Category Archives: Sports Science

2016 NBA Draft – How do you put together a winning team?

The 2015-2016 NBA playoffs have just begun meaning 16 fortunate teams are still playing ball while 14 others are preparing for the 2016 Draft and beginning to set up the structure of their team for next season (“There’s always next season”).

The concept of drafting players is an interesting one. So much goes into it – athleticism, physical stature, game smarts, college performance, and the player’s mentality (IE, will they be able to handle the pressure, will they fit in with the guys and have good team chemistry, etc). Recently, Motomura and colleagues (2016) discussed the role the draft can playing in building an NBA franchise. More importantly, they set out to understand whether having more or higher draft picks actually made an NBA team better. They concluded,

“We find that the draft is not necessarily the best road to success. An excellent organization and General Manager better enable teams to succeed even without high draft picks.”

This got me thinking – could we potentially try and understand which teams are “excellent” organizations in terms of selecting players that enjoy success at in the NBA? Additionally, I am really interested in the Philadelphia 76ers. Year after year they always seem to be in the conversation of tanking at the end of the season, in order to increase their chances of obtaining higher round draft picks in the NBA Draft Lottery. In fact, they have been so good at this over the past few seasons that the 2016 season is supposed to the final season of the tanking era in Philadelphia. Unfortunately, their efforts to tank and stock pile great players has not payed off. They seem to have a hard time either:

  1. Selecting good players. If you are going to tank you better not miss on your draft picks!
  2. Developing players or bringing in veteran players who can surround the young stars so that they don’t have to play a high number of minutes their rookie season and carry the team (something also addressed in the Motomura above).

The Data

2011 – 2015 NBA Draft data was obtained from basketball-reference.com.

Aims

  • With 60 picks in the NBA Draft (300 total over the 5 year period) how many players, on average, do teams pick up?
  • What is the average value of players selected in each of the draft number spots?
  • Which teams have been most successful at picking players that added a high amount of value to their team?
  • What is going on in Philly?

Number of Draft Picks

Over the 2011 – 2015 NBA Draft 300 total players have been chosen, with teams averaging 9 players drafted during that period. The 76ers certainly are leading the way, selecting 21 players over this 5 year stretch. (NOTE: You will notice there are 34 teams in the table below. This is because I left in expansion teams and teams that moved from one city to another during this 5 year period. I did this to just represent what took place in the draft between 2011 – 2015).

Screen Shot 2016-04-16 at 8.02.38 PM
What is the value of a draft pick?

Value or Success metrics are often one of the more difficult things to pin down when studying team sport athletes. Lots of things players do can add value to a team without ever making it into the box score (which primarily consists of count metrics). The writers at basketball-references.com display two metrics which I used to quantify a player’s value – Win Shares and Value Over Replacement Player. Both of these metrics are the type of metrics that were born out of Baseball’s Sabermetrics as a way of trying to provide more context to the box score metrics presented to fans everyday on websites or in newspapers. Win Shares is a metric that takes the teams success and divides up credit for that success among the participating players. Value Over a Replacement Player is a metric which projects the player’s value versus a fictitious replacement player. Both of these metrics have limitations and people argue frequently over which is more useful or whether we should use a different metric to represent value (E.g., Player Efficiency Rating or something like +/- or Adjusted +/-. Both of which have their own limitations). I simply chose these metrics because they were readily available and they would provide me with a quick way to represent player value. Any metric one deems important would suffice, though.

To reflect value per pick I summarized the data in a few ways:

  • I binned the picks into groups of ten (Picks 1-10, 11-20, 21-30, 31-40, 41-50, and 51-60). Because I was dealing with a five year period it meant that there would only be 5 picks for each selection (1-60), which wouldn’t provide enough data. Thus, binning it this way helped me group more players together.
  • Since I am using 5 years of data it isn’t really fair to look at something like Win Shares for all of the players, since players who were drafted in 2011 have a much longer time to contribute to their win share compared to a player drafted in 2015 (a rookie). Thus, I reflected Win Shares over Games Played, to attempt to look at each player’s contribution to their teams success relative to the amount of games they participated in.
  • Finally, I added in Minutes Per Game, simply because I wanted to see what the participation differences were between the bins of draft picks.

The data in the below table is the average of each metric for the six different draft pick bins.

