Analyzing NEC Similarity Scores

For an NEC fan, a star’s potential breakout is the best thing to dream about during the offseason. Imagining a player turning into the next Julian Boyd or the next Shane Gibson can carry fans through an entire summer. But how realistic are those dreams?

As always, the answer is: It depends. But I wanted to quantify what “depends” means in this case, so I built a Similarity Score model for NEC basketball using tempo-free statistics from 2008 to the present. If you want more about the nitty gritty details check out the Methodology section at that bottom of this post. But here are the basics: The model works much like the similarity scores at the heart of FiveThirtyEight’s CARMELO algorithm. If two players were exactly alike they would score 100, but no two players are. For last season’s players the closest comparisons range from a high of 82.6 to -75.1 with a median of 60.6. Anything above 70 is a rather strong comparison, above 60 is relatively strong (the average is 54) and after that the results are much noisier.

Here are the most similar players for the preseason NEC First Team:

  • Keith Braxton, Saint Francis: Marcel Pettway (Bryant, 2016) – 60.8
  • Junior Robinson, Mount St. Mary’s: Jesse Steele (Monmouth, 2012) – 62.2
  • Darian Anderson, Fairleigh Dickinson: Shemik Thompson (CCSU, 2010) – 65.0
  • Isaiah Blackmon, Saint Francis: Corey Henson (Wagner, 2016) – 64.9
  • Joe Lopez, Sacred Heart: Kinu Rochford (FDU, 2012) – 74.4

None of these are the most flattering comparisons. Pettway tailed off after his strong freshman season, losing more than a point and a half rebound per game. Braxton though is a tough player to gage after such a strong freshman season, his other comps include Quincy McKnight and Cane Broome.

Steele was a nice player in the NEC, but didn’t have the dynamic talent of Robinson. Because height is a variable in my NEC model, it’s really hard for Robinson to project out to an elite level. Other comparisons included Yunus Hopkinson and former Wagner guard Chris Martin.

Rochford played on bad FDU teams, but his statistics were actually quite good. He averaged nearly a double-double during his senior season, with 14.7 ppg and 8.9 rpg. If Lopez puts up production like that on a contending Sacred Heart squad it would put him right in the conversation for NEC Player of the Year. Lopez and Rochford possess the fourth highest similarity score of all players that played in 2016-17. Other players the Sacred Heart big man is close to include former Wagner forward Michael Carey and the NEC Player of the Year last season, LIU’s Jerome Frink.

Of course no post about the offseason dreaming of NEC fans would be complete without a look back at the talent the league lost. Here are the closest comparables for some key guys that transfered out of the league:

  • Isaiah Still, Robert Morris: Kyle Johnson (LIU, 2009) – 71.1
  • Marcel Pettway, Bryant: De’von Barnett (SHU, 2015) – 77.5
  • Nisre Zouzoua, Bryant: Shane Gibson (SHU, 2011) – 71.6
  • Elijah Long, Mount St. Mary’s: Darian Anderson (FDU, 2016) – 72.6
  • Miles Wilson, Mount St. Mary’s: Nisre Zouzoua (Bryant, 2016) – 63.6
  • Josh Nebo, Saint Francis: Dan Garvin (Bryant, 2015) – 47.3
  • Nura Zanna, LIU Brooklyn: Jordan Allen (SHU, 2015) – 49.2

Nebo and Zanna are unique cases, but the guards that left the league all had strong comparisons to past greats. It’s certainly a big talent drain for the NEC to lose those players to Nevada, the ACC and beyond.

But there will be guys that step up and take their place. Here are a few players who aren’t on the preseason NEC first team but have interesting close comparisons.

  • Mike Holloway, Fairleigh Dickinson: Julian Boyd (LIU, 2011) – 73.3
  • Adam Grant, Bryant: Nisre Zouzoua (Bryant, 2016) – 69.9
  • Ikenna Ndugba, Bryant: JoJo Cooper (Wagner, 2015) – 70.4
  • JoJo Cooper, Wagner: Dave Johnson (QU, 2012) – 72.4
  • Blake Franics, Wagner: Byron Ashe (MSM, 2014) – 75.5
  • A.J. Sumbry, Wagner: Amdy Fall (SFBK, 2014) – 69.1
  • Rasheem Dunn, St. Francis Brooklyn: Darian Anderson (FDU, 2015) – 65.8
  • Randall Gaskins, Jr., Saint Francis: Zaid Hearst (QU, 2012) – 68.7

Holloway’s 72.4 similarity score to Julian Boyd is the seventh highest score of any player that played during the 2016-17 season. It’s not a stretch to think that Holloway—who shot 55 percent on his two-point attempts last season and had a nationally ranked offensive rebounding rate—could emerge as one of the league’s elite players this season. Holloway’s next closest comparisons are Shawn Atupem, Ike Azotam and Justin Rutty. Atupem, who played for Mount St. Mary’s, represents the downside of Holloway’s season and he even averaged 10.7 ppg and 3.5 rpg as a junior at the Mount. Holloway expect big things from himself this season.

