Three-Point Shooting Teams and The 2014-2015 NBA Playoffs

Major Update: June 22, 2015.  I have now published a formal article on the arXiv proving many of the assertions made earlier in this blog post. It can be found here: http://arxiv.org/abs/1506.06687

Some controversy was stirred up today when Knicks President and Basketball coaching legend Phil Jackson made the following tweets regarding three-point shooting teams not doing so well in the second round of the playoffs:

Advertisements

Data Analytics and The 1995-1996 Chicago Bulls

It is without question that the greatest team in NBA history was the 1995-1996 Chicago Bulls. They went 72-10 that year and went on to win the NBA Championship against a top-notch Seattle Supersonics team.  

Phil Jackson’s system and first-class coaching were the major reasons why the Bulls were so good, but I wanted to analyze their reason for winning using data science methodologies.

The results that I found were very interesting. First, I mined through each individual game’s data to obtain patterns in the Bulls wins and losses, and this is what I found:

One sees that the Bulls were a defensive nightmare, and if you look at these results in detail, it makes sense that the Sonics were really the only team that ever posed a threat to them. This shows that to beat the Bulls, the opposing team would have to simultaneously:

  1.  Ensure Ron Harper had a FG% less than 44.95% in a game,
  2. Ensure Dennis Rodman would have less than 17 total rebounds in a game,
  3. Ensure Luc Longley had less than 2 blocks in a game,
  4. Ensure Michael Jordan had a FG% less than 46.55% in a game.

If any one of these conditions were not met, the Bulls would win!

This analysis on some level also dispels the notion espoused by several sports analysts like Skip Bayless of ESPN who continually claim that the Bulls’ sole reason for success was Michael Jordan. Ron Harper’s contributions although of paramount importance are rarely mentioned nowadays.

This analysis also shows that the key to the success of the Bulls was not necessarily the number of points that Jordan scored, but the incredible efficiency with which he scored them.

A boosting algorithm also allows us to deduce the most important characteristics in the Bulls’ quality of play and whether they would win or lose a game.  The results are as follows:

We see that a key feature of the Bulls’ quality of play depends on how efficient Ron Harper in terms of his FG%.

It is quite interesting that this analysis shows that winning a championship is not about one player, sure, every team needs great players, but the Bulls were a great team, consisting of many great components working together.

Data Analytics and The Raptors 2015 Loss

          Based on several internal statistical models that my colleagues and I developed, we all have concluded that the Raptors losing the way they did in the first round was somewhat of a statistical anomaly. Through an extensive analysis, I present evidence below that shows it was due to several coaching breakdowns in strategy that lead to the Raptors’ collapse. 

Optimal preparedness would have been to prepare and utilize an extensive analysis of the Washington Wizards’ style of play. Using advanced machine learning techniques, we generated two results, first based on tree boosting, and the other based on classification trees that found the weak points in the Wizards’ system that would have greatly helped the Raptors in this series. 

First, one should be interested in the most important commonalities and characteristics in the Wizards’ play. This result is as follows:

  
One can immediately see that out of several factors, the two most important factors in determining whether the Wizards will win or lose a game is their team FG% and the number of points their opponent score in a game. From this analysis, we obtain that to beat the Wizards, the Raptors should have focused on particularly strong interior defense, and in particular, stopping penetration. From an offensive point of view, the Raptors should have played a strong and slow half-court game focused on getting close-to-the-basket, high-percentage shots, instead of “high-octane” running up and down the court as they seemed to do very frequently. 

Going deeper in this analysis, one also has as a result the following classification tree:

  
In this tree, “W” and “L” denote whether the Wizards will win or lose a game, “FG.” denotes the Wizards’ FG%,  “OFG.%” denotes the Raptors’ field goal percentage, and “OPTS” denotes the number of points in a game the Raptors should score. One sees that for the Wizards to lose games, the coaching strategy should have been designed to ensure that the Wizards would shoot below 45.25%, while the Raptors should have shot at least 40.3% each game. Complementary to the above analysis, one notes that since three point shots are not fundamental to the Wizards’ offense, to accomplish this, the Raptors should have had strong half-court defensive schemes (including traps and trapping zones), combined with slow-paced, interior offensive schemes. 

