/cdn.vox-cdn.com/uploads/chorus_image/image/29433749/20140227_mta_se9_385.0.jpg)
There has been a lot of grumbling about how Mark Few has toyed with the lineup this year, and I'll be the first to admit that some of this substitutions (or lack thereof) were a bit confusing.
The easiest way to measure how well a group of players is performing is to measure its points per possession and compare it to how well the opponent scored against that unit. I figured that this was a simple enough task considering that Few doesn't have the most expansive lineup and is pretty predictable with it.
Part of the problem doing this is that the tracking systems for college basketball aren't quite what they are compared to the NBA. I used statsheet.com's play-by-play to keep track of when players came and went, and I ended up with an overwhelming spreadsheet full of about 30 different combinations of players that saw the floor.
Luckily, Few, like other coaches, has lineups that appear more often. Much of the math I was doing also coincided with the percentage of lineups used by Gonzaga according to Ken Pomeroy. For reference, here are the most commonly used lineups over the past five games according to his site.
lineup | percentage used |
Stockton, Pangos, Bell, Dower, Karnowski | 21.4 |
Stockton, Pangos, Dranginis, Barham, Karnowski | 4.8 |
Pangos, Bell, Dranginis, Dower, Karnowski | 4.8 |
Stockton, Pangos, Barham, Dower, Karnowski | 3.8 |
Pangos, Bell, Coleman, Barham, Karnowski | 3.5 |
The thing that is also hard to measure is that each of these units aren't on the floor for extended amounts of time. Often times, players are on the floor as a set of five for a minute at a time once the game and substitutions really start going. Still, in tracking the past 10 games, there were definitive squads that were out the most often (according to my rough calculations with rounding).
lineup | approx minutes over past 10 games* |
Stockton, Pangos, Bell, Dower, Karnowski | 82 |
Pangos, Bell, Dranginis, Dower, Karnowski | 14 |
Stockton, Pangos, Bell, Barham, Dower | 14 |
Stockton, Pangos, Barham, Dower, Karnowski | 10 |
*note: also for sanity's sake, I rounded between a minute and half a minute while keeping track of the rotations. The minutes could have a decent deviation from the actual total, but in reality it doesn't matter that much -- they are just benchmarks for which lineups to focus on.
Those were the three bench utilized lineups that saw the most time on the court. After that, multiple different squads were in there for eight, seven, six and five minutes. But it would get convoluted to include every single one of the options here, although I may reference them later.
Measuring possessions in basketball isn't some new fangled idea and it is one of the basic benchmarks in a lot of statistical analysis. There are basically four ways a possession can take place: (1) a field goal attempt, (2) free throw attempts, (3) offensive rebound (which continues the possession) and (4) a turnover. The general consensus for determining possessions in college basketball is the following formula:
possession = FGA + 0.475*FTA - ORB + TO
Offensive rebounds are subtracted because that carries the possession for the team. If Sam Dower takes a shot, misses it, grabs his own rebound and makes the put back, he didn't really have two separate possessions of the ball. The free throw attempts get a modifier because often times in basketball you don't have to have possession of the ball to head to the free throw line. For my sanity's sake, I made a slight adjustment on the equation since everything was going by hand. My rough possession formula is as follows:
possession = FGA + 0.5*FTA - ORB + TO
As you can see, I didn't make too severe of a change. It is enough of a change that I wouldn't go citing these statistics all over the place, but for the purposes of the number crunching, every lineup (on both sides of the ball) got skewed in the exact same direction.
So now down to the nitty gritty. I used the past 10 games as my benchmark (and would have liked to do the entire season but ugh, that is a lot of work to do in one sitting). The past 10 games is a small sample size, but trends have to start somewhere and there were a few that definitely emerged.
lineup | points per possession | opponents' points per possession |
Stockton, Pangos, Bell, Dower, Karnowski | 1.06 | 1.01 |
Stockton, Pangos, Bell, Barham, Dower | 1.00 | 0.97 |
Pangos, Bell, Dranginis, Dower, Karnowski | 1.22 | 1.09 |
Stockton, Pangos, Barham, Dower, Karnowski | 1.00 | 1.15 |
Here are the points per possession and the opponents' points per possession for each set of five over the past 10 games. Right off the bat, of the more popular lineups, the thing to notice is the one that didn't include Stockton and Pangos was the one that offensively operated the best. I'm not going to place the blame at the feet of Stockton here, because throughout the 30 or so different lineups I put onto the spreadsheet, the common theme was when Pangos and Stockton were on the floor at the same time, the efficiency went down.
Dranginis isn't the reason that lineup is better at scoring, but it makes sense. The Bulldogs three guard lineup isn't an athletic lineup. Pangos and Bell are great at driving to the hoop but they also enjoy the long shot. Stockton isn't an outside threat, and is hardly an interior threat either. This leaves Gonzaga as a feast or famine unit, and when the threes are falling they are tough to keep up with. When the threes don't fall, the team struggles to put points on the board.
Often times, the first sub off the floor is either Dower or Karnowski. Circumstances are different in each game, but it shouldn't always be Barham coming in and moving to the four spot. Gonzaga hasn't made good use of the three spot at all this season, and any of the lineups that operated with a two guard lineup featuring either Pangos and Bell or Stockton and Bell performed at a better offensive rate. The exercise was limited in its scope, admittedly, but it was also consistent throughout, which lends at least some credence to the notion that the lineup might be amiss.
The slight difference in points per possession adds up. Gonzaga averages 67.8 possessions per game this season. The starting lineup would score almost 72 points per game. The one featuring Dranginis would score almost 83 points per game if it stayed in for the full 40. Obviously, that isn't feasible, but there are wins to be lost there by throwing out a lineup that isn't the most efficient one.
The peculiar thing about it all is that those trends with Stockton and Pangos are more recent trends, because the entire team has suffered since the start of conference play. The Bulldogs posted an average offensive rating of 122.53 in the non-conference slate. During conference play (Memphis included), that average dropped to 108.30. Injuries to Dower and Bell might have played a small role, but there wasn't a consistent uptick in offensive quality when both players returned.
It is understandable that guys become starters and they stay starters until playing their way out of the position. Players go through slumps and you can't just pull the plug on someone if they have an off-shooting night (or month in Pangos case). But, you can make adjustments to the subs and how they are used. Just because they are the starting five doesn't mean they need to play as a unit for a majority of the game. Gonzaga has some very capable bench players who could probably be starters on plenty of good D-I teams. Each game calls for a different situation and a different player, but the Gonzaga offense hasn't rolled like it did early in the season for a long while.
I'm not coming away with any recommendations to submit to Few and the coaching staff to take under consideration, because I'm just trying to understand why the team has seemed off as of late. In March, the offense cannot be off, because in March that means the season is over.