Statistical projection models have Nuggets as heavy NBA Finals favorite

The era of sports analytics is well underway. The analysts have the Nuggets winning the championship — and rightly so.

Basketball is one of the least-variant and most accurately projected professional sports. While basketball analytics provide digestible ways to understand how the game is played, fans love narratives.

The way human beings think about probabilities and math is often misguided. The easiest example of this is what we often see at the Roulette table, wherein people look at the board, and observe that a red number has won 10 times in a row. Some justify a bet on red because “red is hot at the moment”, while others, with as much assurance, place a wager on black because “there have been so many reds, black is bound to hit.”

Both judgments imply a misunderstanding of probability — that the probability of the ball landing on red or black is the same regardless of the context of the prior spins.

The Miami Heat’s playoff run is a great story. After losing their first play-in tournament game to the Hawks, the Heat bounced back from the verge of elimination to beat Chicago and make it to the NBA Playoffs. Since then, the eighth-seeded Miami Heat successfully beat the top-seeded Milwaukee Bucks, fifth-seeded New York Knicks, and the second-seeded Boston Celtics to make it to the NBA Finals.

Miami will have a chance to finish the narrative as they get ready to face off against the Western Conference’s best team this season — the Denver Nuggets.

The statistical project models favor the Nuggets in the NBA Finals

NBA statistical projection models are a bit different than a roulette wheel in that context does matter somewhat. How much it matters is subject to the model inputs and how different input variables are weighted.

ESPN Analytics projects the Heat have an 11.3 percent chance to win the championship based on their BPI model, which is a power index model that weighs factors such as efficiency, location, and pace, among other variables.

FiveThirtyEight’s RAPTOR-based forecast, which predicts games based on individual player projections generated through nuanced models, projects more favorably for the Heat, assigning them a 27 percent chance to win the series. FiveThirtyEight’s other forecast, which is based on Elo Ratings, which takes into account head-to-head game results and factors in recent performance. The Heat’s Elo score in the FiveThirtyEight model has risen from 1500 on April 1st to 1640 today, and gives Miami a 43 percent chance of winning.

The best judge of a statistical model’s accuracy is not the outcome of one game, one series, or one season, but is subject to a test of its accuracy, or goodness-of-fit in statistical terms, over the long term. Often considered the most accurate of all models is those Vegas oddsmakers use, based on statistics and, crucially, an understanding of how to interpret betting market activity, to set lines that directly correlate to implied probabilities of events occurring. The implied probability of the Heat winning the NBA Finals is approximately 22.5 percent, as of the time of this article being written.

While the projections indicate the Denver Nuggets are the favorite to win the series, this is merely a reflection of probability. On occasion, a roulette player will pick a number and win despite a one-in-38 chance to win. If that same person bet that same bet 100,000 times, the likelihood he or she would come out as a winner is slim.

If this championship series were to play out 100,000 times, statistics tell us the Heat would likely win somewhere close to about 22,500 times. But this is basketball, not a statistics experiment. This series will be played once — and it should make for an interesting story.

Check out The Step Back for more news, analysis, opinion and unique basketball coverage. Don’t forget to follow us on Twitter and Instagram and subscribe to our daily email newsletter, The Whiteboard.