Sunday, April 5, 2015

Don't Miss It

Are you having fun yet?

The Badgers are headed to the title game, and once again the story is Offense vs. Defense. Duke's defense has nothing—nothing—on Kentucky's. But today Jahlil Okafor is preparing for what's likely his final college game while Willie Cauley-Stein obviously has some other important business to attend to. And the reason Duke's been so successful in the Tournament has been the unexpected play of the defense, which was not a strength in the regular season.

Wisconsin's Defense vs. Duke's Offense

Of course billing this game as "Offense vs. Defense" seems silly considering we're talking about two top-3 offenses. Duke's offense goes through Okafor, the most devastating two-point weapon in the country. I suspect the Devils will go to him early and often, and Frank Kaminsky will need to be at his best to avoid fouls.

Okafor's sidekick in the offense is Justice Winslow, who looks more like a Wisconsin big man in that he's just as comfortable shooting 3s (making 42 percent of them on the season) as he is scoring in the paint. He's also a plus ballhandler, so he's not easy to pressure into mistakes.

But I'm optimistic that Wisconsin will be able to slow this Duke offense just enough. A team that shares a lot of the same defensive principles (although that team was much better on that side of the ball this year) is Utah. Both teams are very good at forcing opponents to settle for mid-range jumpshots, although the Utes tend to challenge shots more aggressively (with the expected quid pro quo showing up in the rebounding). It might be no coincidence that Utah was the only team that held Duke to under a point per possession in the Tournament.

I expect Wisconsin to pretty much give Okafor whatever he wants outside of 8-10 feet, with some weak side help defense if he establishes position in the post.

Wisconsin's Offense vs. Duke's Defense

On the other side of the ball, where Duke has transformed itself from a middling unit to an elite defensive team—let's look at what's changed. For one, we can probably rule out any philosophical about-face—Mike Krzyzewski has been at this a while, and he's actually been a halfway-decent coach. Also, Duke's opponents are getting the exact same shots now as they did across all games:


For those who are not familiar with the Blue Devil defense, the short version is that the Devils have an extreme (though less so, in recent seasons) overplay defense that gets into passing lanes and extends very far into the perimeter. This approach tends to minimize 3-point shots, and create turnovers (though less so on the latter, in recent seasons). Of course, that leaves the perimeter defender susceptible to a blow-by. So what's changed in the postseason? Well, opponents are simply missing their jumpshots.


It's likely that Wisconsin will attempt a fair number of jumpshots. If they make them, they will win (#analysis). But will those shots be open jumpers that are largely behind the three-point line? For that to happen, Wisconsin needs to rotate the ball.

So far, Duke's defense sounds a lot like the basic framework of the Purdue defense. Both are overplay systems that extend the defense well outside the three-point line. But instead of relying on a JaJuan Johnson or A.J. Hammons swat shots, Duke's system employs rotations that step in to take charges.


This possession (at 3:03, action starting at 3:10) highlights a good defensive outcome using Duke's core principles. With the defense extended so far, the offense cannot get a good outside shot, so the ballhandler easily drives past the perimeter defense. But the rotation is there, in part because instead of sticking with his man in the opposite corner, Tyler Thornton spends most of the possession with both feet in the paint, so he's in position to help.

Thus, the challenge for Wisconsin's offense will be whether it can rotate the ball faster than Duke's defense can help and recover. But the Badgers have an advantage here that most teams do not—almost always, all 5 players on the floor have the green light to shoot a 3-pointer. That means Duke's help side defense will be put in a tough situation, as sticking with the defensive plan means leaving a 3-point shooter wide open.

And it doesn't take much—here's a possession against Purdue. Watch as A.J. Hammons shows help defense just a little too much, leaving Hayes open for a 3 (16:06):


Simply put, the best way to beat Duke's defense is to pass the basketball. Heroball is a surefire way to end up with turnovers and empty possessions. But that's why I like this matchup for Wisconsin, this is a team that passes. It's not a surprise, as one of Bo Ryan's sayings is "if you can't pass and catch, you can't play."

