[Power Rankings] Are the Rockets in trouble?

I noticed that there were some request or comments stating that I should include the strength of schedule (SOS) into account. I have several rankings that I have tried out over the years. All of the power rankings include OE and DE.

Luckily, I have one coded with SOS so it is just as simple as frobbing a bit. The euclidean distance discussed in the original post is weighted by the strength of schedule (i.e., teams with weaker schedules are penalized and vice versa). In my experience, the power rankings without SOS are just as robust as the ones with it.

Here are the power rankings using OE and DE with and without SOS for the first 13 games of the season:

<style type="text/css">
table.tableizer-table {border: 1px solid #CCC; font-family: Arial, Helvetica, sans-serif; font-size: 12px;} .tableizer-table td {padding: 4px; margin: 3px; border: 1px solid #ccc;}
.tableizer-table th {background-color: #104E8B; color: #FFF; font-weight: bold;}
</style>

<table class="tableizer-table">
<tr class="tableizer-firstrow"><th></th></tr> <tr><td>OE, DE</td><td>OE, DE, SOS</td></tr> <tr><td>&nbsp;</td><td>&nbsp;</td></tr> <tr><td>1. Miami Heat</td><td>1. Miami Heat</td></tr> <tr><td>2. Los Angeles Lakers</td><td>2. San Antonio Spurs</td></tr> <tr><td>3. San Antonio Spurs</td><td>3. Los Angeles Lakers</td></tr> <tr><td>4. New Orleans Hornets</td><td>4. Boston Celtics</td></tr> <tr><td>5. Boston Celtics</td><td>5. New Orleans Hornets</td></tr> <tr><td>6. Orlando Magic</td><td>6. Portland Trailblazers</td></tr> <tr><td>7. Portland Trailblazers</td><td>7. Orlando Magic</td></tr> <tr><td>8. Dallas Mavericks</td><td>8. Dallas Mavericks</td></tr> <tr><td>9. Chicago Bulls</td><td>9. Chicago Bulls</td></tr> <tr><td>10. Atlanta Hawks</td><td>10. Denver Nuggets</td></tr> <tr><td>11. Denver Nuggets</td><td>11. Atlanta Hawks</td></tr> <tr><td>12. Utah Jazz</td><td>12. Utah Jazz</td></tr> <tr><td>13. Oklahoma City</td><td>13. Oklahoma City</td></tr> <tr><td>14. Milwaukee Bucks</td><td>14. Milwaukee Bucks</td></tr> <tr><td>15. Indiana Pacers</td><td>15. Phoenix Suns</td></tr> <tr><td>16. Charlotte Bobcats</td><td>16. Indiana Pacers</td></tr> <tr><td>17. New York Knicks</td><td>17. Charlotte Bobcats</td></tr> <tr><td>18. Phoenix Suns</td><td>18. Houston Rockets</td></tr> <tr><td>19. Toronto Raptors</td><td>19. New York Knicks</td></tr> <tr><td>20. Houston Rockets</td><td>20. Memphis Grizzlies</td></tr> <tr><td>21. Philadelphia 76ers</td><td>22. Philadelphia 76ers</td></tr> <tr><td>22. New Jersey Nets</td><td>22. Toronto Raptors</td></tr> <tr><td>23. Memphis Grizzlies</td><td>23. Golden State Warriors</td></tr> <tr><td>24. Golden State Warriors</td><td>24. New Jersey Nets</td></tr> <tr><td>25. Detroit Pistons</td><td>25. Detroit Pistons</td></tr> <tr><td>26. Sacramento Kings</td><td>26. Cleveland Cavaliers</td></tr> <tr><td>27. Cleveland Cavaliers</td><td>27. Sacramento Kings</td></tr> <tr><td>28. Washington Wizards</td><td>28. Washington Wizards</td></tr> <tr><td>29. Minnesota Timberwolves</td><td>29. Los Angeles Clippers</td></tr> <tr><td>30. Los Angeles Clippers</td><td>30. Minnesota Timberwolves</td></tr></table>

Also, I am reposting the plot that contains the Houston Rockets power ranking as a function of games played and the Houston Rockets strength of schedule as the season has progressed.

