Search This Blog

Showing posts with label Stephon Marbury. Show all posts
Showing posts with label Stephon Marbury. Show all posts

Saturday, January 29, 2011

Misleading Stats

Anyone take statistics in college?  Well, any stat expert knows very well that you can make stats tell you basically anything.  Anybody can manipulate stats into telling you some ridiculously misleading things.  Car dealers, real estate agents, and any other shady salesperson knows how to hustle people, and part of their tactics are to use bogus stats - even if the stats themselves may be true.  News anchors, politicians, and other public figures heavily use stats to reflect their viewpoints.  The trick is to look for stats that actually mean something.  For instance, let's use the real estate example.  Any agent who is desperate for money these days (very common nowadays, especially in this economy) can give you seemingly convincing numbers to try to get you to buy a house quickly.  They will feed you the usual bs:  "This house was once valued at <some previously listed price that is ridiculously higher than what it was actually worth before coming down> - this is a great deal!"  Or, "So-and-so numbers in today's economy show that if you don't buy now, the price will start to come back up - you better jump on it!"

Basketball stats, in the same manner, can be just as deceiving.  In fact, a lot of player agents use this tactic to their advantage when trying to get a player as much money as possible (thus as much commission for themselves as possible).  They try to trick NBA GMs into thinking that their player is worth way more than what they are actually worth, using deceiving facts to help them buy into it.  How on earth did Darko Milicic land a $20 mill contract last offseason?  I think we would all like to know, though everyone in the league knows that the Minnesota Timberwolves' President of Basketball Operations, David Kahn, is not the smartest manager out there.  In fact, watch this hilarious clip of Chris Webber interviewing Kahn.  C-Webb couldn't hide how offended he was that Kahn actually thought that Darko is the best passing big man since Vlade Divac.  He also was shocked that Kahn had the nerve to compare Darko's growing pains to C-Webb himself.  I think even casual NBA fans know that there is no comparison whatsoever between the two.  I was about to put a comparison chart together, but then wised up and realized how much time and effort I would have wasted trying to prove the obvious.  So C-Webb couldn't hide his disgust over Kahn's comments, and Kahn reportedly called him a schmuck afterwards.

Anyway, the point of all this is to read between the lines.  It's difficult to judge how good players are just by looking at a stat sheet.  You need to dig deeper, just like John Hollinger does.  Sometimes the stats tell the story, but a lot of times they don't.  Guys that carry your fantasy team are not necessarily carrying their own REAL team.  Here are some of the biggest misleading stats in basketball:

Points

This is the most obvious one.  A lot of guys can score in the NBA.  But are all of the top scorers efficient?  Do they help their team win by jacking up so many shots?  Not necessarily.  There are so many examples of guys that scored a lot in their careers but never won (or got close to winning) a title.  Some guys just like to be gunners and take as many shots as possible.  Some are forced to carry a high scoring load, because the rest of their team can't score.

Let's take Monta Ellis and Kevin Martin, for instance.  They currently have the 4th and 10th highest scoring averages in the league.  Part of it is that the Warriors and Rockets are both terrible (requiring Ellis and Martin to jack up more shots), but the other part of it is that these guys need to take a lot of shots to get their points.  The best way to filter out gunners is to see how many field goal attempts (FGA) they need to get their points.  Truly great scorers that know how to win get their points from being aggressive and drawing fouls.  If they need a ton of FGAs to get their points, chances are they are not that efficient.

A.I. knew how to win scoring titles, but not championships

Since Ellis and Martin are victims of being on bad teams, I'll give them the benefit of the doubt.  Let's pick on Allen Iverson.  For simplicity's sake, let's pick the 2001-02 season, fresh off his MVP season as well as an appearance in the NBA Finals.  He scored 31.4 ppg - his highest season scoring average in his career, and led the league in scoring that year.  If he were to be a free agent that following offseason, his agent would be going bonkers and shoving those stats in GMs' faces.  "The league's best scorer needs a max contract!"  Let's dig deeper into that stat, though:  was he all that efficient?  To get his 31 ppg, he needed 27.8 shots per game to get there.  His shot percentage?  39.8%, with a horrid 29.1% touch from beyond the three point arc.  Ouch.  He does get some assists, but I will delve into that misleading stat in the next section.

You can argue that his team lacked offensive talent and needed him to carry that load.  But is being an inefficient high scorer benefiting your team all that much?

