Those who even somewhat know me have probably heard me gush over basketball. I love playing, watching and even measuring basketball. To the last point, there have been many new efforts in the basketball community to attempt to better understand the game in a more mathematically rigorous and statistically relevant way. In much the same way Sabermetrics and Bill James revolutionized baseball, many new basketball advanced statistics and statisticians are revolutionizing basketball (As an aside if you are interested in learning the fundamentals of basketball statistics, you should get the book “Basketball on Paper” by Dean Oliver). So here goes my foray into basketball statistics.
Ever so slowly I have been building my own statistical model that simulates in real-time the NBA regular season as well as the playoffs and draft lottery. This model can then project which teams are likely to make the playoffs, how far they will advance once they get there, and other outcomes. The model is mostly based on offensive and defensive efficiencies, or points scored and points given up per possession and takes into account home-court advantage and rest days. The model is only slightly unique and off the top of my head I can name a few others, of which the one at basketball-reference.com is closest to mine. The data is rather noisy in the beginning of the season but becomes less so as more data points (games) are taken into account.
After running 5000 simulations of the remaining of the 2013-2014 NBA season (as of 02/15/2014), I have projections for the playoffs and lottery. The seeding and playoff odds take into account the divisions and tie break rules. However, the lottery odds don’t reflect lottery picks that have already been traded. For instance, the New York Knicks have almost certainly no lottery picks in the 2014 NBA draft (they traded both their first and second round picks to Denver and Houston, respectively, and only receive the second round pick of Sacramento, if the Kings end up as a top 15 team in the league). The actually probability the Knicks will win the lottery is effectively zero and the probability for the Nuggets then is 3%.
There is a lot to digest in these tables and I won’t describe everything. Unsurprisingly, the playoff picture at this point in the season is very much in focus. Starting in the Western Conference six of the playoff teams are clear. The Thunder, Spurs, Trailblazers, Clippers, Warriors and Rockets all have playoff probabilities of over 90%, pretty much guaranteeing them one of the eight spots in the Western Conference playoffs. The last two spots are being fought over by Minnesota, Phoenix, Dallas and Memphis. Earlier in the year it was projected that the Timberwolves would leapfrog these other teams on the strength of their plus/minus, but this simply hasn’t been borne out as the season has progressed. Now they sit with a 20% chance of making the playoffs.
The Eastern Conference is only slightly more cluttered at the moment, mostly among the bottom six seeds. Clearly Miami and more notably Indiana have distanced themselves from the rest of the East, and as many expect there is a decent chance that both will make it to the Conference Finals. Atlanta, Toronto, Washington and Chicago are also locks to make the playoffs (>90% chance), although the model gives them a combined 8% chance of making the NBA Finals. The bubble teams Detroit, Charlotte, New York, and Brooklyn are fighting for the last two spots to round out the playoff teams in the East.
The other side of the story is the race for last place and it is clearly being won by the Milwaukee Bucks. Currently, the model gives them a 25% chance of winning the draft lottery (the fact that this number is slightly greater than 25% is evidence that not enough simulations have been carried out to completely sample the statistics. The team with the worst record in the league is guaranteed a 25% chance at the top lottery pick).
One last thing…the NBA Championship! With nearly a 50% chance of winning the Finals, it looks as if Indiana is the favorite mostly due to the rather easy path through the rest of the Eastern Conference teams (including the Miami Heat) and also since the Western Conference is so strong, the good teams will pick each other off. I’ll leave it here for now, but look out for updates periodically until the end of the season as well as more context outlining the exact techniques used here and potential ways to improve them. Thanks for reading and enjoy some basketball.