But win or lose, this will be your last chance to see the best offense of the past fourteen (or more) seasons. And it's being executed by a bunch of clowns. If this is what a broken sport looks like, please don't fix it.

Wednesday, April 1, 2015

Who invited these guys?

The Final Four consists of the number 1, 3, 4, and 15th-ranked teams by Pomeroy measures. It's not hard to find the ugly duckling, in that context. But even before the ball was tipped in Michigan State's first round matchup against Georgia, this was a curious team. I'm on record that I was dead wrong about the Spartans' prospects coming into the season. I don't think my outlook was borne in I Hate Tom Izzoville, either. MSU lost its three best players from last year (Gary Harris, Adreian Payne, and Keith Appling), and replaced them with a trio of non-impactful freshmen and a transfer who hit 30 three-pointers in conference play, but didn't do much else (considering Harris, Payne, and Appling combined for 72 three-pointers in Big Ten play last year, that's a deficit). Gavin Schilling and Alvin Ellis were the only two returning sophomores, and while Schilling definitely improved, we're still talking about a severely foul-prone center that has a scouting report that simply reads "Rebounds. Can Dunk." As for Ellis, he was better as a freshman.

Everyone else on the roster was an upperclassmen, which usually means "known commodity." So I don't think I was terribly off-base when I wrote the following: 
Michigan State is not going to compete for the Big Ten title this season, that much is clear. But we’re still seeing a lot of projections putting the Spartans in the top-25, though I’m not sure I can sign onto that, either. For one, I’m not sure where the improvement comes from. The only returning sophomores are Gavin Schilling and Alvin Ellis. Schilling had his moments last year, but he, like Ellis, is an extreme role player that seems unlikely to grab a larger share of the offense. Travis Trice is a three-point specialist who sometimes plays a point guard on TV, though Valentine might actually be better-suited for the role. Valentine and Branden Dawson are nice enough players, but I’m not sure there’s a lot more to see out of either of them. Both players were third and fourth options while on the floor last year, but now they’re expected to be the centerpieces of this attack. While Valentine should improve on his finishing (just 49 percent at the rim last season), the added responsibility will likely lead to more bad shots. As for Dawson, he’s incredibly effective around the rim, where over two-thirds of his shots have come from in his career. But he’s not a jumpshooter. When the shot clock is winding down, he’s not someone I want to see with the ball in his hands at the top of the key. 
And ultimately, that’s the problem I see with MSU’s offense this season. A lot of players who are good in the right role, but all role players to be sure. I see a defense first, second, and third team here, and one that appears headed for the bubble.
Obviously, that's quite the miss on both Trice and Valentine. Trice was probably the best point guard in the Big Ten not named "D'Angelo" (sorry, Yogi), and Valentine was good enough to merit comparisons (at least, offensively) to Draymond Green. Trice's leap was especially miraculous, as he increased his usage by 79% while keeping most of his efficiency (a tall task, given that last year he was well suited to the corner marksman role now occupied by Bryn Forbes). Seniors—especially those not regarded as one of the nation's top-100 players in their high school class—generally do not undergo this kind of transformation. It's probably not going to happen for Bryn Forbes, or Anthony Clemmons, or Joey King, or Benny Parker. Travis Trice is a unicorn that rides the Loch Ness Monster to work after feeding his pet Bigfoot. You will not witness such a thing often.

The massive leaps made by Valentine and Trice therefore explain how Michigan State maintained its conference efficiency margin from last year. But how did these Spartans make it all the way to Indianapolis? Simple—they stopped played like Michigan State:

For years, it's been a struggle for Tom Izzo's teams to keep its turnovers in check. I'm sure that's as much a function of the style of offense the Spartans usually choose to employ (one that traditionally has included a lot of offensive rebounding and fewer 3s) as anything else, though even when the team plays out of character (for instance, by shooting a lot of 3s last year, or in 2004 when the team's offensive rebounding was relatively poor) it's still a struggle. I suspect the fact that Michigan State runs so many set plays also plays a role, but that's mere conjecture on my part.

