clock menu more-arrow no yes mobile

Filed under:

Kentucky Basketball 2008: Looking at "Floor Value"

Btn_ball_medium Btn_stats_medium

Whenever I get into a basketball season, I start questioning authority.  I usually reconcile myself to dealing with the conventional wisdom, but sometimes I feel unsatisfied with what conventional basketball stats tell us.  So I find myself trying to invent new measures to describe trends that I think just might be relevant.

Today, I have invented a new "stat" called "Floor Value."  Floor value is a weighted combination of several statistics:  Offensive rebounds, turnovers, steals and assists.  Why did I do this, you ask?  Well, I got to thinking -- what is more valuable, a steal or a turnover?  Generally speaking, steals lead to a possession change in which the defense is seriously disadvantaged -- i.e. they are set up for offense, and then possession is changed creating a transition play which favors the team that just got the steal.

Case in point -- on a turnover resulting from a violation, for example a walk, a palm, a double-dribble, 3-seconds, etc., the defense always has a chance to set, forcing the offense into the half court.  On steals, the reverse is true -- the defense has to hustle to make a play, making scoring easier for the stealing team.

Additionally, I considered offensive rebounds.  For some reason, the fact that the formula for calculating possessions from box scores always bugged me in that they consider an offensive rebound a continuation of the same possession.  Now, that does make sense, except when you consider a non-trivial number of OR's occur very near the basket and result in a point-blank score.  An offensive rebound, therefore, would seem to increase the chances of a successful possession significantly.

Putting all this together, what I have done is essentially multiply by a fudge factor to each of the statistics where there is added value in whichever direction.  In the case of turnovers, I calculated the percentage of turnovers resulting from steals, averaged them, and adjusted the value of turnovers accordingly.  This methodology is somewhat arbitrary, but produces an interesting result.  What this really is intended to measure is who is doing the "little things" that add significant value to the team effort, but are hard to quantify specifically.

So let's look at my new creation, and see what it tells us:


Name OR/40 Steal/40 TO/40 Asst/40 Floor value
Patrick Patterson 2.44 1.83 2.14 4.58 7.32
Perry Stevenson 4.23 1.15 1.92 2.69 6.88
Josh Harrellson 5.22 0 0.87 1.74 6.78
Darius Miller 3 1 4 3.5 3.79
Michael Porter 1.58 1.58 5.26 5.26 3.25
DeAndre Liggins 0 2 8 8.5 2.18
Ramon Harris 2.14 1.79 4.64 1.43 1.01
Jodie Meeks 0.57 1.43 4.86 1.71 -1.2
A.J. Stewart 3.33 3.33 13.33 0 -6.7

What I have done here is calculated each statistic as a "per forty-minutes" stat and multiplied each value-variable statistic such as steals, turnovers and offensive rebounds by a fudge factor.  Assists are then simply added to the resultant calculation (turnovers are obviously negative) and the result is the floor value derivative statistic.

What is interesting about this is that big men have an advantage in this statistic, primarily because of two factors: They handle the ball less and are therefore less susceptible to turnovers, and they get more offensive rebounds.  I have also discarded any players who have less than 10 minutes played so far in the season.

What I wonder is why players like Meeks and Liggins are so low down the totem pole.  In Jodie's case, it is two major issues --  he has little value as an offensive rebounder and he turns the ball over a lot.  Worse, Jodie is not a particularly effective defender when it comes to steals -- Michael Porter, DeAndre Liggins and Patrick Patterson all get more steals than Meeks.  Meeks' overall floor value is far and away the lowest of any starter, or for that matter, significant bench player.  I actually think this is a problem that needs some attention.

As far as Liggins is concerned, he has no offensive rebounds at all this year, which is not really a bad thing for a point guard -- his main responsibility is to provide defensive balance, not rebound.  Liggins' main fault is an inordinately high number of turnovers, by far the most on the team other than A.J. Stewart, who we will discuss later.

Patrick Patterson's stats are interesting in that his assists vs. turnovers is extremely high, the highest on the team.  Of course, we all knew that Pat was a very good passing big man, but just how good is illustrated here.  Where Pat is not getting the job done is on the offensive glass.  Perry Stevenson, though, is picking up that slack, almost doubling Patterson's OR output.

Josh Harrellson is a bit of a surprise in the higher reaches of this stat, but he is there primarily because he has turned the ball over so little and leads the team in offensive rebounds/40 minutes played.  It is somewhat significant, though, that Harrellson has the lowest minutes played of anyone considered other than Stewart, and most of his time has been garnered against inferior competition.

Another surprise is how high Darius Miller is.  He is contributing more overall in these "hustle" stats than any other non-front court player, and now we see why Gillispie has been so high on him

Michael Porter also surprises a bit -- he is just below Miller on the wing-player totem pole.  So while we have all been hard on Porter, in reality, he has only truly struggled in the North Carolina game.  The other side of that coin is that Porter has yet to prove himself against high D-I competition.

A.J. Stewart is last on the list, and the reason is obvious -- turnovers.  A.J. has an unacceptably high turnover rate per 40 minutes, and if he expects to get on the floor, he is going to have to fix this broken statistic.  According to my measurement, both he and Meeks subtract value with their floor game, but Meeks offsets that by scoring lots of points.  A.J. does not.  To be fair, though, A.J. barely makes the cut minutes-wise with 12, and his relatively small number of stats are exacerbated because of that.  In other words, it looks worse than it probably is.

So there we go.  I have invented a stat that shows ... well who knows what, but it does seem to have some significance in the areas I consider important.  Patrick Patterson and Perry Stevenson have inarguably played very good basketball.  Ramon Harris winds up farther down this list than I expected, but that is because of his high turnover rate and relatively few assists.  Harris does many good things that don't show up in the stat sheet, like forcing turnovers and getting deflections.  My perception is that he does these things better than anyone, but there is no way for a layman like me to measure those intangibles.  Jodie Meeks is offsetting some of his great work scoring by doing the little things poorly.  Those are the extremes, and everyone else is somewhere in-between.