clock menu more-arrow no yes

Filed under:

Villanova Basketball Advanced Stats: Hit the Clutch

New, 16 comments

One of the biggest concerns for Villanova basketball heading into the 2015 campaign is maintaining its stats-defying record in close games.

Mark Konezny-USA TODAY Sports

In 2013-14, Jay Wright's Villanova Wildcats went 7-1 in games settled by 5 points or fewer, which allowed them to march to a 28-3 regular season record and run away with the regular season Big East title. It is commonly accepted in the advanced stats underworld that teams deviating from the norm eventually fall back towards it. It is hard to distinguish between teams that are ‘lucky’ for a season, and teams that create their luck (to some degree) with skill.

Villanova went 7-1 in close games last year, and probability demands they work their way back towards .500 territory this season. It is accepted convention in sports that teams winning the majority of their close games can not sustain their winning percentage. History and its metrics has taught us that team's records eventually regress to .500 in close games. However, this tendency is not exactly a coin flip, as there are aspects of the outcome in the team's control.

There are a series of factors that went into this unlikely success in highly contested outcomes, funneling down from coaching/system, to lineups employed, to the individual player. All this combines into one burning question: should Villanova (and its fans) be worried about coming back to earth in close games? Furthermore, what led to last year's win disparity in close games, and can it be repeated?

In an attempt to see the sustainability of this success, I first went to the one constant at Villanova for the past decade-plus: Jay Wright. If luck is somehow being manufactured by his system, it should show across the years. There are VERY few examples of teams bucking the norm for prolonged periods of time, and it almost always is due to their system (in Belichick we trust). Below are his Wright's ‘luck’ percentages by year, as borrowed from KenPom.

Season Record Luck % Luck Rank Wins Added
2014 29-5 0.075 30 2.55
2013 20-14 0.011 149 0.374
2012 13-19 -0.097 339 -3.104
2011 21-12 -0.04 276 -1.32
2010 25-8 0.025 102 0.825
2009 30-8 0.021 115 0.798
2008 22-13 0.037 85 1.295
2007 22-11 -0.036 266 -1.188
2006 28-5 0.081 20 2.673
2005 24-8 -0.054 284 -1.728
2004 18-17 -0.049 266 -1.715
2003 15-16 0.059 289 1.829
2002 19-13 -0.055 280 -1.76

You can read here or here if you want to get a better idea of where this ‘luck’ metric comes from. In short, KenPom adopts Dean Oliver’s correlated Gaussian method, which adjusts for competition level and extrapolates from the team’s game by game efficiency. ‘Luck’ tells us how many more games we won than our overall season’s efficiency suggests we should have. Furthermore, luck is a solid alternative to the widely accepted Pythagorean method for wins expected vs. wins won.

Sorry for the mini-lecture, especially since all this data told me was what I most feared: luck, as applied to Villanova basketball, has been entirely random during Jay’s tenure. Obviously the players have been variable, but system is the better indicator for predictability. In Nova’s case, system suggests they are just as likely to have another ‘lucky’ string of close wins as they are to inverting their record to 1-7. That’s enough of a flip to turn a 28-3 season into a 22-9 season, which effectively moves a team from a 2 seed to a fringe tournament team. I also planned on completely leaving out the 2012 team, because that season didn't happen, but stats-integrity won out.

Luck, as applied to Villanova basketball, has been entirely random during Jay’s tenure

Aside from injury (just kidding, our players don’t get injured), a bipolar flip should be Villanova’s biggest fear heading into the season. Regression to the mean is undeniably inevitable over time, so maintaining such a high degree of success in close games is at all times an uphill battle. However, considering the high player turnover in college basketball, system is not the only (or even main) factor to keeping up with yourself. There is undoubtedly a player element to all of this, which is much harder to fairly gauge.