Screen Shot 2016-04-17 at 4.29.23 PM

As we would expect (or should expect) there is a monotonic decrease in each of the three metrics as we move from Pick 1-10 to Pick 51-60. This is to be expected and tells us that the quality of player begins to decrease as we move down the draft board (better players are being selected higher up). The only place this doesn’t seem to happen is in Pick 41-50 for the Average Value Over Replacement Player. I’m not really certain why this is. It could be that during this five year stretch there were a lot of players selected from those picks that had minimal to no contribution to their team.

Draft Pick Value Per Team

First, we look at the sum of Win Shares Per Game for each draft pick bin. I added up the win shares per game for each player the team selected in each of the draft pick bins and then summed those up to obtain a 5 year “Value Add”. I then standardized the scores in order to see how each team did relative to the average Value Add during this 5 year stretch.

Screen Shot 2016-04-17 at 4.35.42 PM
NOTE:
There is a limitation with this analysis in that I didn’t have a way of going through each player to see if they played for their draft team over the entire 5 year period. It is entirely possible that some players moved on or maybe got drafted and immediately traded and never had a chance to play with their draft team (as we will see when we discuss Philadelphia). That being said, what quickly jumps out is that 6 teams appear to be very good at identifying those who will be valuable NBA players, whether they still play on their draft team or not – Houston, Cleveland, Detroit, Denver, Minnesota, and Utah. It is important to keep in mind, however, that some of these scores might be coming from one or two players during this five year period. For example, guys like Karl-Anthony Towns (Minnesota) and Kyrie Irving (Cleveland) make significant contributions to their teams in terms of Value Add. Both players were also #1 draft picks.

Another interesting observation is the value Houston, Cleveland, and Detroit were able to find in Picks 31-40. Those three teams stand alone in that draft pick bin as all of the other teams seem to lack the ability to find valuable players. Houston looks to be pretty incredible at identifying talented players as they are green in several of the draft bins and have had the most success in drafting (using Wins Shares as the metric of success) compared to other teams over this period. Houston also happens to be a team that is praised for their analytic savviness and perhaps this helps contribute to their ability to scout talent.

In looking at this chart, Philadelphia doesn’t appear to be doing too bad (7th ranked team). However, it is important to keep in mind the limitation of this chart in that some players might be adding value for teams other than the team which drafted them. I do give Philly credit for identifying some of the players as potentially successful players but trading them away doesn’t help. This will be discussed later in the article.

 Next, we turn our attention to the Value Over Replacement Metric. For this analysis I took the average Value Over Replacement for each of the draft pick bins for each team. I then took the average of every draft pick bin for each team and created a 5 Year Average Value Over Replacement Player. This metric was then standardized for all teams to investigate how they did relative to the rest of the league.

 Screen Shot 2016-04-17 at 4.54.03 PM

Now we get a little bit of a different look at the league and how successful teams draft players. As in the above analysis, there is a similar limitation in that players may have moved on from the team that drafted them; however, the main goal is to understand who is good at identifying talent.

We still see Houston in the top 6. Not only are they selecting players that are adding win value but these players are also contributing more than the replacement player would. Golden State, who was in the top 10 on the previous chart, looks to steal the show here with players above the replacement level player. Philadelphia takes a bit of a hit in this chart.

So What is Going on in Philly?

This is a tough one to sort out. As I alluded to above, sometimes teams draft players and then move those players on to other teams. Philly has been accused of tanking in order to get better draft picks and if you are going to try and go out of your way to get better draft picks then you need to ensure those draft picks actually turn into great players. Otherwise, you just end up being in the same position next year. Philly drafted 21 players over the past 5 years – well above the norm for an NBA team during this time.

  • Of the 21 players drafted only 7 of those players actually ended up playing for the team in some capacity.
  • Of those 7 players, only 4 of them remain with the team.
  • Of those 4 players, one is Joel Embiid, who has not played a game in his first 2 seasons with the team due to injury. Embiid was the 3rd round pick in the 2014 draft and has proven, thus far, to be a very costly selection for the franchise.

Here is an overview of the 21 players Philly has selected in the past 5 years:

Screen Shot 2016-04-17 at 5.26.42 PMPlayers in red are players that are no longer in the NBA or never even made it into an NBA game. That is 10 out of Philadelphia’s 21 picks (48%) who either don’t play in the NBA anymore or never made it in the first place. Stockpiling picks in the hope that a few of them turn into something valuable might not be a horrible idea, but when almost 50% of the players have washed out of the league it may be hard to justify this strategy. Moreover, 33% of the players drafted no longer play on the team. This is including the former Rookie of the year, Michael Carter-Williams and Maurice Harkless (8.5 win shares and a value above replacement player of 1.9) who was traded for Andrew Bynum (who turned out to be an NBA bust). With only 19% (4 out of 21) of the drafted players still on the team (counting Embiid who has made no contribution at all due to injury) it appears to have been a pretty unsuccessful 5 years of drafting. The team was 10-72 this season and didn’t show much improvement over years past. Perhaps the tanking era isn’t over yet in Philly?