Assists

If you get a lot of dimes, you are seen as an unselfish player.  Unfortunately, it's not that black and white.  Using the efficiency analogy I used earlier with shot efficiency, there is also such a thing as passing efficiency.  One good way to measure this?  Weighing turnovers against assists.  Using the same 01-02 Allen Iverson statline, he had 5.5 apg.  At what cost?  He averaged 3.95 turnovers per game, good for worst in the league.

Now, you can argue that assist-to-turnover ratios can be deceiving as well.    If you're thinking that assist-to-turnover ratio is a better stat than assists or turnovers alone, one can argue it can be just as misleading.  Do other players have worse assist-to-turnover ratios than Iverson?  Of course.  But point guards obviously have so many more opportunities to get assists than other positions.  It's hard for someone like Dwight Howard to rack up a high assist to turnover ratio, as he's not running the offense and he constantly commands double and triple teams.  He's going to lose the ball almost as often as he dishes out assists.  Unlike Iverson, it's not his job to run the offense efficiently.  There is a double standard there, however, as guards are put in a position to commit a lot more turnovers than centers.

As Iverson's athleticism wore down and he became even less efficient, no NBA team wanted him anymore.  And that's why he's in Turkey.  His career is eeringly similar to that of Stephon Marbury and Steve Francis.  All three were looked at as highly talented and gifted scorers, but in the end, all were fairly ineffective and harming their teams as often (if not more often) as they were helping them.

Team Stats

Let's move away from individual stats for a second.  Judging teams based on their stats can be deceiving, as well.  As an example, a team that ranks either last Total Offensive Rebounds isn't necessarily doing anything wrong.  The San Antonio Spurs are usually ranked among the lowest in this stat category.  Are they a bad team?  Obviously not.  Well, try to put two and two together.  They are known as a great defensive team.  Because of this, they opt to run back and sacrifice chances to get offensive rebounds in order to get their defense set up as soon as possible after a shot.  On the contrary, think of great offensive teams - during the "Seven Seconds or less" era of the Phoenix Suns, they were consistently leading the league in scoring.  But were they the best team?  No, but they were fun to watch.

Ask Tim Duncan: the Spurs don't care about stats

Now you can ask, why do we even look at box scores?  Because aside from watching entire games and understanding the dynamics of basketball, stats are the next best way to judge how a team or player performed.  While one stat is not nearly enough to tell the story, multiple stats can help tell some of the story. Like with all things, basketball is a complex game that's way beyond numbers. Unfortunately, most things in life cannot be judged by stats alone.  You can argue that ALL stats are meaningless.  But that's the point - take any stats with a grain of salt.  Instead, use them as tools and read between the lines before passing judgement or making decisions.

Thursday, January 27, 2011

Hollinger's Stats

As an engineer in the space program, I use numbers a lot.  Numbers are a great basis for evaluating many of NASA's launch operations.  We use numbers to make judgement calls on several technical issues.  We use numbers to quantify how "well" every component on a rocket is performing.  Whether it's structural data to see how vehicle components are holding up, hot fire data from an engine, or flow rates of cryogenics and gases between vehicles and ground facilities, the numbers we analyze thereof are constantly used to monitor how things are going in a variety of operations.

Basketball is deeper than just numbers.  That still holds true.  But lately, the value of using numbers to come up with statistical analysis has improved considerably - mostly thanks to John Hollinger.  Using numbers has improved to a point where some stats have become virtually meaningless.  For example, teams used to use the +/- stat to get an idea of how players affect team success.  Now, +/- numbers can be seen as misleading. Numbers are now being used liberally by every team in the league to come up with game plans against opponents.  Numbers are being broken down to a point where players make judgement calls on how to guard certain players.  As an example, players now have statistical proof (to anchor watching lots of film when studying opponents) as to why one should force Jerry Stackhouse or Jason Terry left instead of letting them go right.  Numbers are used more and more in TV coverage of the games, as well.  When you watch the Magic on national TV, you see always see Jeff Van Gundy or Mark Jackson break down Dwight Howard's field goal percentages depending on how far away he is from the basket:  at 0-6 ft, 6-12 ft, and 12+ ft away from the basket.  It's truly remarkable how much has been put into numbers in today's NBA.