Next, let us consider the actual players/lineups used in ‘clutch moments’ for Villanova last year. VU Hipsters are constantly calling this player 'clutch' and that player ‘mentally frail,’ without much to back it up (this is all hearsay, as Hilliard is da Gawd and everyone else falls into place). For this segment, I re-visited the Advanced Stats from last season, filtering out everything except plays in the second half of the game (overtime included) when the score was within five points. While focusing solely on the last few minutes of the game might be a better indicator of the colloquial use of ‘clutch,’ it provides almost no sample size and thus no real conclusions can be made. This alternative time interval is a fairer indicator of when the game is on the line, and what lineups/players are performing best. Below are the most commonly used 5 man lineups (including graduated seniors) followed by most commonly used 4 man lineups (excluding graduating seniors). They were compared by a number of metrics, all synthesized into an Offensive Rating vs. Defensive Rating differential. This gives us an idea of what lineups worked, and what lineups will continue to work (with an active fifth player plugged in). Rarely used lineups were removed, as well as any 'average' lineups that James Bell or Tony Chennault were a part of.

Note: Due to the variability of lineups, I used a ‘usage rating’ instead of ‘minutes’. This is an upscale ranging from a couple of minutes played together (usage rating 1) to a cap of 160 minutes played together (usage rating 10).

5-Man Lineup Breakdown


scroll left-right...

Lineup Player 1 Player 2 Player 3 Player 4 Player 5 Usage Rating (1-10) O Rtg D Rtg Diff
1 Kris Jenkins Josh Hart Darrun Hilliard Tony Chennault James Bell 5 1.5 0 1.5
2 Kris Jenkins Josh Hart Tony Chennault Dylan Ennis James Bell 5 2.034 0.678 1.356
3 Kris Jenkins Josh Hart Ryan Arcidiacono JayVaughn Pinkston Dylan Ennis 2 1.695 0.576 1.119
4 Josh Hart Darrun Hilliard Ryan Arcidiacono Daniel Ochefu James Bell 5 1.187 0.541 0.646
5 Kris Jenkins Josh Hart Darrun Hilliard Ryan Arcidiacono Daniel Ochefu 5 1.705 1.103 0.603
6 Kris Jenkins Darrun Hilliard Ryan Arcidiacono Daniel Ochefu Dylan Ennis 5 1 0.4 0.6
7 Kris Jenkins Josh Hart Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston 4 1.473 0.914 0.559
8 Josh Hart Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston Daniel Ochefu 2 1.314 0.8 0.514
9 Josh Hart Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston James Bell 9 1.124 0.732 0.392
10 Josh Hart Darrun Hilliard Tony Chennault JayVaughn Pinkston James Bell 8 1.429 1.053 0.375
11 Darrun Hilliard Ryan Arcidiacono Daniel Ochefu Dylan Ennis James Bell 6 1.371 1.072 0.299
12 Kris Jenkins Josh Hart Ryan Arcidiacono JayVaughn Pinkston James Bell 6 1.2 0.922 0.278
13 Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston Daniel Ochefu James Bell 10 1.074 0.852 0.221
14 Josh Hart Ryan Arcidiacono JayVaughn Pinkston Dylan Ennis James Bell 7 1.343 1.161 0.182
15 Josh Hart Darrun Hilliard Ryan Arcidiacono Daniel Ochefu Dylan Ennis 1 1.649 1.481 0.168
16 Kris Jenkins Josh Hart Darrun Hilliard JayVaughn Pinkston Dylan Ennis 2 1.197 1.236 -0.04
17 Darrun Hilliard Tony Chennault JayVaughn Pinkston Dylan Ennis James Bell 5 1.112 1.152 -0.04
18 Josh Hart Darrun Hilliard Tony Chennault JayVaughn Pinkston Daniel Ochefu 4 1.035 1.103 -0.07
19 Kris Jenkins Josh Hart Ryan Arcidiacono Daniel Ochefu Dylan Ennis 5 0.942 1.024 -0.08
20 Kris Jenkins Darrun Hilliard Ryan Arcidiacono Daniel Ochefu James Bell 4 1.096 1.19 -0.09
21 Josh Hart Tony Chennault Daniel Ochefu Dylan Ennis James Bell 5 0.942 1.06 -0.12
22 Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston Dylan Ennis James Bell 8 0.841 1.105 -0.26
23 Josh Hart Ryan Arcidiacono JayVaughn Pinkston Dylan Ennis James Bell 7 0.675 1.426 -0.75
24 Kris Jenkins Josh Hart Tony Chennault JayVaughn Pinkston Dylan Ennis 4 0.832 1.594 -0.76

4-Man Lineup Breakdown


scroll left-right...