Conclusion

Drafting players is really difficult. There are a lot of things that go into it and some may say it is a lot of luck. That being said, there are some teams that seem to come out on top or near the top, year-after-year. You can have those big luck years where you snag a lot of great talent and hit a home run but I think more importantly you just need to be consistent. The big luck years are good but the years where you are consistently bad end up setting you back. As discussed in the Motomura paper, having a well run organization that understands how to not only develop talent but also bring in veteran players to surround the younger players and take some of the pressure off might be the most important thing. Too often I think teams try and tank with the idea that their first round pick is going to save the franchise next season. Instead, they should consider the things they need to do to help that first round pick develop into the player they need him to be, down the road, in order to save the franchise.

References

Motomura A, Roberts KV, Leeds DM, Leeds MA. Does it Pay to Build Through the Draft in the National Basketball Association? J Sports Economics 2016. 1-16.

 

Three things a high performance team can learn from The Profit

One of my favorite shows at the moment is CNBC’s, The Profit. The basic premise of the show is that millionaire investor, Marcus Lemonis, finds failing businesses, evaluates them, and then  provided he feels the business has potential, invests in the company for a certain percentage of ownership. He then establishes a road map to success by helping them understand what aspects of their business are broken.

In the show, Marcus preaches three main constructs which he feels are necessary for a successful business:

  1. People
  2. Process
  3. Product

In reality, a high performance team working with a sports franchise is no different and these three constructs are actually incredibly valuable for determining what areas your high performance team needs to improve upon or where you may need to make some changes in order to have better success and be more efficient.

People

“Do we have the right people in the right positions?” When thinking about this question it is important to not only think about skill set and ability but also whether or not they work well as part of a team. A lot of times, teams or universities are afraid to let someone go because they have “been there for a long time” or they are “a nice guy”. I understand this can be a tough thing but at the end of the day, keeping people around that are unable to contribute to the level and expectation that is needed is going to create more problems and frustration in the long run. It makes sense to part ways and ensure that the people you are putting together on the staff have a very high level of skill set and interest in continuing to learn and push things to new levels. Additionally, it is important to move along from those who are insecure and create turf wars between departments. These individuals can tear a team apart in a second and create problems within the high performance team. A high performance team is one that is collaborative across the main player support departments – Sports Science, Athletic Training/Medical, and Strength & Conditioning. If the people within the staff are not interested in collaboration and working together then the high performance team will never work. In the Profit, Marcus evaluates people within the businesses he invests in and, at times, is forced to make the decision (with the other owners he has partnered with) to let people go who are not willing or able to satisfy the need of working collaboratively in a successful business.

Process

Marcus is a stickler for process. His famous quote in the show, after handing over a check for his investment and becoming a part owner is, “I may be a part owner but I am 100% in charge”. Oftentimes, where businesses fail is not in the people or the product, but in the process. They can’t seem to put the appropriate processes in place to ensure that product gets manufactured at the right cost, without wasting money, or the product gets ordered at the right amount, without having a back log of inventory. Being a successful entrepreneur, Marcus sets up some very specific processes for these companies to ensure that business is performed in an efficient and timely manner. Within the high performance team environment this process is essential. What is the flow of data – how is it collected, processed, analyzed, and then distributed and discussed amongst all support staff and key stakeholders in the building? Things can be very busy in a professional or university team environment, making these processes even more critical. Oftentimes, information falls through the cracks because there is not a process in place for ensuring that people on the staff get together and meet on a daily basis to discuss the data and develop a plan about what to do with the data – turning data into action.

Product

Finally, product. Obviously a high performance team isn’t making any product; however, a high performance team is serving the athlete to ensure that athlete’s health and wellness is cared for during their time with the team. This “product” is really the outcome of having great people, with a high skill set and standard for excellence, who can work together and having great processes in place, ensuring that the information flow between departments is fluid and efficient.

Collectively, these three constructs will ultimately determine the success that your high performance team has and their ability to adequately effect the athlete’s within the training environment ultimately decreasing injury and improving performance.

