NBA: Playoffs, How Do They Work?

[Latrinsorm]

Fact: NBA playoffs only involve the top 8 (out of 15) teams from each conference by winning %.
Fact: NBA playoff games are played at a slower pace (possessions per game) than regular season games.
Fact: NBA playoff games are decided by about as many points as regular season games.

This does not. Make. Sense. Better competition should drive the margin down. Slower pace should drive the margin down. The margin stays the same. Hence, something must change in the playoffs, but what?

There are Four Factors to the NBA:
-shooting (in the form of effective field goal %, a measure that adjusts for the 3 point shot)
-turnovers (in the form of turnovers per play)
-rebounding (in the form of % of rebounds that are offensive for offense and defensive for defense)
-fouling (in the form of... water! no! in the form of free throws made per field goals attempted)

If we look at the past five years of Finals participants (on the theory that they will have played the most playoff games against the most varied competition in any given year), we find that their values change on a percentage basis as...


efg -2.71
tov -6.75
orb 1.27
ft/fga 6.60
efg -0.46
tov -0.70
drb -0.34
ft/fga 15.82
...that is, during the playoffs a team's offense...

-shoots worse
-turns the ball over much less
-rebounds a bit better
-makes more foul shots

...and a team's defense...

-forces slightly worse shooting
-forces slightly less turnovers
-rebounds slightly less (gets back in transition?)
-fouls A HELL of a lot more

.

Now, a brief digression into causality. No matter what correlation we find with the four factors, do we conclude that a team got to the Finals because it played better or a team played better because it got to the Finals? That is to say, if a particular team hadn't played better it wouldn't have gotten to the Finals, and some other team that did would have: the Wyatt Earp effect.

Of the four factors, I think it's fair to divide them as follows:

Physical effort
Contesting shots
Rebounding (both sides)
Drawing fouls
Forcing turnovers

Mental effort / focus
Making shots
Not fouling
Avoiding turnovers

Thus, depending on the net effects we might be able to see which areas (if any) top tier teams are slacking off in during the regular season. (Though there are potential obfuscating factors: officiating especially with regards to home court advantage, coaching and gameplanning and buying into gameplans, David Stern.)

.

In net (offense-defense), a Finals team will see the following changes from its regular season numbers:



efg -2.25
tov -6.06
rb 0.93
ft/fga -9.23


The biggest difference, by far, is in fouling... but it is surprisingly in the wrong direction, as is shooting. Rebounding gets slightly better, and while I'm not sure how to calculate random error for this sample I'm pretty sure it's bigger than 0.93%. The other big winner is turnovers, which like lower DRB% might be as much an intentional strategy: don't gamble.

In any event, I've caught the scent now and will continue to amass data.

[Latrinsorm]

Over 20 years, the average net differences are:
2.76% worse shooting
2.19% better turnovers
1.24% worse rebounding
7.23% worse fouling

Which begs the question, how the heck are these teams even winning, let alone winning by the same margins as before? Then it occurred to me, what if the pace doesn't slow down in the playoffs? It's definitely true that the playoff pace is slower than the regular season pace, but not all regular season teams make the playoffs. What if playoff teams are in general those that play at a slower pace in the regular season?

And as it turns out, they are! But to a tiny, tiny, tiny degree. The average regular season pace over the last 20 seasons is 91.54, the average regular season pace of teams that made the playoffs is 91.25, the average postseason pace is 88.53. (Also potentially interesting are yearly trends: playoff teams were slower than average from 1994-2001, faster from 2002-2007, slower from 2008-2012, and faster in 2013. The postseason has never been played as fast as either set of the regular season.) Clearly, this doesn't help us.

Alright, what if the effects have different relative values? That is, what if I can come out ahead by getting 1% better at my giveaway rate at the cost of 1% worse on drawing fouls? Or by getting 1% better at takeaways at the cost of 1% worse on fouling? Let us consider:

TOV%
-A turnover means 0 points scored on 1 possession. This is a loss of about a point.
-Turnover rates are about 13%, so a 1% change in turnovers is about .13%.
-Thus, per possession a 1% change in turnover rate is worth about 1 * .0013 = .0013 points.
-A turnover possibly results in a more efficient possession for the other time (transition), but for now we'll stick with this.

eFG%
-FGAs occur on about 75% of possessions.
-Twice eFG% is points generated on baskets per FGA.
-The average eFG% is about 49%.
-Thus, per possession a 1% change in eFG% is worth about .0049 * 2 * .75 = .00735 points.

This suggests that a change of equal relative magnitude in eFG% is much more important than one in TOV%, and given that they are approximately equal this should in turn mean the teams we are looking at perform worse, but the fact remains that they perform just as well (margin of victory) against better competition.

.

This in turn suggests that while teams that reach the Finals play worse than they did in the regular season, the teams they play against play much worse. Why might this be?

-It is easier to find flaws in an existing product than inventing a new one. In this case, it is easier for an opposing coach to find (and exploit) the weaknesses in your plan than for you to invent (and implement) a fundamentally new one, even if you are a much better coach. Thus both teams do worse, though presumably the teams with better coaches would do less worse and thus be ahead.

-The season is long. Everyone gets tired, everyone gets hurt. The best teams do flip the switch and get better, and while this is effect is overwhelmed by the fatigue effect, it still surpasses those teams that are just as fatigued but with no switch to flip.

-Home court advantage. Even though a team at absolute most can get 4 net games of home court advantage (1 each series), 4 out of 20 is a pretty big proportion. Fatigue still dominates, but against fatigue is a bilateral effect and home court is unilateral.

[Gelston]

They work by adjusting the game sliders so that you have an advantage over the NPC team.