Adjusted Point Differerential by Schedule

by TheOdenator on Nov 27, 2009 5:07 PM PST

So one thing has always bothered me about Hollinger's rating system, which is that he basis schedule strength on wins and losses, while he uses point differential for other aspects of determining the strength of the team. As point differential is a stronger stat with respect to predicting future winning, I took the schedule of each team, and created a strength of schedule based on the respective opponents point differential.

Team	Point Differential	Expected Point Differential	Difference from hypothetical ave team
Atlanta Hawks	5	-1.54	6.54
Boston Celtics	8.4	2.166666667	6.233333333
Dallas Mavericks	6.8	1.046666667	5.753333333
Portland Trail Blazers	8.3	2.841176471	5.458823529
Los Angeles Lakers	5.3	-0.142857143	5.442857143
Orlando Magic	6	0.5875	5.4125
San Antonio Spurs	4.2	-0.646153846	4.846153846
Denver Nuggets	7.5	3.446666667	4.053333333
Phoenix Suns	5.5	1.653333333	3.846666667
Oklahoma City Thunder	2.3	-1.326666667	3.626666667
Houston Rockets	0.8	-1.986666667	2.786666667
Cleveland Cavaliers	4.5	2.34	2.16
Utah Jazz	1.2	-0.64	1.84
Miami Heat	1.5	0.75	0.75
Milwaukee Bucks	3.3	3	0.3
Detroit Pistons	-2.7	-2.02	-0.68
New Orleans Hornets	-3.3	-2.4625	-0.8375
Charlotte Bobcats	-1.4	-0.464285714	-0.935714286
Golden State Warriors	-2.2	-0.614285714	-1.585714286
Toronto Raptors	-2.8	-1.19375	-1.60625
Indiana Pacers	-1	1.023076923	-2.023076923
Sacramento Kings	-1.7	0.478571429	-2.178571429
Chicago Bulls	-6.5	-2.964285714	-3.535714286
Washington Wizards	-5	-1.338461538	-3.661538462
Los Angeles Clippers	-3.8	1.28125	-5.08125
Philadelphia 76ers	-4.3	0.786666667	-5.086666667
Memphis Grizzlies	-5.4	0.42	-5.82
New York Knickerbockers	-7.3	-0.26	-7.04
New Jersey Nets	-10.4	-0.746666667	-9.653333333
Minnesota Timberwolves	-14	-2.406666667	-11.59333333

To explain the headings point differential is the difference between the points the team has scored and the points other teams have scored against them. The Expected point differential is the average point differential for every team they have played thus far. An example would be so an average team on X's schedule would have beaten them by 4, so the expectation is a 4 point win, and so if they won by 5, after one game their 'Difference from the Hyopthetical Average Team' would be 1.

If a team has just played difficult competition that everyone else has also been losing to, then their expected point differential would be low, while if the team has had a creampuff schedule the expected point differential would be high. In this based upon point differential the Nuggets have played the weakest schedule so far, the Bucks the second weakest, and the Blazers the third, while the Bulls and Hornets have played the two roughest schedules to date.

While I do believe that this data does provide valuable insights and that I would stand by this as a ranking of the teams as to how they have played throughout the season up to now, this method does have its flaws. The most obvious being that teams that have played a creampuff schedule will have an artificially high Expected Point Differential, which will mess with the average for the couple of decent teams they have played. If anyone can think of a way to deal with this I can send you the data that I used (basketball-reference.com is where I initially downloaded it), or just feel free to take this yourself. The second most obvious being it doesn't take into account home-away or pace, as the previous stat fan-shot explained in a game with a slow pace a ten point lead can be more imperssive than a 20 point one in a Golden State - Knicks game.

Let me know what you guys think, and if anybody else is interested in combining some of these "advanced" stats for a BE power ranking.

Tweet Comment 6 comments | 2 recs |

Do you like this story?

Comments

Nicely done.

I’ve been waiting for someone to do something like this, and by the end you had already addressed the things I was going to mention. It almost makes me wonder if it would make sense to do some sort of recursive thing, where you use adjusted point differential to determine adjusted point differential.

You can measure skill and talent with your eyes, but productivity is shown through statistics.

by austinpwnz on Nov 27, 2009 5:47 PM PST reply actions

I think this is close to the best you can do for a simple metric

I might not include pace, as I just talked about here.

Here’s all the possible enhancements I can think of:
- account for home/away, like you mentioned
- don’t count games against the original team when evaluating an opposing team’s SOS
- go beyond using just opponent’s point differential, and use opponent’s opponents’ differential, or beyond. Take it to infinity.

Disclaimer: everything I know about basketball I learned on Blazersedge.

by pualo on Nov 27, 2009 5:50 PM PST reply actions

Nice!…and rec. I think this is definitely heading down the right path.

Don’t know if anyone else has this problem, but some of your numbers in the right column are chopped by the sidebar. Looks like you sorted by that column so it’s not to tough to gauge about where the numbers should be.

Romance me with that Roy rainbow shot which took flight from way beyond the arc and sailed so high that before it came back down to earth sealing the victory, it kissed the rafters and said "You're mine baby."

by Blazer1342 on Nov 27, 2009 6:37 PM PST reply actions