2014-2015 NBA MVP (Canadian)

A little over a week ago I posted some analysis of the three players currently in the running for MVP of the 2014-2015 NBA season – Curry, Westbrook, and Harden. This week, I wanted to look at the current crop of Canadian stars who are playing in the NBA.

There has been a lot of basketball talent coming out of Canada recently and the 2014 NBA draft was full of young stars, with Andrew Wiggins leading the pack as the number one overall pick (Choosen by the Cleveland Cavaliers who then sent him to Minnesota, in a trade that brought Kevin Love to Cleveland to play with Lebron James and Kyrie Irving). Additionally, some of my best friends happen to make up the sports medicine, sports science, and strength and conditioning staffs for the Canadian Men’s National Team leading up to the 2016 Olympics (so obviously I am cheering them on).

OH CANADA

OH CANADA

Players Analysis

One of the biggest issues with the analysis is that several of the players don’t play very many minutes. That being said, I included them anyway.

Screen Shot 2015-04-26 at 6.20.15 PMThe main players to play significant minutes were Cory Joseph, Tristan Thompson, Andrew Wiggins, Nik Stauskas, Kyle Olynyk, and Robert Sacre; so we will concentrate the analysis on them.

Each player played in over 70 games and over 1000 minutes, with both Tristan Thompson and Andrew Wiggins making an appearance in all 82 games and playing over 2000 minutes (Wiggins played nearly 3000 minutes in his rookie year).

As you can see in the chart below, Wiggins had the highest average points points per game of the group – averaging around 17 points per game.

Screen Shot 2015-04-26 at 6.37.12 PM
However, as discussed in the previous blog article, points aren’t everything. To be an MVP you need to help make others around you better. There are times where players score a lot of points but are actually problematic to their team and cause less winning opportunities. For example, a ball hog who takes a lot of shots, has a poor field goal percentage, and turns the ball over frequently because he is always trying to control the court rather than distributing the ball to his teammates. Allen Iverson was a good example of this, at times, and, if my memory serves, there were three seasons where he led the league in turnovers and despite scoring a lot of points he had a poor field goal percentage and scoring efficiency. In Berri and Schmidt’s research, Iverson actually cost his team wins because of his play, despite the fans enjoying the show – everybody likes to see a guy score lots of points!

Wins Produced

Speaking of Berri and Schmidt’s research, as we did in the previous blog article, we will turn our attention to the Wins Produced model, which allows us to understand the player’s contribution to his team winning games throughout the season. How many things does the player do well and how good is the player at minimizing things that cause the other team to score points?

Going back to Wiggins, while he scored more points than the other guys in the analysis – he had a lot more opportunities to score given the high amount of minutes he played – he only produced about 2 wins for his team (or 0.03 wins per 48min). One reason may be due to his high amount of turnovers.

Looking at Cory Joseph and Tristan Thompson, we see that both players helped contribute about 7 wins to their team. While Joseph’s first two NBA seasons were nothing to write home about, he has put together a great season on a stacked San Antonio Spurs team and might be a guy they look to in the future to run the point guard position as their team continues to age. Meanwhile, Tristan Thompson finished fifth in NBA 6th Man Voting and had a great season coming off the bench on a Cleveland CAVS team led by King James.

Kyle Olynyk pops out in the Wins Produced stat as actually being a bit detrimental to the Boston Celtics. Here is an example of a guy who played a lot of minutes, however, his production is actually less than what an average Center would be able to do given the same number of minutes he played (Olyny played about 100 more minutes than the average for Centers). Olynyk was good for 93 offensive rebounds and 211 defensive rebounds, while Centers, on average, this season pulled down 123 offensive rebounds and 264 defensive rebounds. Kyle did do better than the average in scoring, 656 points to the league average, for Centers, of 522 points. However, he did turn the ball over more than the average, 98 turnovers versus the league average of 74. When looking at all the factors that go into the model, Olynyk didn’t seem to be effective. During his rookie season, 2013-2014, one of the criticisms is that he is not a true center and lacks the ability to defend some of the best big men in the league.
Stauskas produced 0 wins for his team, the Sacramento Kings and Robert Sacre was actually more detrimental to the Los Angeles Lakers than Olynyk was to the Celtics!

Some Other Thoughts

Looking at the stats, it appears that it is a toss up between Joseph (who actually played really well despite playing about 800 minutes less than Thompson) and Thompson for the Canadian MVP. Both had great seasons and contributed a lot to their teams.

