Time for a Change at Point Guard

If there was a bright spot to Saturday’s 74-65 home loss by the Washington Huskies to the last-place Utah Utes — and that requires not just a glass-half-full disposition, but a glass-overflowing one — it was the performance of redshirt freshman guard Andrew Andrews, who scored a team-high 17 points. Down the stretch, it was Andrews, not senior starter Abdul Gaddy, who ran the point for the Huskies. That change ought to become permanent.

Thanks to StatSheet.com’s plus-minus data and Lorenzo Romar‘s tight perimeter rotation, we can see how Washington has played with each combination of its four guards on the court during the first five games of Pac-12 play.

Andrews-Wilcox-Suggs: +17.4 points per 40 minutes
Gaddy-Wilcox-Suggs: +8.4
Gaddy-Andrews-Wilcox: -0.9
Gaddy-Andrews-Suggs: -6.9

Besides indicating that C.J. Wilcox simply can’t afford to rest for more than brief stretches, the plus-minus data also suggests that Andrews has been the superior option to Gaddy at the point. Yet Gaddy has played more than two-thirds of the minutes where the two players haven’t been on the court together.

If you’re not convinced by five games of noisy plus-minus data, I can’t blame you, but in this case their conclusion is backed up both by the eye test and individual statistics. Gaddy’s shooting slump is so far in his head that he’s not just missing, but missing badly. Gaddy is shooting 22.2 percent on threes and 27.0 percent on twos in Pac-12 play. Opponents have taken notice and are sagging off him, gumming up the rest of the Washington offense and making life more difficult for Wilcox and Scott Suggs on the wings and Aziz N’Diaye in the post.

At the same time Gaddy is dealing with a crisis of confidence, Andrews is seeing his grow each game. He’s made 57.9 percent of his twos and 38.5 percent of his threes against conference opponents, providing the Huskies needed scoring. While Andrews is much more prone to turnovers than Gaddy, his decision-making is showing signs of improving. After committing seven turnovers in his first three conference games, Andrews had just two combined against Colorado and Utah this week.

I understand if Romar is reluctant to publicly embarrass his senior leader by benching him for a freshman. Gaddy has been a great teammate throughout a career that hasn’t gone quite the way either he or Washington fans envisioned, and his defensive effort against bigger Spencer Dinwiddie was a major factor in the win over the Buffaloes. So it makes sense to keep Gaddy in the starting lineup to preserve his remaining self-assurance. What I’ll be watching instead is the ratio of Gaddy’s minutes to Andrews’ playing time. That should be close to even going forward, if not tilting toward the freshman’s direction.

Andrews has demonstrated throughout his redshirt freshman season that he’s a key part of the future of this program. Now it looks like that future might be here already.


NCAA Timeouts and the Definition of the #Howland

You don’t have to watch UCLA basketball very long to know that Bruins head coach Ben Howland uses his timeouts differently than most other coaches. Howland rarely chooses to save his timeouts for the end of games and often calls them after his team has scored, even when UCLA ostensibly has control of momentum.

There’s one thing in particular that Howland does that drives me, as a disciple of the Dean Smith school of saving timeouts, nuts. When he does use his timeouts to stop other teams’ runs, he habitually calls them in situations where the next dead ball would be a mandatory timeout. This is my definition of a #Howland, and the man himself is hardly the only coach guilty of calling them on a regular basis.

As an aside, a quick primer on NCAA timeouts. Each team gets five timeouts to spend at its discretion. One of these must be called before halftime or it is lost. In addition, there are so-called “media” or “TV” timeouts — beat writers are clamoring for a stoppage! — that take place at the first dead ball after the 16-, 12-, 8- and 4-minute marks of each half. (Note that if the clock stops at precisely the minute mark, like 4:00, that is not a mandatory timeout. Hence, it’s the “under-4” timeout or so on.)

Now, the NBA has media timeouts too, but the current rules dictate that a timeout called ahead of the mark replaces the mandatory NBA timeout. This means the NBA avoids that scourge of college fandom — a stretch with a timeout called from the bench, one play, a stoppage and immediately the mandatory timeout. It also makes the #Howland impossible.

To maintain a strict definition of the #Howland, I only count timeouts called to stop runs. (Other interpretations may be more liberal.) Basically, calling a timeout in this scenario is something of a waste because the next dead ball will stop the momentum just the same without costing the team a precious timeout. In fact, I’d advocate in situations like this that the team ought to gamble for a foul or try to kick the ball, which would force a timeout at no cost to the team.

There are those who argue in defense of the #Howland, noting that the double-timeout is sure to quiet even the loudest of road crowds. So one a game might be acceptable, especially in the first half when the timeout won’t carry over anyway, but Howland himself has a habit of calling multiple and running out of stoppages with plenty of game time remaining. Of course, even when Howland does save a timeout, he doesn’t always use it. When it comes to timeouts, Howland just can’t win.

Huskies Better Lucky and Good

With just less than two minutes remaining in last night’s 64-54 Washington win over visiting Colorado, Buffaloes forward Andre Roberson banked in a three-pointer. Unless you happen to be an NBA scout taking note of Roberson’s range, you probably didn’t pay the shot much attention. With Colorado down 11 at the time, it mostly served to prolong the final two minutes by encouraging Tad Boyle to foul and Lorenzo Romar to take timeouts to protect the lead.

