Statsketball: Lancing Free Agency

In the final edition of this series of Statsketball -- in which I continue to discuss the biggest names of free agency this season and how they stack up against each other and for the Dallas Mavericks statistically -- it's finally time to look at the guard position.

Dallas is fairly desperate for another guard, which seems weird, given that they have both of their starters locked up under contract for at least two more years, and they have a very good backup for both guard spots ready to be re-signed in Devin Harris.

Unfortunately, the rotation has turned out to be completely untenable, especially long term. Harris was very good this year, shooting percentages aside (the team scored way more with him on the court vs. off, so take that how you will), and even per 82games.com he was a very good defender when he was part of the primary defensive action.

Still, Jose Calderon was even worse on defense than we thought he would be to begin with -- and we came in expecting him to be very bad -- and Monta Ellis was about as bad as we thought he would be.

The truth is, Monta and Jose are both great players, and with the right players around them they can be amazing, but having both on the floor together is an unmitigated defensive sh*tstorm. Devin Harris, while a solid defender, just isn't anywhere near enough to clean up the mess created by either or both.

The Mavericks badly need another guard who can really defend. The interesting thing about this particular need, though, is that the only stipulation needed to cover this role is "defensive guard." The player can be starting caliber -- pushing either Jose or Monta to the bench -- or a 10-15 min per game kinda guy. He can be primarily a guard, or a versatile wing who can play guard or forward minutes.

It really doesn't matter who it is or what in particular he specializes in, as long as it's someone who is good enough defensively to take some of the onus off of Dallas' big men and poor old Devin Harris to clean up a mess that's completely unclean-able.

There are, luckily, a few people who fit that description. Kind of.

This is the most interesting position in terms of the market, specifically, since the variation on the type of player that the Mavs need to sign is so high. With the forward and center positions, Dallas really needs a starting caliber guy, probably costing them between $8-$12 million a year. They can probably try and save a little money and get a facsimile of the type of player that they really need at those positions, but nonetheless that is fundamentally the need.

With the guard, though, the Mavs can do anything from maxing out Lance Stephenson to taking Thabo Sefolosha with the minimum, which opens up a world of possibilities with the other positions, and it makes what they choose to do this offseason that much more interesting.

One final reminder that I'm using this space to only discuss unrestricted free agents in-depth.

Lance Stephenson

The numbers don't actually indicate that Lance is a particularly great defender

As everyone is keen to point out, Lance will inevitably be a very risky signing this offseason. He'll be given a lot of money that he may or not deserve, and he may or may not implode.

His appeal, though, is threefold: he's a solid to very good scorer on a team with no offensive system and no real spacing, which offers a lot of potential; he's very young for an unrestricted free agent, which offers potential and growth; and he's, supposedly, a voracious and hard-toothed young defender.

Unfortunately, the last bit may not be entirely true. Or, at least, it's not corroborated by the vast majority of stats database that's out there right now. I tend to be a little tentative using defensive stats, since defense is so hard to quantify, but it's probably a concern if almost no major measure of defensive strength (that I'm aware of) is on your side.

Per NBA.com, Lance had no noticeable impact on the Pacers defense. Indiana defended at a rate of 96.8 points per 100 possessions allowed with him on the court, and 96.6 with him off.

Lance's DRPM (Defensive Real Plus Minus, ESPN's new stat that tries to track a player's defensive +/- impact on the court without the interference of the stats of the other players on the court) is a paltry 0.15, while his DRAPM (the same thing from a different outlet and calculated slightly differently) is actually -0.44. His impact, per those stats, is either negligible or vaguely damaging.

Per Synergy, too, Lance is a purely middle-of-the-road defender at best, ranking 175th in the league at 87 points per 100 possessions allowed.

Most concerning is that he struggled most in the places that guards, in particular, have to be good at: guarding the ball handler in the pick and roll, defending in isolation, and sticking to shooters. Lance ranked 200th in the league defending the ball handler in pick and rolls and defending players in iso, at 89 and 91 points allowed per 100 possessions, respectively.

Spot up shooters scored 94 points per 100 possessions, too, which is better but still not great (running the ball in to score off of a pick and scoring in isolation are inherently less efficient, which is why a stingier mark on those plays than spot up shooting is still "worse" defense).

On the bright side, he did well chasing down players looking to shoot off of screens, limiting them to a very stingy 80 points per 100 possessions.