Wiggins had a good season as well and, as a rookie, has a lot of room for growth. Controlling the ball is going to be something he will have to work on in the offseason.

Olynyk just finished his second season on a young Celtics team. Perhaps playing Center isn’t his position but he may just need more time to grow into it. Despite having a negative wins produced stat for his team this year, he did start to show promise towards the latter half of the season.

One thing I think about, from a health stand point, is the number of minutes some of these young players are playing. With 3936 available minutes in an NBA season, not counting overtime games, Wiggins logged a massive amount of minutes. Keeping players healthy is the name of the game and managing their health by managing their minutes played (as well as how you help them recover off the court) is going to be critical for these rising stars.

2014-2015 NBA MVP Analysis

Currently, there is an intense debate over who is more deserving of the 2014-2015 NBA MVP. The three main front runners are Steph Curry (PG, Golden State Warriors), Russell Westbrook (PG, Oklahoma City Thunder), and James Harden (SG, Houston Rockets).

All three players had remarkable seasons and you could make a case for each one (despite Westbrook’s team not even making it to the post season).

All three players were leaders in Points scored in their respective positions:

  • Steph Curry = 1900 points
  • Russell Westbrook = 1886 points
  • James Harden = 2217 points

While Harden scored more total points, he actually scored less than Westbrook per game (keep in mind that Westbrook also missed several games in the beginning of the season, due to injury). Both are in the running for the SCORING TITLE this season:

  • Westbrook – 28.2 pts/g [95% CI: 25.8 , 30.6]
  • Harden – 27.4 pts/g [95% CI: 25.2 , 29.5]
  • Curry – 23.8 pts/g [95% CI: 22.0 , 25.5]

The difference between each player and the margin of error of their difference scores is as follows:

  • Curry compared to Westbrook = Difference: 4.40 pts/g with a 2.99 Margin of Error
  • Curry Compared to Harden = Difference: 3.62 pts/g with a 2.79 Margin of Error
  • Harden Compared to Westbrook = Difference: 0.78 pts/g with a 3.22 Margin of Error

From this, it appears that, while Harden scored more total points, Westbrook appears to average more points per game than the other three. Of course we don’t know how we would have performed had he not missed games early in the season. Both Harden and Curry played in 80 games while Westbrook only played in 66. This, along with his team not making the playoffs, may end up hurting him in MVP Voting.

While there are some differences in production between the three players, all of them were incredible at putting up points. In relationship to the average player at their respective positions, Curry and Westbrook were both 3.2 standard deviations better than the average, while Harden was an astonishing 4 standard deviations better than the average shooting guard.

Points Aren’t Everything

While scoring a lot of points is important, it isn’t everything. A key aspect of a player, particularly an MVP, is whether or not he makes his teammates better. What is the value of the player to the team and how is he able to contribute to helping the team win?

One metric that is useful to help answer this question is Wins Produced, from David Berri and Martin Schmidt. The metric is designed to understand a player’s contribution to winning and it factors in not only the players stats for that season but also how an average player would have contributed given the same opportunities (minutes played), allowing us to understand how much more effective the player was then the average player that season.

To compare these players to the average for their position groups, I took every player in the league who played in more than 200 minutes during the season. The Wins Produced for Curry, Harden, and Westbrook are:

  • Curry = 18.8 Wins Produced (0.344 Wins Per 48min)
  • Harden = 16.9 Wins Produced (0.263 Wins Per 48min)
  • Westbrook = 11.2 Wins Produced (0.237 Wins Per 48min)

The Wins Produced and Wins Per 48min for all three of these players is exceptional. When we compare the three, we see that Curry produced a bit more wins and wins per 48min than Westbrook and Harden. Note that Westbrook drops a little bit here which could, again, be reflective of the fact that he played 14 less games that the other two. Curry appears to make larger contributes to help his team win games.

Curry produced approximately 17 more wins than the average point guard. Westbrook produced 9 more wins than the average point guard and Harden was able to produce approximately 13 more wins than the average shooting guard.

Who Should Be Crowned MVP?

I don’t know that there is a simple answer here. All three of these guys played incredible this season. If I were casting a vote, my vote would go to Curry. While he scored about 300 total points less than Harden and he averaged slightly less than both Harden and Westbrook, he did have a slightly less variance from game-to-game points (although not significant). Additionally, because of his consistency and ability to help create more wins for his team per 48min, I believe he is deserving of the MVP. The MVP provides the most value to his team by helping them succeed.