(I'm a few days late to the party, but I guess I'm looking at this from the opposite perspective...)
Neil Paine of Basketball-Reference.com is one of the few wunderkind basketball stat-heads capable of conjuring up a sufficient response to just about any inquiry possible. As such, should a reader of his blog drop him the casual basketball writer's equivalent of an NC-17-rated question -- in this case, "Who are the most one-dimensional players of all time?" (I'm pushing PG-13 on this hypothetical scale) -- you can expect Neil to come through with a few fancy charts, a solid explanation for the data and an accompanying poke at Antoine Walker (okay, maybe that was just this time).
In his post addressing the question, Paine posted multiple charts assessing the most one-dimensional players post-1952 using win-shares. Paine prefaces the post with the following:
I should note, one issue with using Win Shares is that they can sometimes be negative, which totally wreaks havoc on an exercise like this. But as a kludge, I just zeroed out the negative OWS/DWS, and took the percentage of those totals devoted to each side of the ball.
Everyone onboard? Nope? Great!
Actually, it isn't that difficult to follow. In short, we're discussing how much a player impacts a certain side of the ball (defense) as compared to the other side (offense). At first glance, I saw a few names that made sense in the "Lots of Offense, Little Defense" table. Kiki Vandeweghe was someone I knew to be offense-oriented, and Steve Nash, Jose Calderon and Eric Piatkowski weren't surprises, either.
Then Kevin Martin's name popped up in the top ten. Historically, according to the chart, Kevin Martin has the third-lowest defensive win-shares percentage of all time among players with 10,000 career minutes. 83% of Martin's win-shares are accounted for on offense, while only 17% are defensive. Piatowski actually beat him by a percentage point, with 18% accounting for defense.
Among the thirty players listed in the "Offensively Biased" list, Martin has played the fewest games, but 10,000 minutes is a fair cutoff when it comes to sample size. So, what does this mean? Is Kevin Martin the third-worst defender among the long list of players who have logged 10,000 career minutes?
No, not at all.
Rather, my interpretation of this finding is that while Martin plainly does not contribute much on the defensive side of the ball, we are...
A. Looking at a fairly team-oriented statistic (in other words, Martin's defensive WS aren't entirely based upon his individual play), B. Comparing these statistics with players on different planes of productivity, and C. Merely looking at DWS as compared to OWS. On top of that, they are percentages of a whole, not averages or totals.
If you're still confused (or if I'm just needlessly confusing you further), look at Tyronn Lue and Kevin Martin. Lue has a 88% OWS, while Martin has 83% OWS. However, it's fairly obvious that Martin is a far more effective and accomplished player than Tyronn Lue, but in Paine's study, we're only looking at offense compared with defense, and Lue's defense is that bad. Make sense? In other words, were Kevin Martin not so good on offense, his percent-DWS would be much higher. For those of you that understood this immediately, I do apologize.
In assessing Martin's defensive impact on the Rockets this season, I went to 82Games.com, jumped to the Rockets team page and looked at the "5-Man Units" bar. This essentially shows which units are more productive than others. When judging which units are better or worse than others on defense, Martin's name popped up just about everywhere.
In a few of the units Martin has played with, the Rockets defense wasn't any good.
Lowry-Martin-Battier-Hill-Miller allow 1.47 points per possession.
Brooks-Martin-Battier-Scola-Yao allow 1.31 points per possession.
Brooks-Martin-Battier-Scola-Hayes allow 1.30 points per possession.
But in other units containing Martin, the defense performed much better.
Smith-Lee-Martin-Hayes-Hill allow 0.74 points per possession.
Smith-Martin-Battier-Scola-Miller allow 0.96 points per possession.
Lowry-Martin-Battier-Scola-Hill allow 0.98 points per possession.
Take what you will from this. Martin still isn't a very good defender, but he's not historically bad. Perhaps I skewed this out of the realm of interest, but at its core, it's a nice look at how one-dimensional (in the broadest terms, "offense" vs. "defense") Martin's game really is.