There are two bright sides here, though: one, 82games.com registered that when he was directly involved in a play he kept opposing players to a piddling 11 PER. Unfortunately, Lance was also more often than not the head of bench units, and the shooting guard position is weak anyway, meaning that Lance kept opponents who are already well below average to a well below average mark. It's hard to know how much he actually impacted opponents on defense from this measure, but it's a good number anyway.

The second bright spot is that none of his other defensive stats are really bad, per se, either. They're just not impressive. There's no statistic that says that Lance meaningfully hurt his team on defense, he just didn't help them much. For a scoring oriented 23 year old guard, being able to impact the game defensively at any point is a plus, and is not necessarily expected of him.

That everybody has also pointed out that Stephenson approaches defense with a vigor, energy, and excitement is similarly encouraging. It's more than possible that he cares enough to put int he effort to get better on that end of the floor. He has the time, too.

Nonetheless, perhaps it's best if you don't think of Lance as a 23 year old defensive stopper. He's not. He'll be better than everyone else that the Mavericks have had on defense, but that's not necessarily saying much, and he probably won't be enough to plug the hole in the dam.

Lance may not have world-ending upside, but he'll be pretty good

Lance's primary utility, at this point, is that he's a good scorer. The Pacers scored a whole five points per 100 possessions better when he was on the court than when he was off, and he averaged 13.8 points, 7.2 rebounds, 4.2 assists (and an unfortunate 2.7 turnovers) in 35 min per game, but with a usage rate of only 19% (below average), a true shooting percentage of 56.4% (above average) and an impressive assist rate of 21.7% for a scoring wing.

The real question for Lance is whether he'll always be a good third guy or if he can grow to be a huge impact player, and that's a question that I certainly can't answer, but I can look at.

I ran some very basic projections to see what kind of player Lance might grow to be in the next few years. First, I took a very crude measure of Lance's general production in every month where he played significant minutes for the last four years (I used an aggregate of points, rebounds, and assists weighted for efficiency and minutes).

Using that data, I ran a regression on the data to see how, roughly, his production has grown over the last few years. For you math types, I used a logistic regression (what I consider to be the best model for individual growth anyway, in a vacuum) with an R^2 of roughly 80%.

I then used the relationship that I'd found to project what Lance's PER might be over the next few years.

I'm making two major leaps of logic, here: first is that Lance's role on the Mavs will be roughly equivalent to the role he had on the Pacers, considering that he'd likely be the third option (behind Dirk and Monta) playing 35ish minutes per game.

The major difference, as I saw it, between his role on the two teams is that the offensive options other than Lance (Jose and whoever is on the wing) will be less ball dominant, allowing Lance to use a greater percent of the team's possessions than he did in Indiana.

The second major leap of logic is that my basic calculation of points, assists, and rebounds that's weighted for efficiency will be very similar in its growth to PER.

The results of the projection were as follows:

Great, but nothing spectacular, which isn't a major surprise. Lance is a very good player who has some major flaws, including some major gunner tendencies. Remember, his PER this season was only 14.7.

Then, including this data and what we know about the Millsap Doctrine and the relationship between true shooting percentage and Usage Rate, I predicted what Lance's typical box score might look like with the Mavericks.

Again, nice, but not revolutionary.

As a reminder, though: Stephenson could very easily prove this all to be entirely wrong. This is crudely calculated, but it does provide an interesting look into, at least, the path that he appears to be on thus far.

Finally I looked at the growth of his defense. Using data from basketball-reference, I did the same calculation that I had done to project PER to project his defensive efficiency for his next season.

It occurred to me, though, that defensive system is much more important than individual defense, and that good defensive players on bad defensive teams tend to perform far worse.

So, to compensate for that, knowing that the Mavericks are a horrid defensive team, I used basketball-reference to average the difference in defense among many players who were once on good defensive teams, and then later played on bad ones, or vice versa.

I used that data to weight what Lance's defensive rating would likely be on the Mavericks, given some degree of improvement.

The results:

The difference between Lance's rating on the Pacers vs his rating on the Mavericks would be roughly the difference between a very good defender to an average one.

Again, none of this is set in stone, and it mostly just describes the path that Lance is on at the moment. The appeal of a player so young is that they have so much time to get better in ways that no one expects, and that's why it's always so exciting to project what we think those players can be.

