SOS in the NBA?

[eagereyez]

So I was looking at the difference in SOS between teams in the NBA:
http://espn.go.com/nba/stats/rpi/_/sort/SOS

And comparing it to SOS in the NFL:
http://espn.go.com/blog/nflnation/post/ ... f-schedule

So you have a largest difference of .03 in SOS for the NBA, and a massive largest difference of .15 in the NFL. Does this mean that the difference in quality opponents between say SAS and ATL is negligible? Or is a small difference of .03 actually significant?

[giberish]

If you just looked at 16 games stretches in the NBA, there would be much larger schedule strength differences.

In the NBA teams play a full double round-robin schedule (giving a very balanced base), then add 1-2 extra games against the other teams in their conference.

NFL teams play 3 teams twice, 10 teams once, and 18 teams not at all.

[Chicago76]

In addition to sample size, the difference is also based upon how the game is scored and how margins open up. Possessions in basketball involve teams moving the ball around the court through a succession of fairly low risk passing and dribbling actions. Teams execute a simple entry pass very well. The difference between a good team and a bad team might be one point every 20 possessions. In football, possessions are drives. Drives require plays that pick up first downs. The difference between a good team and a bad team per play might be 5% per play in football rather than per possession in basketball.

Simple illustration: assume it takes 10 plays for everyone to get a TD. Team A executes flawlessly every on each play and always scores a TD. Team B executes at 95% of team A and with each sub perfect execution, their ability to score a TD declines by 5%. Team B will only score a TD 60% of the time (.95^10). They are losing 2.8 pts to Team A on every drive. Over 10 drives that's a huge margin compared to 100 possessions in basketball (relative to the good team).

The other issue is that because drives (possessions) are so scarce in FB compared to basketball, strategy changes more quickly. If you fall behind by 3 scores early in FB, teams play O and defend differently. A higher risk strategy is required to try to get back in the game, because it doesn't matter if you lose by 100 or by 1. Defenses also know the pass is coming, so they blitz more. The O is more predictable. This can start in the 3Q in football. In basketball, it typically only happens with 2-3 minutes left. In football, games can open up more and runaway score lines can occur. A 28-14 game might not feel like a runaway game, but that's the same thing in relative terms as a 100-50 basketball game.

Unbalanced schedule + more team variance (team scoring vs. league avg scoring) much bigger SOS #s in football.

[eagereyez]

In addition to sample size, the difference is also based upon how the game is scored and how margins open up. Possessions in basketball involve teams moving the ball around the court through a succession of fairly low risk passing and dribbling actions. Teams execute a simple entry pass very well. The difference between a good team and a bad team might be one point every 20 possessions. In football, possessions are drives. Drives require plays that pick up first downs. The difference between a good team and a bad team per play might be 5% per play in football rather than per possession in basketball.

Simple illustration: assume it takes 10 plays for everyone to get a TD. Team A executes flawlessly every on each play and always scores a TD. Team B executes at 95% of team A and with each sub perfect execution, their ability to score a TD declines by 5%. Team B will only score a TD 60% of the time (.95^10). They are losing 2.8 pts to Team A on every drive. Over 10 drives that's a huge margin compared to 100 possessions in basketball (relative to the good team).

The other issue is that because drives (possessions) are so scarce in FB compared to basketball, strategy changes more quickly. If you fall behind by 3 scores early in FB, teams play O and defend differently. A higher risk strategy is required to try to get back in the game, because it doesn't matter if you lose by 100 or by 1. Defenses also know the pass is coming, so they blitz more. The O is more predictable. This can start in the 3Q in football. In basketball, it typically only happens with 2-3 minutes left. In football, games can open up more and runaway score lines can occur. A 28-14 game might not feel like a runaway game, but that's the same thing in relative terms as a 100-50 basketball game.

Unbalanced schedule + more team variance (team scoring vs. league avg scoring) much bigger SOS #s in football.

So does this lower SOS in basketball mean that records are less dependent on a team's schedule? I'm asking this with the eastern conference in mind, wondering how much playing in that conference will boost a team's record.

[Chicago76]

So does this lower SOS in basketball mean that records are less dependent on a team's schedule? I'm asking this with the eastern conference in mind, wondering how much playing in that conference will boost a team's record.

Compared to NFL, SOS has far less significance in the NBA. If there is a 2 pt/100 possession difference in SOS, that's the difference between winning 41 games and winning 46-47 games. This assumes that the strength of teams is fairly randomly distributed among teams within a conference. If you have some bunching of teams at a certain quality level, the results might be magnified a bit. There might be only a couple of good teams in both conferences that an average team would lose to regardless. There might also be 3 bad teams in each conference that an average team will win. If the middle of the pack is what is pulling the SoS in one conference up, then the win impact might be magnified more.

If you go back to 79/80 through last year, you can take SRS and compare to expected wins. If teams underperform in the win column relative to their SRS score, it probably indicates a combination of two things: 1) difficult SoS and 2) getting unlucky (many close losses/big wins). Just looking at all exactly average to only good (not very good teams)...SRS between 0.00 and 4.00. That's a little over 280 teams. 1 had a win total 10.1 below SRS expectations (MIN last year). The next 5: 9.7, 8.1, 8.1, 7.2, 7.2 below. Everyone else is within +/- 7 wins. Only two of those outliers came from a notably stronger conference (MIN and 12-13 HOU) and all of these outliers are probably down to misfortune as much or more than SoS. The max SoS Win impact in practice is probably 5-6 wins. In theory, it might be 7-8.

In the NFL, there are teams in weak divisions with 10-6 records that finished last in their division the year before. These teams get to play 6 games v. their watered down division, 4 more games vs. another conference division (maybe weak), 4 games vs. a division in the other conference (maybe weak) and 2 games vs. the two other teams that finished last in their divisions in the same conference the year before (often weak). If everything sways the right way, a team can get a really soft schedule. This favorable scheduling for bad teams is why teams can yo-yo between 4 win and 9-10 win seasons. That 10-6 teams playing a soft schedule could easily be a 6-10 win team with a different schedule...maybe even worse. That is the equivalent of a 20 win swing in the NBA due to SoS.