Lineup Player 1 Player 2 Player 3 Player 4 Usage eFG% O eFG% D O Rtg D Rtg Diff
1 Kris Jenkins Ryan Arcidiacono JayVaughn Pinkston Dylan Ennis 3 0.75 0.2 2.041 0.404 1.64
2 Kris Jenkins Josh Hart Ryan Arcidiacono Dylan Ennis 3 0.6 0.16667 1.345 0.553 0.79
3 Kris Jenkins Darrun Hilliard Ryan Arcidiacono Dylan Ennis 4 0.83333 0.14286 1 0.333 0.67
4 Kris Jenkins Darrun Hilliard Daniel Ochefu Dylan Ennis 3 0.83333 0.2 1 0.4 0.6
5 Kris Jenkins Josh Hart Ryan Arcidiacono JayVaughn Pinkston 2 0.75 0.36111 1.439 0.867 0.57
6 Josh Hart Darrun Hilliard Ryan Arcidiacono Daniel Ochefu 5 0.57407 0.27586 1.176 0.699 0.48
7 Josh Hart Ryan Arcidiacono JayVaughn Pinkston Daniel Ochefu 2 0.66667 0.36364 1.277 0.889 0.39
8 Josh Hart Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston 8 0.49057 0.41818 1.157 0.868 0.29
9 Kris Jenkins Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston 4 0.67857 0.39474 1.239 0.987 0.25
10 Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston Daniel Ochefu 10 0.52027 0.46835 1.091 0.841 0.25
11 Kris Jenkins Josh Hart Darrun Hilliard Ryan Arcidiacono 3 0.64286 0.47368 1.184 1.029 0.15
12 Darrun Hilliard Ryan Arcidiacono Daniel Ochefu Dylan Ennis 4 0.75 0.4 1.225 1.087 0.14
13 Kris Jenkins Ryan Arcidiacono Daniel Ochefu Dylan Ennis 6 0.66667 0.4375 1 0.875 0.13
14 Kris Jenkins Darrun Hilliard Ryan Arcidiacono Daniel Ochefu 4 0.41667 0.33333 0.714 0.67 0.04
15 Kris Jenkins Josh Hart Darrun Hilliard JayVaughn Pinkston 3 0.5 0.42857 1.061 1.095 -0.03
16 Josh Hart Darrun Hilliard JayVaughn Pinkston Daniel Ochefu 2 0.33333 0.4 1.136 1.258 -0.12
17 Josh Hart Ryan Arcidiacono Daniel Ochefu Dylan Ennis 5 0.4 0.33333 0.712 0.949 -0.24
18 Darrun Hilliard Ryan Arcidiacono JayVaughn Pinkston Dylan Ennis 3 0.5 0.5 1 1.304 -0.3
19 Josh Hart Darrun Hilliard Ryan Arcidiacono Dylan Ennis 2 0.2 0.4 0.678 1.02 -0.34
20 Kris Jenkins Josh Hart Darrun Hilliard Dylan Ennis 2 0.2 0.45 0.59 1.008 -0.42
21 Josh Hart Ryan Arcidiacono JayVaughn Pinkston Dylan Ennis 5 0.5 0.675 0.838 1.31 -0.47
22 Tony Chennault Ryan Arcidiacono JayVaughn Pinkston Daniel Ochefu 1 0 1 0.914 1.401 -0.49
23 Kris Jenkins Josh Hart Ryan Arcidiacono Daniel Ochefu 5 0 0.5 0 0.808 -0.81
24 Kris Jenkins Darrun Hilliard JayVaughn Pinkston Dylan Ennis 3 0 0.41667 0 1.096 -1.1

The most promising active lineup looks a lot like one projected starting lineup: Arch, Jenkins, Hart, JVP, and the people’s champion Ennis.

Big men are overrated anyway. In limited minutes, this squad had a sizable 1.119 difference between offensive rating and defensive rating. To no one’s surprise, Chief appeared in only 2 of the top 10 lineups. Corollary and I covered the reasons why in his shot chart, and it mainly boils down to: bad fouls, bad foul shooting, and clogging up spacing necessary at the end of the games.