In conclusion, it is important to note that these analytical results and ideas were available well in advance of the NBA playoffs, and the Raptors would have tremendously benefited from using these ideas. I would also like to point out that I have only offered a preview of the results I obtained. I have also developed several results pertaining to optimal offensive and defensive schemes that would not only change the way the Raptors play, but would make them significantly better.

NBA Scores Predictions – April 11, 2015

I am testing out a new algorithm that I have been developing over the past few months that attempts to predict the outcome of sports games, in particular, NBA games. I am taking it out for a “Test Run” today. Here is what I predict:

aprl112015predictions

Probabilities in principle are not too difficult to predict assuming you have the correct algorithm! What is more challenging is trying to predict the scores. Here is my prediction for the individual game outcomes:

Team 1

Team 2

Point Difference

p1

p2

Result

MIA

TOR

2.565

0.420

0.580

TOR

CHI

PHI

6.461

0.701

0.299

CHI

NYK

ORL

0.850

0.467

0.533

ORL

MIN

GSW

8.652

0.225

0.775

GSW

MEM

LAC

2.008

0.432

0.568

LAC

UTA

POR

4.115

0.342

0.658

POR

Note: p1 and p2 denote probabilities of each team winning.

Breaking Down the Knicks’ Season

Like many of my fellow Knicks fans, I am in an absolute state of shock and disappointment as the Knicks are currently 5-29 to start the new year! Many analysts from the standard outlets, ESPN, Yahoo! sports, etc… have given their share of reasons why the Knicks are playing the way they are. Being a mathematical physicist and data scientist, I decided to see if one could deduce any useful information from how the Knicks have been playing to see what is the true reason why they are losing all of these games. Here is what I found. Based on the data available at Basketball-Reference.com,  I designed an algorithm in R to go through each game, and fit regression trees (Here is a link to more on regression trees if you are unfamiliar with the concept) and found the following:

1. The number of points the Knicks score per game:

knicksplot1From this regression tree, we see that if the Knicks for example make less than 33.5 FG’s in a game, and have a 3-Point shooting percentage of less than 0.309, they will be expected to score no more than 79 points in a game. On the other hand, if they make more than 38.5 FG’s in a game, and also attempt more than 19 free throws in a game, they can be expected to score more than 111 points in a game.

2. The number of points the Knicks’ opponents score per game:

knicksplot2From this regression tree, note that first “Tm” denotes how many points the Knicks score in a game. We see that for example, if the Knicks have less than 28 defensive rebounds in a game, also score less than 98 points in a game, and have fewer than 4-5 blocks in a game, their opponents will slightly outscore them, and win the game. In fact, if the Knicks manage to get less than 28-29 defensive rebounds per game, and score less than 98 points in a game, they will be expected to lose every game they play! Now, let’s say, the Knicks do manage to get more than 28 defensive rebounds in a game, if they still only manage to score less than 89 points in a game, they are still almost guaranteed to lose as well.

Although, many analysts have probably pointed these things out, the conclusion one draws from these regression tree analyses, is that the Knicks have a significant problem with defensive rebounding, as that seems to be the number one factor in them not winning games. Further, they also have a significant problem with how many points they score per game, which is a direct result of this Knicks team still not running their offense correctly.

Would Tyson Chandler have made a difference? As the above analyses show, no single factor determines whether the Knicks win games or not. It is reasonable to assume that if Tyson Chandler was on the team, then, the Knicks would get more than 28-29 defensive rebounds in a game. But, according to the above analyses, and the right of the previous regression team, if they still as a team would attempt more than 78-79 field goals, they would still be expected to lose every game. The question then remains would Tyson Chandler’s presence increase the Knicks’ offensive efficiency? In principle, according to his career FG% stats, I would say yes. According to Basketball-Reference.com, Tyson Chandler had a FG% of 0.638 while in New York, and for his career has a FG% of 0.588, which is quite high for NBA standards. It is quite reasonable to assume therefore, that the Knicks would have considerably less FGA’s (certainly less than 78-79) in a game, and their opponents would be held to around 91.0 points per game. One would conclude that from a statistical perspective, trading away Tyson Chandler was perhaps a mistake and had an overall negative impact on the team’s performance both defensively and offensively.