# Data and Statistics – Another Example (Sports)

## Data and Statistics – Another Example (Sports)

/

B: Statistics may help sports be a lot more interesting than they otherwise would be, but one runs into the same kinds of problems here that were mentioned in previous examples.  In fact, the problems may be even more egregious in sports – but thankfully, they’re only sports.  With sports statistics and the interpretation thereof you’ve got people saying outlandish things like Chris Paul potentially being the greatest point guard of all time.

First, however, a bit of historical reflection.  The biggest sports themselves were not invented with any statistical measures to be made in mind; fans, coaches, and others came up with these later on, beginning with the raw data one can take down simply by observation, and later with more technical analysis of the raw data.  One has to remember that the raw data themselves were tallies of things that are easy to see and measure, and may not necessarily be indicative of what contributes the most to a team’s success.  Why remember that?  Because what some have begun to do with statistics is say that the better players have the better tallies or combinations of raw data.  This is basically another instance of correlation implying causation.

For instance.  Let’s take basketball – might as well; I referred to it before – wherein points, rebounds, and assists are fairly easy to keep track of.  But how much do they really contribute to the team’s success?  What would most teams call “success?”  Whether they win the championship, which is most easily done if they win a relatively large proportion of their games.  To win the game, you must score more points than the other team.  Thus it would seem the “points” statistic is an important one.  But anybody worth their salt as a basketball fan knows you can’t just say the best player is the one who scores the most points (even though it may be true in some cases).  “Any knucklehead can score,” quoth Charles Barkley, and I believe he was referring to the fact that any player can take a large volume of shots, giving himself a good chance to score quite a few points.  But the question really comes in efficiency, as statisticians have recognized not only with points, but with statistics in general over the years.

How about “rebounds?”  That refers to the first seizure of the ball after a missed shot, which of course gives a team an advantage because they gain possession of the ball and thus have the next opportunity to score.  But how many rebounds are really that advantageous?  How often do teams simply concede a rebound to the opposition, content without trying terribly seriously at securing it themselves, and try to get back on D instead?  One player might pile up truckloads of rebounds, but perhaps most of those have been conceded to him, not only by the other team but by his own teammates, simply because he happens to be the one closest to the ball.  The real rebounds are earned; i.e., when many fight for it.

Also, it turns out “assists” are somewhat of a judgment call made by the statistician himself, about whether the person who scored was genuinely assisted by the person who passed the ball to him.  But it could be only by the person who passed it to him, since that’s one pre-defined rule about assists.  Never mind whether the scoring player had a teammate set a screen to spring him free; the player setting that pick gets no credit for any assist, even if it was his screen that primarily allowed his teammate to score.  The only player that can get the assist is the player who just happens to be the one who last passed it to the scorer; no previous pass or any other action by another player is recorded as contributing to the assist.  The “judgment” part comes in deciding whether or not that last pass truly helped the player score.

An example of the inaccuracy of this latter statistic comes from Game 4 of the 2014 playoff series between the Miami Heat and the Brooklyn Nets.  Late in the game, with the Heat having the ball and the score tied, LeBron James gets the ball in the lane, drawing a double team, because of the great scoring threat that LeBron is.  Instead of forcing up a short shot between two players, he passes to a wide-open Mario Chalmers at the three-point line, and when another defending player had to come off Chris Bosh to guard Chalmers (because of the double-team on LeBron), the latter passes it to Bosh for a wide-open three, which he nails.  The Heat go on to win, with that particular basket playing a key role down the stretch.  See the entire play here.

But the emphasis I want to make comes in how the play was scored in the statistics.  Bosh gets three points, an improvement on some shooting percentages, and Chalmers gets an assist.  LeBron, however, gets nothing.  As far as the statistics are concerned, he may as well have been on the other side of the court getting a head start on defense.  And yet it’s indisputable that LeBron had a key role in that play – in fact, it’s arguable that the play was made primarily by him, since he drew the second defender and found the first open guy, who was easily able to pass it to the second open guy.  But no statistics give LeBron credit for anything whatsoever on that play.  It’s not that what Bosh and Chalmers did on the play was trivial, it’s that the statistics that are recorded make it seem like those two, and those two only, made the play, which is simply inaccurate.

I should mention that there perhaps are some relatively little-known analyses some statisticians – many of them working for NBA teams – have made about efforts like LeBron’s on the aforementioned play.  Such an analysis may make mention of, say, how often one of LeBron’s possessions of the ball leads to points on the play.  That, I would say, is a step in the right direction; however, there are still difficulties such as determining exactly how much of an impact the action makes on the play.  It’s this same kind of difficulty that should be accounted for when trying to quantify certain coefficients, or weights, for some raw data in the more technical statistics.  How much weight, for example, should be given to points?  Or to rebounds?  Or to assists, etc.?  There are attempts at ascertaining these coefficients, but how accurate are they, really?

Here’s another example: a sharp-shooting player is thrown in the game as a possible decoy, and he sits around at the three-point line.  The guy shoots perhaps more than 50% from three-point range if he’s left wide open, which is as valuable as shooting 75% inside the three-point line, a high percentage indeed.  So his defender would be unwise to just leave him open to, say, go give help defense.  So by standing out in three-point land, the sharpshooter has drawn his defender away from helping out on defending other players, i.e., he has effectively “stretched the defense,” thus reducing the game from 5-on-5 to 4-on-4 merely by his remarkable ability to shoot.  With all that extra space, it may become significantly easier for his teammates to score.  Coaches know this and use the practice regularly.  But “stretching the defense” is not a statistic that’s recorded in any box score, nor included in any technical analysis of the raw statistics that are recorded that I know of.

Well, so these examples raise more questions, like how often do things like this happen?  How often are the really valuable statistics, i.e., the “credits” for helping the team win the game, given to the players who deserve it?  How many other actions in the game are being undervalued, or even completely overlooked and not counted as statistics at all?  Here’s a short and incomplete list – might each of these things, and perhaps many more, still make a difference to the team’s play?

• Screens set by players
• Help defense being applied
• Difficulty of the shot taken, including consideration of defenders present, offensive and defensive schemes, trajectory of the shooter’s body’s motion, venue, time left in the game, score of the game, importance of the game, etc.
• Fulfilled assignments, which could include certain cuts made at certain times (on offense) or guarding certain players or working schemes correctly (on defense)
• Times playing through pain, whether physical, emotional, mental, etc.
• Other difficult inconveniences taken by players, both on and off the court
• General attitude and professionalism
• Leadership and encouragement to teammates (coaches seem to think it’s important, but how in the world do you measure that?!)

Let me concede the fact that most statisticians are aware of the shortcomings of all the statistical measures that are used, and they usually admit that there isn’t really an all-inclusive stat that differentiates between players, or even teams.  They’ve done, I’d say, a remarkable job of trying to include so many of these other details.  Bless their hearts for trying; but I’m afraid that their statistics still fall woefully short of painting a genuine and precise picture of the rankings of players’ or teams’ overall abilities.  The bottom line is, they cannot be relied upon to give an irrefutable assessment of a player’s or team’s ability; they can only make helpful suggestions.