In any event, even this season the Spartans just barely were better than average in the Big Ten. This is not unprecedented, nor is it a sign that turnover problems are gone. Keep in mind that "barely better than average" are not words usually associated with Michigan State basketball, and certainly not going to strike fear in the heart of a Final Four opponent.

But in their Tournament run, the Spartans have posted a turnover rate of just 13 percent over 245 possessions. Now sure, Louisville's defense isn't up to Pitino standards this season, and Virginia and Georgia generally don't try to force turnovers, but keep in mind the story the graph above shows. Michigan State wasn't playing VCU for 14 seasons, they were playing Average Big Ten Opponent, who usually is not one for a pressing style of basketball. And yet, too many MSU possessions did not end with a shot attempt.

Is this newfound respect for possessions a mere statistically anomaly? Newly-discovered offensive wrinkle? Something as simple as fewer minutes for the turnover-prone Lourawls Nairn? We'll find out Saturday. 

Monday, March 9, 2015

Another One in the Books

The regular season is over, and just like everyone figured, Wisconsin came out on top. And not just in the conference standings:

It's official—the 2015 Badgers had the greatest offense in the Tempo Free Era (since 2002). Now, Michigan fans will undoubtedly claim that the 2014 Wolverines actually bested this year's Wisconsin team in Adjusted Offensive Efficiency over the entire season. But, I would counter, that gap is very slim (much slimmer than Wisconsin's lead over 2014 Michigan in the conference season) and, more importantly, puts us in the precarious position of admitting Frank Haith presided over the 3rd-best offense in that time frame. I'm not comfortable with that.

Speaking of standings, Maryland fans are rather upset with my claim that this year's Big Ten was Wisconsin, and everyone else. After all, there's Maryland, a mere two games back, who beat the Badgers in their only matchup of the season. Well, a few points:

  • That was on Maryland's home floor.
  • Curiously, "beating Wisconsin at home" puts the Terps in a exclusive club with one other member—Rutgers. So maybe that should not be regarded as a trump card. 
  • Maryland is now 10-0 in games decided by 6 points or fewer (yes, that's arbitrary, as any number would be. But it's at least a benchmark—two-possession game—that's tied to the proceedings on the court). 
  • Was Maryland even the 2nd best team in the Big Ten? 

Team
Efficiency Margin, Conference Games
Wisconsin
+0.21
Michigan State
+0.09
Iowa
+0.08
Ohio State
+0.08
Purdue
+0.05
Maryland
+0.02
Indiana
+0.00
Illinois
+0.00
Minnesota
-0.03
Michigan
-0.05
Penn State
-0.06
Northwestern
-0.09
Nebraska
-0.11
Rutgers
-0.19

On a per possession basis, Maryland was closer to Nebraska than Wisconsin. That's not a knock on the Terrapins, who had a fine season, but a commentary on how dominating the Badgers were this year. 

Back to that 10-0 mark. I think most on the analytics side would argue that random chance has a lot do to with the success of Turgeon's team in tight games. I certainly would. But I'll allow for the possibility that maybe there's just some It Factor with this year. OK, let's say there is. The problem is that even the best poker players lose a hand every now and again, because they got unlucky. And in the NCAA Tournament, you don't get to play another hand. 

Frankly, I'd rather be in the camp of Wisconsin (2-2 in games decided by 6 or fewer), where almost always, luck will have nothing to do with how the team won.  

While I'm on a roll upsetting Maryland fans, let's take on defense. UMD was, on a per possession basis, the best defensive team in the Big Ten. 

Team
Defensive Efficiency, Conference Games
Maryland
0.993
Purdue
0.995
Wisconsin
1.003
Ohio State
1.005
Illinois
1.009
Michigan State
1.009
Penn State
1.023
Iowa
1.034
Nebraska
1.039
Minnesota
1.065
Michigan
1.071
Rutgers
1.08
Indiana
1.111
Northwestern
1.115

Or were they? Those teams in red—they only had to face the Wisconsin Offense once (or fewer). Of all the ways to praise Wisconsin's offense, this might be my favorite. 