What made Roberson’s triple notable was that it was the Buffaloes’ only three-pointer of the night in 10 attempts. In a game where points were at a premium, the Huskies’ 18-3 advantage from long distance was a major key to the win. Consider it an off night for Colorado, which usually makes 35.7 percent of its three-point tries, but part of a larger trend for UW.

The difference between Washington’s 4-0 start to conference play and their underwhelming 8-5 non-conference mark — which included home losses to Albany and Nevada — has been largely at the defensive end of the floor. The turnaround really dates to the Huskies’ final game before Pac-12 play, a 61-53 loss at Connecticut that was attributable to an off shooting night from star C.J. Wilcox, not the defense. Before that game, UW was allowing opponents to score more points per 100 possessions than their adjusted season-long marks (via KenPom.com). In plain English, the Huskies were worse than the average NCAA defense through their first 12 games. That’s bad news for a power-conference team, and explains the upsets Washington suffered.

Starting in Storrs, the Huskies have been a juggernaut defensively. Their last five opponents have collectively averaged 17.0 points per 100 possessions fewer than their season-long adjusted marks. Only one team in the country (Louisville) has been so stingy over the course of the season. There are multiple explanations for the turnaround. A healthy roster has allowed Romar to trim his rotation. A week of practice during the holidays may have given the coaching staff more time to focus on defense after spending early practice time implementing a new high-post offense. And Aziz N’Diaye has been dominant in the paint, successfully avoiding foul trouble to stay on the court for more than 30 minutes a night in conference play.

Beyond that, Washington has been extraordinarily fortunate in terms of opponents’ three-point shooting. Over the past five games, Husky opponents have shot 12-of-67 (17.9 percent) from downtown. While there have been some middling three-point opponents in the group, weighted for their attempts in each game, this group should have shot 32.8 percent. Colorado and Connecticut both would have scored nine more points had they merely reached their season average in those respective games. The impact was even greater in the Stanford and WSU games. Both teams made two fewer threes than expected, which could have swung close games.

Now, you’re probably protesting that improved three-point defense reflects everything else Washington has tightened up. Unfortunately, that doesn’t square with the research done by Ken Pomeroy, who has found that teams have little control over the three-point percentage their opponents shoot over the course of full seasons, let alone five-game stretches. You know who’s an example of this? The Huskies. When I took a look at the team in December, I noted that one reason Washington’s defense was playing so poorly was bad luck in terms of opponent three-point shooting. Through Dec. 22, Husky opponents should have shot 34.0 percent from beyond the arc. They actually hit 37.5 percent, though it’s hard to find an example of hot shooting actually causing Washington to lose (the two best examples were Seton Hall, which the Huskies won anyway, and a blowout loss to Colorado State).

Using the same method of expected three-point percentage, we can construct ratings for each of Washington’s games independent of three-point defense and summarize the two portions of the season (note that Pac-12 includes the UConn game, as before):

Stretch   DRtg   OppORtg  AdjDRtg   3P%luck  No3DRtg  No3AdjDrtg
NonC     101.6    101.0    - 0.6      - 8      98.6      +2.2
Pac12     88.0    105.1    +17.0      +10      97.2      +8.1
Total     97.7    102.2    + 4.5      + 2      98.2      +3.9

That’s a lot of column headers, so let me explain. DRtg is the actual points per 100 possessions opponents have scored in each stretch. OppORtg is the opponents’ season-long Adjusted Offensive Efficiency — what we’d expect them to do against an average opponent. So AdjDRtg is the difference — how much better the Huskies are than average.

3P%luck is the number of threes better or worse teams have done than expected. No3DRtg is how many points per 100 possessions teams would have scored had they made threes at their season-long percentage against the Huskies, and No3AdjDRtg uses this figure to see how the rest of the Washington defense compares to NCAA average.

Even if you don’t believe Pomeroy’s research, it’s clear that taking out three-pointers produces a more consistent view of defenses. More than 20 percent of the variance in UW’s adjusted defense from game to game can be explained solely by opponents’ three-point shooting. Factoring in the variability in three-point defense, the Huskies were above average all along, though they have gotten better over the last five games at other facets of defense.

The three-independent figure would put Washington about 50th in the nation defensively, which is similar to where the Huskies have been the last two seasons. Add in an offense that has improved slightly from 2011-12 and this Washington team now looks capable of ranking in the upper third of the Pac-12 and challenging for an NCAA tournament berth. That’s as big a turnaround as the three-point defense has made.

AN ASIDE ON LUCK: I know using the term “luck” bothers some readers because of the fact that it doesn’t take into account the skill, strategy and execution involved on the court. I would compare the issue to forms of gambling that include both skill and luck, like blackjack and poker. Over the course of a single hand, or even a night of play, randomness is the dominant factor, which is why we say poker players get lucky or unlucky. It’s only in the long run, over thousands of hands, that the randomness evens out and we can tell stronger players from weaker ones. Three-point defense is similar, except that the college season doesn’t provide thousands of attempts. There’s not enough time for the differences between teams to become evident. So three-point defense is in fact a skill; it just can’t be determined from opponents’ three-point percentage, which tends to be misleading.