Misix Library

Learning to Defend the Three

March 1, 2016

Before we begin this post, let’s start with a tweet from one of our favorite accounts covering Marquette basketball:

Paintouches 3PT Defense Tweet

This got us thinking about why this might have been the case during the early tweet-inspiring games, and also made us wonder if the issue has continued to plague this young Marquette team through the rest of conference play. So we decided to gather data on the season through the second meeting with Villanova and take a look. In the end, we found that that this issue improved as conference play moved along, likely due to better defense from Marquette’s guards. But we wanted to take a look at two possible theories as to why Marquette was struggling to defend the three early on.


Big and Tall vs Quick and Small

Marquette often plays two 6’10” or taller players at the same time, arguably two “big men”. Most often those two players are Luke Fischer and Henry Ellenson, though occasionally Matt Heldt will be teamed with one of them. Given that most teams don’t have the type of players to pull this off, one of the two Marquette “big men” often have to guard a player who is much smaller and  quicker than them, making it tough to stick tight to their man. The size difference could also cause confusion on whether to switch on certain screens. In either case, this would be likely to result in open looks for the smaller and quicker players the big men are forced to guard.


Too Helpful

There is also the chance that Marquette’s guards were having a hard time keeping their assignments out of the lane, forcing the defense to collapse in order to help and allowing for an easy kick to a wide-open player behind the arc. Marquette could also just be too eager to assist, overcommitting to help defense when they would be better off sticking within reach of their assignment.



To test these two theories, play-by-play data was employed to analyze opponent three point shooting percentage when certain players are in the game. We gathered data on all 2015-16 season games through the February 27th matchup against Villanova, with the exception of the Presbyterian, Georgetown and second Xavier games, as there was no substitution information within the play-by-play data available at gomarquette.com. Let’s take a look at the results by player:


Table 1 | Opponent Three-Point Percentage

PlayerInOutIn vs Out
Sacar Anim26.7%32.5%-5.8%
Duane Wilson30.8%34.3%-3.5%
Henry Ellenson31.3%33.9%-2.6%
Luke Fischer31.3%33.1%-1.9%
Traci Carter31.3%32.5%-1.2%
Wally Ellenson31.3%32.0%-0.7%
Jajuan Johnson31.9%31.8%0.1%
Sandy Cohen III33.7%29.2%4.6%
Haanif Cheatham33.3%28.6%4.8%
Matt Heldt38.8%30.9%7.9%
Luke and Henry31.8%31.9%-0.1%

Table 1 shows the percentage of three-point attempts Marquette’s opponents have made with each player on and off the court. This certainly won’t tell us which player is losing their assignment or overcommitting to help defense, but it can give us a peek into how the team defends the three-point line, on average, when a player is in or out of the game. Keep in mind that other factors surely impact these percentages, such as which other players are most often on the court with them, the quality of the opponent’s lineup while they are on the court, changes in defensive scheme and many more.

Right away, we see that the stats for “Luke and Henry” seem to debunk the “Big and Tall vs Quick and Small” theory. Opponent three-point percentage was essentially unchanged when the Luke and Henry duo was on the court, with opponents shooting 0.1 percentage points worse from three with them on the floor together. Not only that, but the team defends better when Henry and Luke as individuals are in the game versus when they are not in the game. When Henry has been in, opponent three-point percentage has been 2.6 percentage points lower than when he has been out and 1.9 percentage points lower when Luke has been in versus out.

When we look at just the initial Big East games that spurred the troublesome tweet from Paint Touches, we find that when Henry and Luke were in together, opponents made just 26.1% of their attempts from three. That jumps to 48.3% when Henry and Luke were not on the court together. This means the team defended the three-point line much more efficiently when both Luke and Henry were in than when either just one or neither of them was in the game. Henry and Luke might have a hard time keeping up with smaller forwards and guards, but that hasn’t been showing up in the percentage of three-pointers opponents make while they have been on the floor.

This likely leaves us with the issue of Marquette perimeter players not being able to keep opposing guards out of the lane, or an unnecessary level of help defense being played. Looking at guards, we see that, to date, opponents made a lower percentage of threes while Duane and Traci were in the game than when they were out, with the difference being most evident with Duane. There has been no real difference in the team’s ability to defend the three when Jajuan is in or out. This wasn’t the case for these three guards during the four tweet-inspiring games in Big East play, where opponents made 43.1%, 41.4% and 40.0% from three while Duane, Traci and Jajuan were on the court, respectively.

Marquette seemed to defend the three much worse when Sandy and Haanif were in versus out. Also, similar to the other three perimeter players, numbers with them on the court during the tweet-inspiring Big East games left room for improvement, as opponents made 36.8% and 35.9% of their attempts from long range when Sandy and Haanif were on the court, respectively. The percentage of threes made by opponents was lower when Sacar and Wally were on the floor versus when they were not, and the opposite can be said for Matt, although it is tough to draw much from these numbers given the amount of floor time for each of them.