<table style="width:auto;"><tr><td><a href="http://picasaweb.google.com/lh/photo/nBuWKljXV6eRz45RPTVErQ?feat=embedwebsite"><img src="http://lh3.ggpht.com/_8u4A05PwTfw/TPFLnhvLIYI/AAAAAAAACOs/WmkprorkJ2U/s640/nba10_pr_rockets.jpg" height="480" width="640" /></a></td></tr><tr><td style="font-family:arial,sans-serif; font-size:11px; text-align:right">From <a href="http://picasaweb.google.com/jsmee2000/PowerRankings?feat=embedwebsite">Power Rankings</a></td></tr></table>

In summary, the SOS didn't seem to give certain teams the boost that they would have expected (i.e., MIA - down or UTA up). In fact, the Rockets went up in the rankings as soon as SOS was included. That's because there are several teams with easy schedules but can't seem to win games.

There were nine teams from the Western Conference in the top 15. I don't know you but that scares me immensely.

 

Since every team has played 14 games, here is an updated OE vs. DE chart.

<table style="width:auto;"><tr><td><a href="http://picasaweb.google.com/lh/photo/jJ3EZzIm7G1xDksU8g0OFA?feat=embedwebsite"><img src="http://lh5.ggpht.com/_8u4A05PwTfw/TPJWfXy5IcI/AAAAAAAACPA/qta-5AtofJE/s640/nba10_pr_norm_chart14.jpg" height="480" width="640" /></a></td></tr><tr><td style="font-family:arial,sans-serif; font-size:11px; text-align:right">From <a href="http://picasaweb.google.com/jsmee2000/PowerRankings?feat=embedwebsite">Power Rankings</a></td></tr></table>

In terms of the Rockets, this chart includes only up to the Golden State Warriors game. Indiana has only played 14 games so I can't include the Bobcats game in here.

Here is the updated Rockets chart (noticed how are schedule keeps getting easier and we are not winning) :

<table style="width:auto;"><tr><td><a href="http://picasaweb.google.com/lh/photo/CKciMsnUGxGQcLudwJQ9DQ?feat=embedwebsite"><img src="http://lh3.ggpht.com/_8u4A05PwTfw/TPJWfrneTLI/AAAAAAAACPE/xMKiySd22FA/s640/nba10_pr_rockets.jpg" height="480" width="640" /></a></td></tr><tr><td style="font-family:arial,sans-serif; font-size:11px; text-align:right">From <a href="http://picasaweb.google.com/jsmee2000/PowerRankings?feat=embedwebsite">Power Rankings</a></td></tr></table>

Notice that three lines appear in the top plot. These are three different power ranking computations that pretty much state the same thing.

Ranking 1 - is based on the euclidean distance between any team and a "perfect team" based on OE and DE.

Ranking 2 - same as Ranking 1 except that teams are penalized or awarded based on their strength of schedule.

Ranking 3 - same as Ranking 2 except that the OE and DE are weighted with an exponential forgetting factor. In other words, most recent games count more.

Initially when I started assessing these rankings, I tried various ways to weight the games based on the date played. For example, I tried linear weighting, before all-star/after all-star break, Hollinger weighting (weigh a team's full-season results by two-thirds and its most recent games by another one-third), etc. None of these seem to work well enough to keep them.

Something new that I am trying this year is the exponential weighting of the OE and DE. However, I will not be able to quantify its effects until later in the season. I thought I would share the results with you. Here is a figure representing the forgetting factor.

<table style="width:auto;"><tr><td><a href="http://picasaweb.google.com/lh/photo/Pd-tceWHmFC-jgMu5fHSwg?feat=embedwebsite"><img src="http://lh5.ggpht.com/_8u4A05PwTfw/TPJbgKH27-I/AAAAAAAACPM/A04H_dnp65I/s640/exp_scale.jpg" height="480" width="640" /></a></td></tr><tr><td style="font-family:arial,sans-serif; font-size:11px; text-align:right">From <a href="http://picasaweb.google.com/jsmee2000/PowerRankings?feat=embedwebsite">Power Rankings</a></td></tr></table>

The way it works is that the OE and DE for the latest game gets scale by the forgetting factor at one. Since the Rockets have played 15 games, their first game would currently get scaled by the forgetting factor at 15.