Since the NBA playoffs are starting today, I figured I’d throw out my own predictions. I used something I call “True Winning Percentage” (which I will explain some other time, but essentially, it takes into account opponent’s records, and opponents’ opponents’ records, and opponents’ opponents’ opponents’ records, and so on, and determines what a team’s “true” winning percentage would be if they played each other team infinite times, at least, theoretically), to calculate odds for each first round matchup, and then compared those projected winners and so on. Here are my predictions, along with my estimated probabilities of each team I pick to win actually winning the series:
|Likely victor||Probability of likely victor winning series|
Note that the matchup between San Antonio and LA is essentially a dead heat. However, Boston has essentially identical odds against both, so the final outcome is not so much in doubt.
I also calculated overall odds of each playoff team winning the title, given their probability of winning against each other playoff team, these odds are as follows:
|Team||True Win%||Prob of championship|
As you can see, Boston has better than 3:1 odds of winning before any basketball has even been played. They are just that much better (even taking into account that they played most of their games against weaker competition) than everyone else. Incidentally, to see the effect of calculating True Win%, notice that Detroit, which had the second best win-loss record, falls to fifth when you take into account their opponents’ (and so on…) success. To conclude, given that the odds of any other team winning come in at 22.47%, I feel pretty safe picking the Celtics this year.
Notice that I don’t predict much success for Utah, contrary to several other prognostications that have them performing very well in the playoffs. I understand that these other folks are using scoring efficiencies and the like, but I’m sticking to my guns. Consider this my bold prediction: Houston will actually beat Utah in the first round (although it will be close).