As mentioned earlier, whom each player is teamed up with likely has an impact on how well the team defends the three while they are on the court. Now let’s take a look at opponent three-point percentage with different player combinations. Table 2 includes opponent three-point percentage, for the year, for lineups where at least 20 threes were attempted against:


Table 2 | Opponent Three-Point Percentage by Lineup

LineupAverage Opp. 3P%
Henry Luke Duane Traci Sandy25.0%
Henry Luke Duane Jajuan Traci27.6%
Henry Luke Duane Hannif Sandy31.0%
Henry Luke Duane Jajuan Hannif31.6%
Team Average31.9%
Henry Luke Hannif Traci Sandy34.5%


Some things to note:

  • Henry and Luke play a lot.
    • The difference in ability to defend the three seems to come down to guard combination.
  • It seems that when the more experienced guards- Duane and Jajuan- are in, Marquette defends the three better.
    • Looking deeper, teaming up Duane with Jajuan or Traci has produced solid numbers:
      • In all lineups:
        • Duane and Jajuan: 29.7%.
        • Duane and Traci: 29.3%.
      • We also see evidence of poor three-point shooting defense when Sandy and Haanif are in the game together.
        • In all lineups:
          • Sandy and Haanif: 36.1%.
          • Sandy no Haanif: 29.3%.
          • Haanif no Sandy: 29.8%.
          • Neither: 27.1%.


Now let’s take a look at how things have progressed over conference play. To do this, we will examine opponents’ three-point percentage from each game and compare that to what they have averaged on the year. This is done in an attempt to control for some of the differences in the quality of each conference foe.


Opp. Three Point Percentage Conference Play


With the exception of the Xavier game, when the Musketeers refused to miss and logged their best three point percentage of the year, Marquette’s three-point defense has been trending in the right direction. Even with the Xavier game included, a linear trend line is able to help display the improvement seen throughout conference play.

So by just how much did things improve since the first four games? Marquette allowed their opponents to average 6.3 percentage points better than their season average in the first four games of Big East Play. Since then, Marquette has kept their opponents to an average of 0.2 percentage points above their season average. Remove the Xavier game and that improves to an average of 2.5 percentage points below their season average. That is a substantial improvement, with or without the Xavier game.

Let’s take a look at Table 3 to see this improvement by comparing the team’s defense during the first four games to that in the following twelve, while different players were on the court (not including the Georgetown and second Xavier games, as mentioned earlier).


Table 3 | Opponent Three-Point Percentage Change

 TeamDuane WilsonJajuan JohnsonTraci CarterHaanif CheathamLuke FischerHenry EllensonSandy Cohen III
First Four Games38.5%43.2%40.0%41.4%35.9%35.5%30.0%36.8%
Following 12 Games31.2%27.2%29.5%33.3%30.5%32.0%30.0%38.8%


Opponent three-point percentages fell the most after the first four Big East games when Duane, Jujuan and Traci were on the court, ending with three-point percentages below 30% when Jajuan and Duane were playing. This helps to explain the year-long stats from the first table in this post and suggests developing guard play might have something to do with the improvement seen in Marquette’s three-point defense. The change seen when Haanif and Luke were on the floor was not far behind. Things remained steady while Henry was on the court through all of conference play, with opponents hitting just 30.0% of their attempts.

A potentially surprising result is that opponents made a high percentage of their three-point attempts throughout conference play while Sandy Cohen was on the floor. This doesn’t necessarily mean that Sandy himself is bad at defending the three point shot, but it does suggest that, for some reason, the team struggles to defend the three point line when he is on the floor. Sandy was once called Marquette’s best perimeter defender by coach Wojciechowski, and therefore, you would be likely to expect Marquette’s perimeter defense to be better with him in the game. This hasn’t been the case, and while it could be that he has often been teamed up with worse defenders, played with a group that doesn’t meld as well defensively, or has even been asked to play against better three point shooting lineups, it is clear that something isn’t allowing the team to efficiently defend the three point line when he is on the floor.


Impact on Two-Point Defense?

One thing you might worry about with a team showing signs of improved three-point defense is its impact on two-point defense. Did Marquette’s improvements come at the expense of its two-point defense? It seems as though the answer to that questions is most likely no.

One way to test this is by taking a look at the correlation between how Marquette’s opponents performed from two-point range and three-point range compared to their season averages. When doing so, we find the two series have a correlation coefficient of -0.29. This coefficient is approaching the border of being considered a weak negative relationship, meaning that in games where Marquette’s three-point defense is better, you can expect their two-point defense to be just slightly worse. The opposite would also be true. This seems to make sense, as when a team sells out to stop the three, they are more likely to leave open opportunities closer to the basket. In the end, this isn’t considered a strong correlation in the world of statistics.

We can also take a look at the average opponent two-point percentage in comparison to their season average and compare the same four-game to twelve-game periods. In the first four games of Big East play, Marquette held their opponents 1.5 percentage points below their two-point season average. Over the next twelve games, that number improved, as the Golden Eagles held their opponents 2.1 percentage points below their season average.

A visual representation can also help to analyze the data:


Opp. Three-Point and Two-Point Conference Play


In all, the Golden Eagles’ two-point defense has remained relatively steady, with the exception of the last two games against Creighton and Villanova, in which a slight upward trend seems evident. This could have had something to do with an injured Matt Heldt being unable to relieve Luke Fischer over that stretch, especially considering the foul trouble Luke experienced versus Villanova. Nonetheless, there is no clear evidence that Marquette’s three-point percentage improvements have come at the expense of its two-point percentage defense. This is good news for a young team trying to make a statement to end the 2015-16 season.