(Also, it's been noted that Wisconsin got to face an easier conference slate than some other teams. Of this, there can be no dispute. After all, Bo Ryan's team had the enviable position of never having to face Wisconsin. What are the odds?)

As for postseason awards, I think this was an easy year for the media and coaches. Frank Kaminsky was excellent all season, and I'm hopeful we'll be able to watch him play in college for another month. Next stop, Chicago. 

Monday, March 2, 2015

Three Shots

One theme of tempo-free analysis is that basketball can be distilled into Four Factors: shooting (accounting for point rewards for converting 2s versus 3s), rebounding, turnovers, and free throw attempts. These items are, of course, normalized for opportunities of each—you cannot grab a defensive rebound on a made basket, for example.

But of those four factors, it's no secret that shooting dominates. A decided edge in shooting can and does trump deficiencies in the other three (ask Michigan fans about that). But we can do better than just chalk something up to "shooting," and thanks to Shot Analytics and Hoop Math, we now have the data to do just that (all numbers below courtesy of those two sites).

First, let's talk about the dreaded "mid-range" shot. The biggest issue with that shot is that it counts for just two points, and college players are really no better at making mid-range shots than 3s. They also do not generate as many free throws as shots around the rim (which also have the benefit of higher field goal percentages). In short, if a team is taking a lot of mid-range shots, it needs to make up for that with a very high shooting percentage. And that's the problem with Illinois:

The Illini lead the conference in the percentage of field goal attempts that come from the mid-range. The fact that Illinois is also tied for the worst-shooting team on mid-range shots suggests that Illinois might not be attempting these shots by design (it's a bad shot and the team isn't good at it at this volume—we should give John Groce the benefit of the doubt). 

For Illinois fans, this is sadly nothing new—the final years of the Bruce Weber regime featured offenses that had all of the trappings of a perimeter-oriented team (more on POTs in a bit) except, you know, that extra 50% of scoring that comes with a made shot. 

The Illinois results are interesting in that Groce's team is actually shooting 3s with relative accuracy: 

The Illini rank 6th in the Big Ten in 3-point percentage, which is a departure from Groce's first two teams (ranking 11th and 10th, respectively). This backs the notion that Illinois is not a team without shooting touch, but rather one that is taking too many mid-range shots. This is something like a marginal cost argument, so bear with me—as Illinois attempts more mid-range shots, those "extra" shots become more difficult. There's a baseline number of "open" shots that an offense can generate. But ultimately, every offense includes a number of contested shots. If a team (like, for instance, Wisconsin) has lots of players on the floor that can execute every kind of shot, that team will be in a position to shoot more open shots than others. But not every player can hit every shot, meaning some open shots are passed up. When an open shot is passed up, it increases the likelihood of a contested shot later in the possession. So what kind of shots is Illinois passing up, leaving it with too many difficult two-point jumpers?


Illinois either does not create enough chances at the basket for players that are capable of converting them, or is simply reluctant to take those shots (again, giving John Groce credit, the former seems more likely). The Illini are basically trading shots at the rim (which they convert at a 60 percent rate) for shots in the mid-range (32 percent). That's the biggest reason why the team ranks 10th in offensive efficiency, at present.

Indiana has also made a similar trade, but it's trading mid-range shots for 3s. Here's the mid-range graph again:
IU is "last" in the percentage of its FGAs from the mid-range. But rather than trade those shots for at-the-basket looks, the Hoosiers are playing to their strengths:

A little over 40 percent of IU's field goal attempts are 3s, which the team hits at over a 40 percent rate. No one has hit the 40/40 Club in the Tempo Free Era (Michigan and MSU came close last year, IU almost did it in 2007, the 2005 Illini came the closest), but IU is on pace to do it, with Maryland right there and Northwestern even with a puncher's chance.

