What Wins Championships? (598 views)

[eric]

In this post I will define ORtg+ and DRtg+ as the team's ORtg and DRtg relative to the league mean that year. A positive number is always better: a team with 3 ORtg+ had an ORtg 3 points better than the league average, a team with 3 DRtg+ had a DRtg 3 points better than the league average.


Year ORtg+ DRtg+ Pace+ Team

3000 3.62 3.13 1.15 Jazz

3001 8.39 0.44 5.30 Pistons

3002 11.97 5.22 3.77 Jazz

3003 9.07 7.45 -4.13 Mavericks

3004 9.86 8.70 2.85 Grizzlies

3005 11.85 7.52 2.05 Mavericks

3006 8.26 4.19 4.01 Kings

3007 6.61 3.74 0.41 Cavaliers

3008 9.46 9.19 0.20 Magic

3009 4.74 4.68 0.24 Grizzlies

3010 11.03 5.12 -1.09 Rockets

3011 7.97 6.99 0.19 Rockets

xxxx 8.57 5.53 1.25 Average



Offense and defense are both important, I assume this comes as no surprise. Pace is also important, which might. Consider last year: the Bulls and Rockets were the cream of the league with 15 Net Rtg+, but the Rockets reached the Finals and the Bulls did not. Why? When the Bulls and Cavs played, the average pace+ was -.82; when the Rockets and Mavs played, the average pace+ was -.03. Coincidence, or does slow pace increase the odds of the better team losing? Clearly in the extreme case where each team takes one shot, it is very easy for the better team to lose, and in the opposite extreme of infinite shots the better team will always win. The question is if there are other pace-related phenomena (fouling? better/worse shots?) that muddy these effects.

Tomorrow I will tabulate the playoff results, I anticipate finding a linear function of NetRtg and Pace to explain Success.

[Odin]

very fast/always/always/outside for life

[RWisoursavior]

I have no fucking clue.

[ANK1990]

We need to get this guy a spreadsheet.

[20s Navidad]

Scoring more points than your opposition.

[Swaggy Z]

Eric..... Slow down, you're confusing me.

[eric]

If I'm right, then we should observe that faster paced matchups are less likely to produce upsets. We have 12 years * 15 series per year = 180 data points, but if we take all the series by average Pace+ and bin them into intervals of one, we in fact find...


Pace+ Wins Losses Pct
-5 3 1 75.0%
-4 4 2 66.7%
-3 7 1 87.5%
-2 12 4 75.0%
-1 25 5 83.3%
0 22 5 81.5%
1 24 10 70.6%
2 19 7 73.1%
3 12 7 63.2%
4 4 5 44.4%
5 0 1 0.0%


...which regresses to Favored Win Pct = -.05 * Pace + k, or that SLOWER paced matchups are less likely to produce upsets. This qualitative finding holds even if we restrict ourselves to those bins with 10+ members. Although the coefficient is reduced to -.03, the R^2 increases to .5325.

.

But not all favorites are as favored. We are equating the narrowest upset with the most shocking, let us instead sum the Net Rtg+ of all outcomes. If there are for example three series at 4 Pace+, we get the heavy favorite right and two slight favorites wrong, we will get a positive number. If instead we miss the heavy favorite and get the two slight favorites right, we will get a negative number...


Pace+ Count Avg
-5 4 2.17
-4 6 3.81
-3 8 3.41
-2 16 3.96
-1 30 3.72
0 27 3.73
1 34 1.72
2 26 3.05
3 19 0.99
4 9 0.67
5 1 -0.46

...again, we get the opposite of what we expected. Pace+ is still negatively correlated with reaching the expected outcome. I believe the explanation is that while some teams are slower than others, every team in this league plays at an extremely high pace. The lowest last year was the Raptors' 107.4, which if it took place in the NBA would be the fastest pace in 25 years... and they're the TORTOISE of this league. The binomial distributions of each team are thus about as separate as they are going to get, but perhaps pace can impact other statistical accumulation, specifically in the form of personal fouls. Especially in a league where robot coaches have significant autonomy, it stands to reason that even if team A has better starters than team B, both teams getting in foul trouble will instead pit backups or end of bench guys against each other where team A's advantage is reduced.

The next step is to go through the box scores of the playoffs to see if this is actually what happened, to see if higher paced teams (a) played at a higher pace and (b) suffered more fouls as a result.

.

The injury records are insufficient for this research, but it is interesting to wonder if the software assigns injuries at a per-possession rate, which would also disfavor higher paced teams.

[Souper Troopers]

I'll give you a 5 to stop.

[Heebs]

Awesome work.

[IanBoyd]

fuck all this spreadsheet mumbo yumbo

[eric]

Thank you all :)

[Lazy Pete]

Those raptors are fast now, like, um, raptors