We can begin with the Power Rankings and work our way back:
|Team||Conference||Wins||Point Margin||Power Rk||Conf|
|San Antonio Spurs||West||56.1||5.4||3||2|
|Los Angeles Clippers||West||55.0||5.0||4||3|
|Oklahoma City Thunder||West||52.0||3.9||7||5|
|Golden State Warriors||West||44.7||1.3||12||8|
|New York Knicks||East||42.6||0.6||15||7|
|New Orleans Pelicans||West||34.2||-2.4||21||11|
|Portland Trail Blazers||West||34.1||-2.4||22||12|
|Los Angeles Lakers||West||29.1||-4.2||26||14|
I'm betting some of you are already familiar with my method for doing Power Rankings. I take the results of all the games and use a rather nifty variation of a Kalman Filter to come up with a running estimation of every team's strength (Editors note: a much more in-depth look at a similar method can be seen here courtesy of Jirka Poropudas). Done right, this method does a very good job of measuring the state of the NBA in real time and we can use it for forecasting the relative ranking of all teams. It lets me quantify some general expectations for matchups and build a playoff odds report.
One of the keys to making this method work is the initial seeding and the strength given for each team. It can overcome a bad initial seeding, but it'll make the first few weeks somewhat useless. This is where you readers come in.
A prediction market is speculative market created for the purpose of making predictions using a select group of informed participants. Typically, the accuracy of the prediction is tied to some incentive (like the winner getting a free custom t-shirt, a post on any topic they want [within reason], and eternal fame and glory). At this point you should be realizing that a fair amount of you just participated in our first (but certainly not last) prediction market exercise by picking the win totals for every NBA team. Of course we made sure that you where all as informed as possible before the contest.
We think so highly of your predictions that we've already compiled them and I'm about to use them as a key part of my initial seeding. Yes, that's right: you just helped build the inagural BoxScore Geeks Power Rankings.
We start with the results of our season model using the final projected rosters:
You should be seeing a nice visual interactive app that can show the full detailed projected win totals for each team this season. It shows the selected team's league-wide rank, conference rank, over/under odds, odds of winning their division, odds of finishing as the #1 seed in their conference, and odds of finishing as a top four seed. Miami, the number three team in our model, should be the default selection, but feel free to examine any of the others. You can visit our previews page if you want to see how we got to those numbers for each team.
That's almost all the information we need to get our power rankings together. We have the model representing the math's opinion, we have the Vegas numbers representing the sharps, and thanks to our readers, we have a gauge on public sentiment (well, informed public sentiment).
My initial Power Rankings are just the average of these three numbers.
For every selected team, you should be able to see actual wins last year, the Vegas over/under, the reader projection, the model projection, and our consensus win projection. Note that Miami is number #1 and finishes atop our inaugural Power Rankings (they shouldn't get too comfortable though).
Of course, the next step is to start incorporating results from actual NBA games. For that, you'll have to wait untill next week.