Analytics Approach to Georgetown’s Minutes Distribution

With the 2019-2020 season right around the corner, the long wait for another Georgetown Hoyas season is almost over. There is more excitement and anticipation in the fanbase this year than in several years. A common topic of offseason discussion has been the Hoyas’ starting lineup and the allocation of minutes for each player for the season. The purpose of this article is not to speculate on that but rather to take a look back to see what combinations worked and did not work last year. It also examines how well a couple of the new roster additions performed at their old schools and how they may fit in. Unfortunately, we can only pull stats for those players that played Division 1 basketball, so the incoming freshman and Galen Alexander (who played at the JUCO level) will not be included.

Much of the analysis focuses on points per possession, or PPP, which is a common stat used to measure effectiveness both offensively and defensively. PPP demonstrates how many points the Hoyas scored on offense and gave up on defense per possession when certain players and lineups are on the court. For instance if the Hoyas played a game with 75 possessions and scored 75 points that would be a PPP of 1. The same goes for defense. On offense, the higher the number, the better. On defense, the lower the number, the better. I am also including the differential, which is simply the PPP on offense minus the PPP on defense. While the differential is usually small, keep in mind this is for each possession, and thus, is important over the course of hundreds of possessions. The Hoyas played on average about 76 possessions per game last year as a point of reference. I have broken the data sets down into two groups. One will be all games and the other will be just Big East games. All of the data in this article was gathered from HoopLens.

Chart 1 - Shows results for when each player was on the court

● The higher amount of possessions was used in cases where offensive and defensive totals were different. Green boxes show players that outperformed the teams overall PPP differential while red shows those that didn’t and white were those that were equal.

Observation - Since this chart is sorted by total possessions it would be ideal if all the green shaded players were near the top and red shaded players at the bottom. Put in simple terms, you want the best players playing the most, and the least effective players playing the least. Two of the predominant starters (Mac McClung and Jamorko Pickett) were in the red. Jamorko was eventually replaced in the starting lineup at the end of the season by Jagan Mosely, while Mac was left as a starter despite having two backups that outperformed him on a PPP basis. The chart also shows that the Hoyas did not have very good backup big men as Trey Mourning and Kaleb Johnson both came in as net negative players.

Chart 2 - Shows the results for when each player was off the court in Big East conference play.

Observation – Examining how the Hoyas performed without each player on the floor is informative because it shows the net effect of each player. Here, a negative differential is ideal. As most would probably expect, James Akinjo was by far the most valuable Hoya on the team in conference play. Once again, the team generally performed better when Mac and Jamorko were off the floor (with Trey and Kaleb just above them).

Chart 3 - Best 2-man combos based on a minimum 100 possessions

Observation - This shows which two-man combos performed that best when they were on the court together. Jahvon Blair shows up in the two best combos and his combinations are the best by quite a large wide margin. Greg Malinowski shows up in four of the eight best combos while Josh LeBlanc shows up in 3. The combo of Josh, Jahvon and Greg formed the top three, two man combos. Even though his overall numbers were poor, Jamorko showed up twice in the best combos list. The biggest take away from this is when the Hoyas had two shooters on the floor (or Josh) the results were much better than they did not. Yes, this includes Jahvon even though he was only an average shooter because he seemed to get the respect of being a good shooter, which is reflected in his propensity to get fouled on 3’s, which helped with spacing.

Unsurprisingly, the worst combos all either have Trey (5) or Kaleb (4) so it is no wonder why they were by far the worst combo. One of the biggest reasons for optimism is essentially addition by subtraction – one would hope that one of the new bigs can step in to fill the “backup big” slot that was had been manned by Trey.

Oddly, Kaleb shows up in some of the best and worst two-man combos. The key difference was the good defensive numbers when Kaleb was on the court with Jamorko and Jagan, whereas the bad defensive numbers when he was paired with Trey, Jessie, Mac and Jahvon.

Chart 4 - Best three man combos based on a minimum 100 possessions

Observation – Once again, Jahvon once again leads the best three man combos—indeed, he was part of seven of the best eight combos. The Jahvon, Greg, Josh combo had an amazing +.40 PPP differential. Given how strong this combination was, and how little it was used, arguably it should have been used more. Josh was in the top 3 combos and shows up in 5 of the top 8 as well. LeBlanc clearly outplayed Mourning, and these numbers confirm that he was was kept behind Trey for far too long.

Again, Trey and Kaleb show up in the worst combos. The biggest difference in the two and three man worst combos is that Mac shows up in the top three combos. Also, Jahvon shows up on the worst list twice, as well. These stats demonstrate how valuable Josh is when you consider the Jahvon and Jessie combo show up on both the best, with Josh, and worst with Kaleb.

Chart 5 - Best four man combos based on a minimum 100 possessions

Observation – In looking at the best four man combos, one begins to see that most of the key lineups have the starters and high-usage bench players. The backups no longer make the list simply because they did not play enough possessions.

Greg and Josh headline the best four man combos, present in three of the top four spots. The Jessie, Greg, Josh combo was included in the top 2. This made me dig a little bit deeper as those two combos had a guard that wasn’t James in them. When Jessie, Greg and Josh were on the floor with James they were a +.04 PPP when they were on and James was off they were +.32 PPP which is quite astonishing considering what the rest of the metrics seem to show us. The team shot an insane 62% from 2 and 45% from 3 when that combo without James was on the floor compared to 50% and 39% when he was. It’s possible that when James is not on the court the team relies more on passing and movement which created more open looks for the spot up shooters which Greg and Jessie were. If there was one knock on James last year it was that his overdribbling caused the offense to stagnate which those numbers back up. That is the basis for a possible lineup that will be discussed later.

