there are fewer teams playing, and therefore a smaller number of data points goes into computing the average. Table 1 breaks out the details below for the playoff team matchups only.

Difference: The playoffs are between 0.5 and 1.75 possessions slower on average than the regular season games, for the playoff team matchups only.

Significant/Meaningful: Aggregating the data across years does produce a statistically significant difference, but the lack of consistency or trends by year leaves me a smidge skeptical that the above 0.5-1.75 estimate is all that accurate.

Expected: Its repeated often enough that playoff games are substantially slower that a 0.5-1.75 possession difference does not seem that major.

Player Speed

Building off the pace discussion, there’s a way you could have the same number of possessions per game, but literally just be a slower game. I’ll be mean standardizing the speeds by individuals, so the graphs will show on average if speeds are trending up or down taking into account the normal speeds for each player. I’ll use a 15 minute cutoff as well. This is exclusively playoff team matchups.

Interesting. It certainly looks as if in generally the players literally move slower on average in the playoffs, and generally as the season goes as well. Probably a combination of fatigue and boredom during the regular season. 2016-17 sure was a postseason full of tryhards though.

Fouls per 48

I limited this to starters and those who played 15 min or more per game to remove any of those two fouls in 45 second type situations that can throw per minute data off.

What’s interesting to me is that there are more fouls per 48 at the start of the season, possibly due to rust or officiating emphasis (freedom of movement this year). In 2014 and 2017, fouls per 48 went up in the playoffs, and in 2015 and 2016, fouls per 48 went down. The details are presented below in Table 2.

Difference: None

Significant/Meaningful: None

Expected: Conventional wisdom is that it is harder to draw fouls in the playoffs as refs are less willing to whistle ticky tack calls. While obviously we cannot look at non-whistles that should have been calls, the actual number of fouls per 48 committed by starters remains essentially the same.

Minutes

Again, this one is pretty easy to determine, simply the by game average of the starters for each game over the course of a season.

Yep, this more or less squares with conventional wisdom. Here’s the table with specifics.

Difference: The playoff minutes are between 2.2 and 2.7 greater than the regular season minutes.

Significant/Meaningful: The difference is statistically significant and is approximately an eight percent increase.

Expected: Definitely aligns with what is commonly stated.

Rotation Size

We’ve already established that starters play more minutes, but do those minutes then get decreased amongst the rotation equally, or are the deep reserves totally eliminated?

This was not what I expected. Just to narrow down the scope a bit, let’s try again in Figure 7 with limiting this strictly to games between playoff teams in each given year, and for players who played more than five minutes.

That’s slightly more what I expected, with the table below for the more filtered data. You can certainly see that the bench shortens in terms of meaningful minutes when two good teams play each other, regardless of the time of year.

Difference: Based on the filtered numbers (limited to playoff team matchups and greater than five minutes on the court), the playoffs have between 0.4 and 0.7 fewer players on the court per game than in the regular season

Significant/Meaningful: The difference is statistically significant, but it’s hard to tell exactly how meaningful it is.

Expected: Agrees with conventional wisdom that the rotation shortens in the playoffs, but not as drastically as one might expect - if you change the minutes threshold to 15 minutes, the playoffs have between 0.7 and 0.9 fewer players per game.

TOUGHNESS! GRIT! EFFORT!

As we all know, tough, physical, gritty teams always pound the ball down low and get to the rack because jump shots are for soft children too afraid to get fouled. Actually this part will cover the distribution of points from the various areas on the floor. In order to get everything on the page, I’ll be plotting multiple metrics per graphs so that the seasons will have to be aggregated. Each line in this case includes all four seasons under consideration.

You can certainly see the Warriors effect here. As the number of teams/games decreases, the Warriors three point barrages pull up the green (threes) and push down the yellow (twos). Next, in Figure 9, we’ll look at a couple other breakdowns of scoring types.

To spare you from yet another time series graph, here’s all the scoring distribution data in one table, again aggregated by year and for playoff matchups only.

Now, this table is intentionally misleading (also unadjusted p-values). When we compare regular season data, there is an 82 game sample for each team. The Warriors and the Suns each provide 1/30th of the data. When we get to the playoffs, this is no longer the case - especially in the last few years. Check out Figure 10 below.

So, when we consider that playoff data from 2014-2017 as a whole is heavily skewed to the play styles of two specific teams, we can start to see how comparing regular season to playoffs kind of ends up as a “how similar is this to the Cavs/Warriors” test. Sidenote: don’t forget that in the 37th year of The Process, the Sixers have played in more playoff games than Charlotte, Brooklyn, Minnesota, Detroit, Atlanta, Phoenix, Sacramento, Orlando, the Knicks, and the Lakers.

Wrap-Up

Again, this was a pretty quick and dirty look into some of the common play type differences that I have heard assumed to be different in the regular season and postseason. There are numerous ways to examine/test for differences, and this was just one. If you have strong thoughts as to why I’m wrong, I am certainly willing to have my mind changed.