“I definitely believe that I can compete for a first team spot,” Holloway said at NEC Social Media Day. “I’ve put in the time at the gym. I’ve worked on my body.”

Another wrinkle that could help move Holloway into the NEC elite is a reworked jump shot.

“I know everyone thinks of me as a back to the basket guy, but I believe I can expand my game this year and that’s something everybody will be shocked to see this year,” Holloway said about his new face-up game.

It is quite convenient for Tim O’Shea that despite Zouzoua’s transfer Grant is waiting in the wings. Grant turned the ball over slightly more than Zouzoua did during his freshman season, but otherwise their freshmen campaigns were extremely similar. There’s no guarantee that Grant will automatically score 20 ppg during his sophomore season, but the signs are certainly there. His other top comparisons are all elite scoring guards, including Cane Broome, Julian Norfleet, Ricky Cadell and Robby Ptacek.

O’Shea though is most excited about Grant’s attitude and his desire to improve.

“He’s a really unselfish player, he always makes the right play whatever situation he’s in,” O’Shea said about Grant. “He’s got a competitive gene in terms of he never takes a play off even in practice every day. The guy is a tremendous competitor and you can’t teach that as a coach. … He’s going to get better every year he’s here.”

Finally, the Wagner trio is one of the reasons the Seahawks are so intriguing. If Francis develops into a dominant scorer on the wing given more of an opportunity, Cooper becomes a steady hand both offensively and defensively, and Sumbry provides an anchor in the paint then Bashir Mason will already have an extremely solid foundation to built an NEC contender upon. Sumbry quietly had a 9.4 percent block percentage during his sophomore season. Putting up those types of numbers, even while only playing 15 mpg, is not easy to do. More playing time could help him become an NEC Defensive Player of the Year candidate. B.K. Ashe went from a 7 ppg scorer as a freshman to one of the focal points of the Mount attack as a sophomore and junior. Francis was an extremely lethal scorer as a freshman (his 122.9 offensive rating ranked 70th in the country) and while he probably won’t be that efficient as a sophomore, his scoring average should rise with more opportunities.

All of these comparisons give NEC fans some reasons to dream that the right pieces are ready to take their team to the next level.


To build the similarity scores I scraped the tempo-free statistics for every NEC team, including Monmouth and Quinnipiac when they were still in the league, from Bart Torvik’s excellent site. I then cleaned up a few statistics, including converting height to inches and calculating three-point rate.

After gathering the data, the next step was to build a similarity score for each player. The similarity score is built on 13 components, which I weighted in hopes of making sound basketball sense moreso than building a perfect model.

Statistic Weight
Height 2
Minutes% 3
eFG% 3
Usage 3
FT% 1
Three-point Rate 2
Assist% 1
TO% 2
OR% 1
DR% 1
Blk% 2
Stl% 2

The weights were divided into vitals (5 points), shooting/scoring (11 points), other offensive attributes (4 points), defense (5 points). Therefore this model probably slightly overweights offense, but it’s also the easiest part of basketball to track through boxscore statistics. (And can also tell you about a player’s other physical attributes.) There is no star-rating component in the model because the vast majority of NEC players are either 2-star or fewer recruits.

To find the most similar player to a specific player for a specific season I then take all of the players of the same class year of the original player (freshman are compared to all freshmen for instance) and find the player that is most similar using a weighted nearest neighbor calculation. (There is a great explanation about how FiveThirtyEight does it available here.)

Finally, I retrieve the top comparison for a player along with anyone in the top five comparisons that played the next season in the NEC (that’s between 3 and 5 players total for all players). Those are the comparisons that were discussed during this post.

3 thoughts on “Analyzing NEC Similarity Scores

    1. The Bart Torvik data goes back to 2008, so that’s what I used for comparisons. I’ve considered going back further with College Basketball Reference or another source, but I thought these were a good start with about a decade of data.


      1. Thank you. In the early-to-mid ’90’s, the NEC was loaded with talented point guards … MSM (McGuthrie), Rider (Hames), FDU (Dasher), Monmouth (Geraldo) and Marist (Basile) had PG’s who all came in the same year and at some point in their careers earned 1st or 2nd team all-conference despite the depth at the position. I can only imagine the debates that would have taken place if we had social media back then. I compare Junior favorably to Rider’s Deon Hames from the crop. I’d be curious how those players numbers would stack up against the more recent generation in the NEC. I know I sound like the old guy on the block, but that was an unbelievable time for the conference. Those five teams were also the best 5 teams, so the games often were big showdowns, but took a back seat to the point guard wars.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s