web analytics

NBA Adjusted Efficiencies 1-11-2011

January 11, 2011

Last week, I unveiled the first iteration of my adjusted efficiencies Google Motion Chart. This week, I’ll revise and expand on it.

The main thing I didn’t like about the chart was the 5-game moving averages. The games dropping off the far end of the moving average add just as much movement as the newest game adds. The solution? Use a weighted moving average. The scheme I have selected is a 7-game weighted moving average, which weights the most recent game 7, the next 6, and so on to the 7th-most recent 1 and the 8th 0.

This has 2 effects: the movement is more smooth, and the most recent game has a bigger impact. Here’s the latest chart, again with all efficiencies adjusted by opponent, rest advantage, and location. No, I did not locally adjust the opponent adjustment; I feel using a full season-worth of data is more valuable than local precision.

This week’s featured team is Orlando.  Orlando has been on quite the hot streak since making the blockbuster trades.  I questioned the trades at first, but they seem to be working out famously.  I’ll probably do an in-depth analysis on the results of the trade, but first I’ll simply show how impressive this hot streak has been.

Team Chart Orlando 1-10-11

Orlando's performance through the season

That’s 9 games in a row of above-NBA-average play. Most of the improvement seems related to Hedo Turkoglu, who is a different player in Orlando. (Can a statistician say that? Should I just say “small sample size?”)

Here’s a table of Orlando’s games this season and their offensive and defensive numbers. Remember, negative is good for defense:

GmGameAdjusted OEAdjusted DEAdjusted MarginResult
110/28/10 vs. WAS4.0-15.419.4W
210/29/10 @ MIA-21.7-5.5-16.2L
311/03/10 vs. MIN16.1-17.233.2W
411/05/10 vs. NJN9.62.17.5W
511/06/10 @ CHA-5.9-8.42.5W
611/08/10 vs. ATL-6.8-9.62.9W
711/10/10 vs. UTA-7.86.5-14.3L
811/12/10 vs. TOR-0.611.5-12.1L
911/13/10 @ NJN1.00.20.8W
1011/15/10 vs. MEM-3.0-23.019.9W
1111/18/10 vs. PHO1.2-7.78.9W
1211/20/10 @ IND-2.9-8.25.3W
1311/22/10 @ SAS0.12.8-2.6L
1411/24/10 vs. MIA14.8-1.916.7W
1511/26/10 vs. CLE6.310.5-4.2W
1611/27/10 @ WAS8.06.01.9W
1711/30/10 vs. DET-3.7-9.86.1W
1812/01/10 @ CHI27.4-15.943.3W
1912/03/10 @ DET13.70.313.4W
2012/04/10 @ MIL-7.1-1.2-5.8L
2112/06/10 vs. ATL-24.3-17.1-7.2L
2212/09/10 @ POR-11.7-2.8-8.9L
2312/10/10 @ UTA9.612.7-3.1L
2412/12/10 @ LAC0.2-9.59.7W
2512/14/10 @ DEN-4.39.6-13.9L
2612/18/10 vs. PHI-14.8-0.3-14.4L
2712/20/10 @ ATL-19.4-8.8-10.5L
2812/21/10 vs. DAL1.56.0-4.5L
2912/23/10 vs. SAS20.2-5.125.3W
3012/25/10 vs. BOS-8.7-20.511.8W
3112/27/10 @ NJN7.1-5.412.5W
3212/28/10 @ CLE5.4-7.412.8W
3312/30/10 vs. NYK8.5-2.410.9W
3401/03/11 vs. GSW1.2-12.914.1W
3501/05/11 vs. MIL1.5-1.73.2W
3601/07/11 vs. HOU5.7-7.012.7W
3701/08/11 @ DAL24.72.122.6W

Tags: , , ,

3 Responses to NBA Adjusted Efficiencies 1-11-2011

  1. Crow on January 11, 2011 at 3:19 pm

    A 7-game weighted moving average which dropped weight half as sharply as the new system would appeal to me more as the middle ground between the previous version and the new one. It would weight the most recent game 7, the next 6.5, and so on to the 7th-most recent 4 and the 8th 0. Weighting the most recent game at 7 times the game about to drop off seems like a really high emphasis. The proposal I suggest for your consideration would weight the most recent game 75% more than the the game about to drop off. It that isn’t enough for your tastes you could descend to 3 instead of 4 or even 2.5 or 2 instead of all the way to 1.

  2. Greyberger on January 11, 2011 at 4:31 pm

    Thanks for putting these together. I think the Utah motion chart is also very strange.

    Other events that seem to show up on the chart: Phoenix feeling the effect of their side of the trade, which was to lose whatever mojo they had. Denver suffering during the trade rumors.

    I had seen the numbers but hadn’t quite realized how much Chicago has become the defensive team of the league. When they’re playing well it’s defense defense defense for Tibs’ squad.

  3. Jerry on January 11, 2011 at 5:04 pm

    RAPM had Turkoglu as the best offensive player in the league before the trade. I guess VanGundys’ defensive schemes, or DHoward, are what makes Hedo not look super terrible on defense like he did in Phoenix/Toronto

Leave a Reply

Your email address will not be published. Required fields are marked *

Current day month ye@r *

DSMok1 on Twitter

To-Do List

  1. Salary and contract value discussions and charts
  2. Multi-year APM/RAPM with aging incorporated
  3. Revise ASPM based on multi-year RAPM with aging
  4. ASPM within-year stability/cross validation
  5. Historical ASPM Tableau visualizations
  6. Create Excel VBA recursive web scraping tutorial
  7. Comparison of residual exponents for rankings
  8. Comparison of various "value metrics" ability to "explain" wins
  9. Publication of spreadsheets used
  10. Work on using Bayesian priors in Adjusted +/-
  11. Work on K-Means clustering for player categorization
  12. Learn ridge regression
  13. Temporally locally-weighted rankings
  14. WOWY as validation of replacement level
  15. Revise ASPM with latest RAPM data
  16. Conversion of ASPM to" wins"
  17. Lineup Bayesian APM
  18. Lineup RAPM
  19. Learn SQL