The POT Line lies at 38 percent, meaning a team that dedicates 38 percent or more of its field goal attempts to 3s. As far as I know, it's arbitrary and an edict from the digital scrolls of the Wonk, which I think makes it unassailable law at this point.

Thursday, February 19, 2015

Let's talk Coach of the Year

In just a little over two weeks, the regular season will be over and we'll begin the best month in all of sports, where over 300 teams will essentially enter a tournament where a national champion will be decided by results on the court. It's a comfort to know that all those weekly polls, power rankings, and quasi-scientific indexes will all fall into the trash bin of irrelevance as we decide a singular winner on the basis of having won the final game of the season. The riddle I can't solve, however, is why those things continue to exist in the first place—every argument I've heard from the defenders ultimately boils down to the contention that these made-up rankings "generate discussion," which is A Good Thing. My retort has always been that college basketball is plenty entertaining in its own right, and examining the extreme nature of Joe Scott's system holds my attention far longer than the observance that Baylor lost twice last week, so hey, let's knock them down a couple of pegs.

Along with the close of the regular season comes the hardware, and almost all of the awards make sense—putting aside the objections of self-anointed wordsmiths who wax poetic over the difference between "valuable" and "outstanding" (a debate which centers on who played with the biggest collection of bums).

But the one that doesn't sit well with me is Coach of the Year. For one, I suspect that, unlike players, coaches don't really have good years and bad years. These guys have learned their craft over at least a couple of decades, and any changes from one season to the next tend to be marginal. Sometimes what they try works, sometimes it doesn't. That's sports.

Also, it seems like the award tends to be less of an accomplishment for the coach, but rather an indictment of prediction models, experts, and the media, which underrated a team from the outset of the season. Case-in-point: I suspect that Matt Painter is going to be a popular candidate for COY, because many (myself included) figured this team would be a lot worse than its turned out to be (but not everyone. Kudos, Dan.).

And maybe Painter should win it. But then again, Wisconsin is applying income inequality to the Big Ten in a matter we haven't seen since, well, at least the 2005 Illini (although Michigan State was at least in screaming range of that team, efficiency-wise. Today's Badgers are so far ahead of anyone else, it eclipses anything we've seen in the tempo-free era. So I'd be guessing. Flintstones? The 17-1 Calbert Cheaney team?). So does Bo deserve the nod?

There's also John Groce, who is on track to improve on last year's performance despite losing two rotation players for most of the conference slate. And what about someone like Thad Matta? Ohio State probably isn't over-performing relative to preseason expectations (most had the Buckeyes 2nd), but that's largely because Thad Matta has a track record of excellence and because he convinced D'Angelo Russell to attend Ohio State. But because both of those things were accounted for in October, Matta will not be regarded as highly? That seems unfair, so maybe Matta deserves consideration.

All of these interpretations of the award (and others) are entirely fair and defensible, which is really the problem. When the criteria for an award has nothing resembling guidelines, it's not so much an award as it is a psychology experiment. Maybe it's time we stop handing this one out—the merits of discussion generation be damned.

Tuesday, February 10, 2015

Badgers' Pursuit of History

Thanks to some timely shooting by Rutgers, and a couple of injuries to Wisconsin starters, most of the media probably regards this year's iteration of Bo Ryan Ball as a very good team but not one worthy (barring a deep tournament run) of any lengthy discussion in a historical sense.

That would be a mistake, because this year's Wisconsin team is a sight to behold on offense:
(If you want to design a great offense, have the team wear red. Every little bit counts.)

For years, the 2004-05 Fighting Illini have been the gold standard by which all other offenses should be measured. Barring a late collapse, that changes this season. The Badgers are assaulting defenses to the tune of 1.28 points per possession in conference play, the highest such raw mark in the Tempo Free Era (also known as "as far back as kenpom.com goes"). How have the Badgers achieved such lofty heights? By addressing last year's deficiencies while opting not to indulge in any of the conventional tradeoffs.


