Editor’s Note: the following post is from Ed Feng, founder of The Power Rank, a sports analytics company that focuses on college football and basketball.
There is an absolute certainty when traditional sports media picks a team to win the NBA championship. Whether the choice is Miami or Oklahoma City, they show utter confidence in their prediction. This team gets a 1 while the other 15 teams take a 0.
But you come to Wages of Wins to cut through that crap. There’s no certainty in the playoffs, especially when a team’s fate can change as fast as a torn meniscus in Russell Westbrook’s knee. This blog embraces that uncertainty, as Arturo gave us the probability for each team to advance past the first round. Even Milwaukee had a slim 0.1% chance to upset Miami.
This post takes it further and forecasts the odds for each team to win a title. But before diving into the methodology, here are the results for the remaining 14 teams.
The predictions page at The Power Rank will have updated results daily.
How to Forecast an NBA Game
This projection method is based on the NBA team rankings at The Power Rank. Inspired by Google’s PageRank algorithm and its use of Markov chains, I developed an algorithm that uses the margin of victory in each game to rank teams. In essence, it takes a team’s raw margin of victory and adjusts for strength of schedule.
While these adjustments are crucial in college sports with its wide disparity of teams, they might not seem important in a professional league with a salary cap. However, the rankings think otherwise. Of the top 10 teams, only 3 (Miami, New York and Indiana) come from the Eastern conference. While Houston was the 8th seed in the West, the Rockets were the 8th best team overall in the team rankings when the playoffs began.
The algorithm assigns each team a rating, and the difference in the ratings of two teams gives a predicted margin of victory on a neutral court. To predict NBA games, a home court factor, constant for all teams, is added for the home team. In the playoffs, the home court advantage is 4.5 points, significantly larger than the 3.2 points in the regular season.
These predictions may seems too simple, since they do not account for injuries, coaching changes or whether either team played the night before. However, from the 2007-2008 through 2010-2011 regular seasons, the team rankings predicted 70% of the game winners after 3 months into the season. The Vegas line from Oddsshark.com predicted 70.8% of winners in these games. These estimates over 2330 games have an uncertainty of 1%.
How to Forecast Each Round of the Playoffs
The predicted margin of victory discussed in the previous section corresponds to a win probability for each team. From these numbers, I developed a way to calculate the probability of winning a series using recursion.
In general, a recursive function calls itself as a way of repeating a calculation. In this sports calculation, the recursive function is called for each game in the series. This function then calls itself twice, once for each team’s winning the game. This self repeating process iterates over all possible outcomes of the series. It is also possible to simulate the series with random numbers, but this recursive function gives the answer much more quickly.
With these series win probabilities, the probability that a team advances through the later rounds is the product of these series probabilities. For example, San Antonio could meet either Denver or Golden State in the second round, so their probability to win that round includes a contribution from each of these teams.
Again, it’s also possible to simulate the playoffs by generating a random number for each series. I’ve also done this to check the accuracy of the direct calculation. However, the direct calculation has less error and runs in much less time.
Not surprisingly, Miami and Oklahoma City have the highest odds to win the NBA championship. These results are particularly interesting for 3 teams.
New York Knicks
Despite being 7th in the team rankings, the Knicks have the 4th highest chance to win the NBA title at 4.8%. They have an advantage playing in the weaker Eastern Conference. While the team rankings have New York and Memphis about the same, the Knicks will probably face Indiana, #8, in the next series while Memphis gets Oklahoma City, #1.
Of course, the Knicks must first advance past Boston, a chore they found quite difficult on Wednesday night when their contested jump shots weren’t falling.
Oklahoma City Thunder
The 30.4% chance for the Thunder to win the title does not consider Russell Westbrook’s injury. While this might seem like an egregious omission, I don’t think it matters.
Westbook took more shots than Kevin Durant this season despite shooting only 43.8% from the field (the NBA average was 45.3%). His departure lets Durant rightfully be the focus of the offense. The Thunder might get even better if Westbrook’s shots get distributed to Thabo Sefolosha and Kevin Martin, two players that both shoot better than 41% on 3 pointers.
Of course, the Thunder struggled from 3 point range (24%) Wednesday night in a loss to Houston. Expect them to shoot better when the series returns to Houston for game 6.
San Antonio Spurs
The Spurs are flying under the radar. One might think they only swept the Lakers because of the rash of injuries that crippled Kobe Bryant, Steve Nash and others.
However, the Spurs are 3rd in our team rankings, and they might move up past Miami and Oklahoma City if the rankings only used games in which Tim Duncan played. San Antonio has a 16% chance to win the title.
Thanks for reading.
Ed Feng is the founder of The Power Rank, a sports analytics company that focuses on college football and basketball. However, he is more excited about this year’s NBA playoffs than any season since Charles Barkley suited up for his hometown Philadelphia 76ers.