That said, this data is not necessarily discouraging, but it's not a ringing endorsement of a guy who might command a max deal.

Thabo Sefolosha

Thabo's entire utility is as a defender, and he's good, but he might not be the same player anymore.

It seems weird to be going from "free agency goldmine Lance Stephenson" to "minimum or midlevel player Thabo Sefolosha" as the main answer for the Mavs' backcourt defensive woes, but here we are.

The big problem is that -- despite the Mavs' great flexibility with how they go about addressing this need -- there just aren't many players who are particularly interesting or helpful in this free agency class, and that's a problem.

Thabo's entire billing as a player, and the entire reason that he's still in the league, is that he's supposed to be a really, really good defender who can occasionally knock down some threes. It is not apparent, however, that he can still do either of those things.

His slippage defensively is obviously the biggest concern, because it's literally all that the Mavericks would need Thabo to be able to do.

Like Stephenson, per NBA.com, Thabo has absolutely no impact on the team's defense. Despite starter's minutes, the Thunder defended at the same 101 points allowed per 100 possessions mark with Thabo on the court as they did with him off the court, and that's a major concern. Also, per Synergy, he's a bad defender overall, registering a worrisome 94 points per 100 possessions allowed, seven points worse than Lance.

Here's the bright side, and it is noticeably bright: per Synergy, Thabo is very good at the things a guard's supposed to be able to do, ranking 70th and 50th in the league at defending isolations and the ball handler in the pick and roll at 73 and 71 points per 100 possessions allowed, respectively.

Plus, in contrast to his apparent lack of impact to OKC's defense on NBA.com, ESPN's RPM -- which, again, adjusts for teammates -- indicates that his impact on defense is quite positive, at a score of 2 points per 100 possessions, a distinct improvement. RAPM, too, has his defense impacting the team to a tune of 1.19 points per 100 possessions, also a distinctly positive measure.

The takeaway here is that Thabo is not a Tony Allen type anymore, not the kind of guy who can shore up a perimeter defense on his own. He is, though, good, and he can go a decently way towards improving the Mavericks' porous backcourt.

The real concern, I would think, is that he seems to have lost his shooting touch. Sefolosha shot a way below league average 31% from three this last season, and he can't really do anything else offensively. He kills OKC's offense, they're significantly better offensively with him off the floor than when he's on. It was so bad that the Thunder couldn't play Thabo at all in the playoffs this year.

It's possible -- probable, even -- that his shooting won't really matter on this Mavericks team with offensive weapons galore for 10-15 minutes per game, so he could still have a ton of utility. In fact, his shooting woes could be a blessing that drives his price down while he stays useful to the Mavs.

Nonetheless, it is a head scratcher that's worth keeping in mind.

Final Notes

• Avery Bradley, a restricted free agent in Boston, is the biggest name other than Lance that I've heard thrown around for filling the Mavs' big "defensive guard" hole. To be honest, though, I have no interest. Bradley is pretty decidedly overrated: defensively he can wreak havoc by being active and hounding ball handlers early in a play, and is a great on-ball defender, but overall he has his serious defensive warts. He gets lost on his second or third rotation quite often, he struggles knowing when and how to hedge on a pick and roll, and like most over-active defenders he gets flattened by picks way more often than he should. He's smart, and he's probably roughly the equivalent to Thabo Sefolosha defensively in the end, but given that he's had more shots than points on half of his seasons so far (an astoundingly hard thing to do, and a damning thing to have on your resume), Bradley may actually be a worse scorer than Sefolosha. Don't get too encouraged by his solid three point shooting numbers, since he doesn't take nearly enough of them to make up for his offense elsewhere. Don't get me wrong, Bradley could be good player, and he'd be a better get than Thabo when you take upside into account, but Boston is really, really high on him, and he wouldn't be worth what it would take to get him.

• Mario Chalmers is another name that might get thrown around. For the entire run they've had so far, he's been the forgotten member of the squad, and might easily get ignored this free agency, despite frequently being a killer offensive weapon for them. He'd be a brilliant backup guard to soak up some minutes, like Darren Collison with a shooting touch and a little less "run and the basket and pray." Unfortunately, he's also not a very good defender, and he probably wouldn't be any better at assuaging the defensive issues than Devin Harris has been. Plus, he's really, really annoying.