Whatever tools John Hollinger has to crunch numbers, I certainly would love to get my hands on those (it's gotta be something better than Excel, although it wouldn't surprise me if that's all he uses).  One of the most valuable player statistics that most people are aware of is Hollinger's Personal Efficiency Rating (PER).  One of the things that makes the numbers accurate is that playing time is well accounted for (unlike the potentially misleading "per-48 minutes" stats).  This also seems to filter out the "gunners" of the NBA, as historically, guys like Allen Iverson, Stephon Marbury, and Steve Francis were rarely too high up on the PER ratings despite their huge numbers.  Efficiency is truly overlooked when glancing at the" regular" NBA stats, so it is truly nice to be able to crunch numbers in such a way to truly define a player's value and filter out the deceiving inefficient players.

How do these numbers work so well?  I wanted to point out some of the stats he compiles which the PER is based off of.  Here is the formula below:


  • Players on pace to play 500 or more minutes
  • *To qualify: a player must have played 6.09 MPG.
  • TS%: True Shooting Percentage - what a player's shooting percentage would be if we accounted for free throws and 3-pointers. True Shooting Percentage = (Total points x 50) divided by [(FGA + (FTA x 0.44)]
  • AST: Assist Ratio - the percentage of a player's possessions that ends in an assist. Assist Ratio = (Assists x 100) divided by [(FGA + (FTA x 0.44) + Assists + Turnovers]
  • TO: Turnover Ratio - the percentage of a player's possessions that end in a turnover. Turnover Ratio = (Turnover x 100) divided by [(FGA + (FTA x 0.44) + Assists + Turnovers]
  • USG: Usage Rate - the number of possessions a player uses per 40 minutes. Usage Rate = {[FGA + (FT Att. x 0.44) + (Ast x 0.33) + TO] x 40 x League Pace} divided by (Minutes x Team Pace)
  • ORR: Offensive rebound rate
  • DRR: Defensive rebound rate
  • REBR: Rebound Rate - the percentage of missed shots that a player rebounds. Rebound Rate = (100 x (Rebounds x Team Minutes)) divided by [Player Minutes x (Team Rebounds + Opponent Rebounds)]
  • PER: Player Efficiency Rating is the overall rating of a player's per-minute statistical production. The league average is 15.00 every season.
  • VA: Value Added - the estimated number of points a player adds to a team’s season total above what a 'replacement player' (for instance, the 12th man on the roster) would produce. Value Added = ([Minutes * (PER - PRL)] / 67). PRL (Position Replacement Level) = 11.5 for power forwards, 11.0 for point guards, 10.6 for centers, 10.5 for shooting guards and small forwards
  • EWA: Estimated Wins Added - Value Added divided by 30, giving the estimated number of wins a player adds to a team’s season total above what a 'replacement player' would produce.
Derrick Rose could use some more talented off-guards
than Keith Bogans (#6 above).

The last metric in this formula, Estimated Wins Added (EWA), is particularly interesting.  If you sort that column on Hollinger's Stat home page, you find some really interesting numbers. It comes in handy to find out the most value-added players, as well as the players who drag your team down. For example, I see tons of complaints about Keith Bogans being the NBA's worst starter.  If you look at the EWA and sort it by lowest EWA value, you see Bogans as the 2nd worst in the league.  According to this number, he's losing 1.5 games per season.  I understand the coach's argument for wanting to bring scorers off the bench and starting good defenders, but are Ronnie Brewer or CJ Miles worse defenders than Bogans?  I have my doubts.  There are many cases where teams benefit from bringing scorers off the bench (e.g. Dallas this year backing up DeShawn Stevenson with Jason Terry, Denver last year backing up Aaron Afflalo with J.R. Smith).  But I have a hard time seeing what value Keith Bogans has in Tom Thibodeau's eyes.

D-Rose could use more offensively talented off-guards than Bogans

I also highly recommend looking at the team stats and Playoff Odds.  Halfway through the NBA season, these numbers resemble the big picture of the NBA season pretty darn well.  Unfortunately, there aren't enough games in the playoffs to make the numbers work.  Plus, the playoffs are a totally different game;  the big disparity is in the Western Conference, where every round is a heavyweight slug-fest.  In the East, the top 3 teams have incredibly inflated numbers feasting on the bottom-feeders of the East.

It's amazing how much numbers can tell the story.  But are Hollinger's statistics an exact science?  Of course not.  No matter how advanced numbers get, basketball will always be unpredictable.  And that's the beauty of it.