The Pitfalls of Statistical Analysis: Basketball vs. Baseball

Vicious D takes a two-part look at how statistics are used in basketball and how they can sometimes be problematic, ineffective, or misleading (but seldom boring).  In this first part, he contrasts two opposite ends of the statistics spectrum: Basketball and Baseball.

Stats.

Fantasy players love them, apologists ignore them, and just about everyone acknowledges their usage in analysis. 

But just how useful are statistics in a game like basketball?

When I was starting out in university as a Computer Science major at the University of Waterloo, statistics was a class that was required, but it was interesting and showed me a new way to quantify various objects in life.  When I later switched my major to Economics, statistical analysis was required to analyze trends and figure out the "big picture".  However, there were also a couple harsh lessons that almost everyone in my first year class took away from the year:

1. Statistics can lie.  Which statistics we chose to measure and which ones we choose to ignore will often paint us a picture that can be detrimental to our analysis.  It is always preferable to have a larger sample size and more measurements, but it is sometimes unavailable or too time-consuming and cost-prohibitive to do so.  
2. Statistics will be used to lie.  What we choose to show to people from available statistics becomes important as people have an inherit trust in numbers.  How we present the numbers also affects how people perceive a situation.
   

There's little that we can do to dissuade the second point.  People inherently believe numbers thrown at them.  Numbers are often used to great effect as they are seen as undeniable truths and are factual cornerstones. On the flip side numbers can be far from objective if used improperly.  To combat this, we need to be aware of how people choose their numbers to create their "skewed" statistics.

However, it's the first point that I wish to bring to your attention. 

Basketball is first of all, one of the harder sports to analyze.  We, of course, have the standard statistics available from the NBA.com websites which are replicated across other sites such as Yahoo Sports.  For further statistical analysis, sites such as Wages of Wins, Hollinger's ESPN Statistical Analysis, and 82games.com all provide their own unique look at the numbers breakdown.  I'll be covering the individual statistics and their effectiveness next week, but as for why it's so difficult to analyze basketball, we'll have to look at another sport: baseball. 

On MLB's own page for Suzuki Ichiro, we can find 29 offensive statistic categories and 13 defensive/fielding stats.  On the other hand, Chris Bosh's page on NBA.com displays 15 combined offensive and defensive stats.  It's quite a staggering difference. It can be easy to argue that these two sports are not a fair comparison as they're as different as apples are to oranges.  However, baseball is the perfect sport to illustrate why basketball statistics can sometimes be flawed.

In baseball, we have a sport that is clearly defined between offensive and defensive categories.  What one player does when he is out in the field or at bat seldom affects the other.  A baseball player's role is well defined.  When you are hitting the ball, you can measure certain kinds of statistics that only pertain towards hitting the ball and when you're running the bases, there's another set of statistics that only measure your effectiveness at running the bases.  In fact, it can be argued that baseball is a team sport that relies on the individual performances of players rather than players all working together at the same time, all the time.  A play can be broken down into a series of moves being passed from one player to the next.  For example, a pitcher pitches out so a catcher can take the pitch and then toss it to the short stop covering second base in order to tag a runner trying to steal second.  Each player has a part to play, but statistically, each player's contribution to the play is compartmentalized.  Their role before and after their individual contributions to the play have little bearing upon the success or failure of the play.  

In contrast, basketball is a sport that is always reliant on the other players on the court who are with you at the same time.  On offense, having a clear lane to the hoop for an easy basket relies on your 3 point threats drawing their defenders out to the perimeter, a perfect pick set by your big man, and a partial clear out under the basket so your wing can have room to score.  Each player has a part to play at the same time in the possession, but in the stats registered in NBA.com, we only ever concentrate on the points scored or the assist made.  Defense can be just as light on statistical analysis as a rebound grabbed by one player is reliant on having multiple players box out their man.  In fact, it's not surprising that the champions of the NBA teams that are the best collection of players that have learned to work together for a common goal.  It wasn't until Kobe Bryant learned to play with his team that the Lakers won their championship, and it's why a superstar player like LeBron James has still yet to wear a championship ring.

It's also why analysts in basketball often talk about the "intangibles" that players bring to the game.  Jorge Garbajosa is a recent Raptors example.  Despite having fairly average numbers across the board, Garbajosa brought things that we still talk about to this day and was a big factor in ensuring the Raptors won the Atlantic Division in 2007.  His ability to play more athletic players, his will to go after loose balls, and his overall smarts for what to do in various defensive situations all led up to statistics that are not regularly calculated or released to the general public by the NBA.  Players such as Garbajosa and Shane Battier are often applauded for their "intangible" work, but at the end of the day, isn't that just a way of glossing over the fact that the NBA statistic records are incomplete?

After all, what more are "intangibles" than statistics that are not completely calculated?

All that being said, statistics are incredibly important with the work we do at the HQ.  We use them to show a team from different perspectives and in order to give people a hard look at the team we all love.  We quibble over what statistics mean and whether a player will prove his worth on a roster.  However, if we were to simply only look at statistics in our analysis, we would be failing in our work to understand the players on our roster.  The picture would be incomplete because the statistics available are incomplete themselves. 

Next week, I'll take look at the NBA statistics available to us, which are beneficial, which are flawed and how to improve upon them.