Wisconsin 2014
Wisconsin 2015
Change
2P%
50.9
55.6
+4.7
3P%
35.2
38.5
+3.3
Off. Reb%
25.9
32.1
+6.2
Turnover %
12.4
10.5
-1.9
FT Rate
44.9
44.5
-0.4
3PA/FGA
38.6
38.5
-0.1
 Generally, if a team wants to make more 2s and get more offensive rebounds, it means it needs it needs taller players that are better at making shots in the paint. These are usually not the same guys that are skilled with ballhandling and making 3s. Presented with these downsides, Bo Ryan apparently gave a pithy "I'm good, thanks" and went on his way. Maybe no facts illustrate this point better—Frank Kaminsky and Nigel Hayes own the highest assist rates on the team in Big Ten play, and they're also shooting a combined 45 percent from three-point range. Exactly who is capable of defending that?

After last season's Final Four run, the Badgers are getting more respect these days. But I still think the team's slow tempo (58 possessions, slowest in the Big Ten) is doing a yeoman's job of masking just how superb Bo Ryan's crew is on offense. It's been at least 13 years since we've seen an offense this good (Wisconsin also holds the Tempo Free title for best adjusted efficiency for all games—by a much narrower margin over last year's Michigan team). It might be another 13 before we see it again. Skipping over the Badgers because of the perception that they're "boring" or "plodding" would be a mistake—this is offensive perfection.

Friday, February 6, 2015

Fixing Iowa

Iowa really should be better. Not only is Aaron White having an All-Big Ten caliber season, but he has solid frontcourt running mates in Gabe Olaseni and Jarrod Uthoff. The backcourt will not be confused with Notre Dame's anytime soon, but offensively at least Mike Gesell and Peter Jok rank well above-average in the Big Ten (take note—this is something like damning with faint praise. As the backcourts of Michigan & Illinois have not exactly worked out as planned, guardplay is the a primary reason why the Midwestern conference is down this season). And Adam Woodbury—when he isn't running his hack-a-thon camp—has been serviceable in conference play.

It's easy to throw accolades out there when the Hawkeyes trail only the the Best Offense Ever (more on Wisconsin's sizable lead over the 2005 Illini in a later post) in Big Ten efficiency. But defense has been an issue throughout Fran McCaffrey's tenure, in ways that are difficult to characterize as unlucky:

The purple represents the conference averages over the span of McCaffrey's tenure at Iowa. As you can see, there's one season that really stands out (2013), but I honestly have no idea why. The two-point defense was especially improved, but I'm unable to come up with a reason why subbing out Bryce Cartwright and Matt Gatens for Woodbury and Olaseni can hurt one's interior defense.

Indeed, the presence of tall guys willing to block shots can (and probably does) explain why Iowa has never been particularly good at forcing turnovers ever since the tall guys started arriving in 2012. But it's worth noting that Iowa has always ranked 3rd or 4th in the percentage of opponent two-point shots blocked.

Ditto for defensive rebounding. The team has enough athleticism to be elite on the boards at the other end of the floor, so it has to be something other than frontline personnel.

Indeed, if you go back to McCaffrey's later Siena teams, you see a similar issues if you are looking for it. Prior to the 2008-09 season, McCaffrey had a couple of very good defenses fueled by turnovers. Those teams were quick, pressing, and didn't block many shots. In other words, they look nothing like what Iowa looks like today. In 2008-09, that began to change. The team blocked more shots, got fewer steals, and generally started this strange mix of bad 2P defense with solid shotblocking. The defensive rebounding also did not go all that well...at first. The next season, Ryan Rossiter was on the court all the time, and rebounding at DeJuan Blair levels. Indeed, even after McCaffrey left, and just about everything went south under the new regime, the defensive rebounding stayed solid with Rossiter in the paint. After he left, that slipped too.

Iowa's not good at very much on defense. They're dead last in defensive efficiency in Big Ten play. As long as the two-point defense is a mess, the ceiling on this unit is mediocrity. But in order to get there, the Hawkeyes need to rebound with the same fervor on defense that they do on the other side of the court. McCaffrey needs a new Rossiter.