There were only a few five man combinations that had more than 100 possessions.

Chart 6 – How Terrell Allen and Omer Yurtseven Performed on Their Old Teams

Observation - Aside from the freshman and Galen Alexander (who only played 55 minutes at LSU before being dismissed), Allen and Yurtseven logged significant Division I minutes. While Omer missed last season as a transfer, his stats at NC State show that NC State played better offense with Yurtseven, but worse defense. Sound familiar? Omer should definitely be able to fill the offensive void left by Jessie Govan. While the Hoyas’ fanbase seemed very discouraged by Govan’s defense, there is little reason to believe Omer will be a better defensive player. While he may have improved after the tutelage of Ewing over the past year, his defense at NC State was average, at best.

One of the most exciting potential additions is Allen, who was one of the best defenders in all of Division 1 last season. UCF, with Allen, defended significantly better with Allen, than without. Some may say that was because UCF had 7’6 Tacko Fall, and while Allen was a better defender with Fall on the floor, Allen seems to have defensive chops that are unmatched by anyone else. While Allen likely falls behind Akinjo and McClung in the depth chart, if his defense is as strong as it was last season, it will be hard to keep him on the bench, especially when facing tougher guards like Powell or Howard.

What Does It All Mean?

One can examine this data and conclude that Coach Ewing frequently used lineups that were not very effective. That said, Ewing has indicated he wanted to get Akinjo and McClung more experience for the future. In this sense, Ewing had longer term goals in mind, however, with higher expectations and visions of NCAA bids, Ewing will have to focus more on what lineups work, than getting certain players experience. In my view the fan base has excused the underwhelming team performance the past 2 years, but that will change if the team misses the NCAA tournament this year. There appears to be good reason for optimism, as the key pieces are back, Yurtseven should fill Govan’s shoes, and our freshman should outperform the graduated seniors.

Most would agree that James, Mac, Josh and Omer are extremely likely to start. Statistically, there are strong arguments for all except McClung, who needs to become more efficient and a better defender to justify his playing time. If Mac starts off the season cold, will he Allen take some of that time? Ewing has more tools this season than last, and a shorter leash may be more advisable, especially with a stronger out of conference schedule.

The data shows that Jahvon absolutely deserved more time last year, which would have come at Mac’s expense. In some sense, it is odd that Blair got solid minutes as a freshman and had much better stats than Mac, yet he played less than Mac. Malinowski was also a better option statistically, though he is irrelevant for 2019-2020 given his graduation. It is unclear who will replace Greg’s shooting, but hopefully Mac or another guard improves. Galen shot 35% from 3 in JUCO last year, as well, similar to Jamorko and Jahvon while Greg shot 39%.

Now we have Allen who might be better than all of them. His PPP stats were better on a NCAA tournament team that was an unlucky bounce from beating Duke. He was brought in to be the backup point guard, but that would only provide about 8 minutes per game. Given how poor the Hoyas’ guards defensed last year, he will almost surely be the best defensive guard, which would be incredibly valuable.

The 3/SF spot is other main question mark in the starting lineup. Jamorko started in the spot for most of the year until he was eventually benched for Jagan. That move was prudent based on the data of +.09 PPP differential in the starting lineup combination with Jagan and -.04 PPP differential with Jamorko. Jamorko’s performance varies significantly based on who he plays with. Overall, his stats were poor but in certain lineups he did play well, though he did not perform well with the likely starters. Last year, James, Mac, Josh, and Jessie played better with a third guard. Assuming that holds true, Jagan would be the obvious choice with Jahvon being an option. The James, Mac, Jahvon, Josh, Jessie lineup was a very good +.72 PPP last year. Granted it was only for 22 possessions, too small a sample to make solid conclusions, but it does show that a three guard lineup should be experimented with in the early games.

Another option Ewing could experiment with is having Terrell playing point and James off the ball. This could accomplish a few things. One it would help with the defense of playing three guards. Two it could get more open looks for James who was our best shooter last year as well as whomever proves to be a capable shooter if paired with James. I described above how James could hinder shooters. Terrell has shown to be a very willing and good passer. Three it provides more minutes for Terrell, who as we described above, is likely one of our best guards this year. Lastly it would provide more rest for James while still being on the court. Playing off the ball would allow him to not exert as much energy. Since we all hope for a long season, it would be prudent to make sure our best players are fresh at the end of the year when the games are most important.

Needless to say there are plenty of options and playing time at the 2 and 3 positions. Let’s hope Coach Ewing tries them in the non-conference and uses all the info provided to him to find the best combinations.

The backup post positions are completely up for grabs. To put it lightly our backup bigs last year were not very good. While Mourning and Johnson are gone, none of the returning players were very good last year backing up Josh and Jessie including Jamorko. Jamorko’s PPP showed that he was not just as bad as Trey and Kaleb when he was asked to play the four (backup PF) position. Pickett thrived when he was on the court with either Trey or Kaleb and Jessie. When he was only on the floor with Jessie and no other PF his PPP differential significantly dropped. This seems to point that his most effective position is the 3/SF. That’s unfortunate as Jamorko being able to fill the backup PF role would ease some of the reliance on the three new freshman. Undoubtedly, the three freshman bigs bring a ton of size, and hopefully one of them is the answer to this need. That would greatly aid the journey toward achieving Ewing’s first appearance in the Big Dance as a coach.