All you can do is consistently put yourself in a position to win them and hope you win the coin flip.

The worst active line up was Jenkins, Hilliard (uhhh?????), Arch, Chief, and Ennis. As we have discussed ad nauseum, Ennis did not engage in many productive basketball-related activities in the 2014 calendar year. In order for the Wildcats to succeed moving forward, they need better production out of this line up, especially in crunch time. Part of that must come from Ennis’ maturity and increased efficiency. This means never shooting a (contested) three ever again. The improved spacing allowed by Jenkins upcoming 3 point explosion will only help matters, and Ennis developing into a drive and dish threat will be a cherry on top. But it is expected that this lineup will see extended minutes this season, so improved efficiency on both ends of the floor is a must.

The 4 man lineups are similar in rank to their 5 man analog, minus one star (Bell) and one plug (Tony C). Again, the most used lineup consisted of Chief, Arch, Hliliard, and JVP. This lineup had middle of the road success, however, with average offensive (1.09) and defensive(.841) ratings. The biggest cause for the offensive struggles was likely the floor spacing issue with Chief and JVP on the floor together. Jay loves putting them together, and they have shown signs of above average chemistry. However, each live in the same areas of the floor, and JVP’s presence generally forces Ochefu even further away from the basket. This does not help the flow of the offense. On the defensive end, none of the four are above average defenders at this juncture, but all are capable and should form a more cohesive unit this year. There is far too much length among these four to not be wreaking havoc on defense. Improvements must be made, or the starters might not end up as closers.

These charts mainly allow for self-extrapolation, but also give decent insight on what worked and what didn’t during the most vital parts of the game.

Lineups, and their respective cohesion, matter. But basketball at its most empirical form is one player shooting over/defending another player, and there is no doubt that certain players/attitudes are better suited for the ‘moment.’ I consider ‘clutch’ more than just specifically making a big three that everyone remembers; even if you missed the last three ‘clutch’ shots you took. Feel free to disagree with this definition... but don’t.

Winning close games on the offensive end comes down to drawing fouls, making free throws, controlling the clock, taking and making good shots, and preventing turnovers. Defensively, it boils down to not wasting a good possession with a bad foul at the end of the shot clock, rebounding, and causing turnovers. These are the tangible necessities for sustained success in close games, and Villanova did almost all of them well throughout 2014 (free throws are somehow still an issue). There is definite variability to winning/losing games at the last second, but maintaining success in close games is almost always dependent on these basic basketball fundamentals. Darrun Hilliard made a game winning floater to beat Seton Hall in the Big East tournament. Unfortunately, Sterling Gibbs returned the favor before the buzzer sounded. That is the nature of close games. All you can do is consistently put yourself in a position to win them and hope the stars align.

Player PTS 2PA 3PA FG% 3P% eFG% FTA FT% AST TO A/T Steals PF
Kris Jenkins 24 6 10 0.438 0.5 0.594 6 0.833 3 3 1 1 2
Josh Hart 49 14 23 0.378 0.217 0.446 27 0.593 4 3 1.33 3 21
Darrun Hilliard 107 42 32 0.432 0.406 0.52 43 0.698 26 23 1.13 12 25
Tony Chennault 11 7 0 0.286 NA 0.286 11 0.636 4 5 0.8 1 17
Ryan Arcidiacono 86 23 40 0.381 0.325 0.484 32 0.781 27 12 2.25 11 18
JayVaughn Pinkston 117 65 5 0.529 0 0.529 59 0.729 10 8 1.25 3 21
Daniel Ochefu 53 35 0 0.6 0 0.6 18 0.611 9 19 0.47 7 20
Dylan Ennis 14 2 7 0.444 0.429 0.611 5 0.6 4 2 2 1 4
James Bell 118 37 54 0.385 0.315 0.478 41 0.756 17 16 1.06 10 31

Keys for Success

  • Villanova will need Kris Jenkins to continue his above average production behind the 3 point line in these situations, and cut down on his turnovers (1.0 assist to turnover ratio).
  • Josh Hart, for the umpteenth time, should not take a three point shot when it matters (22% from deep). Stick to making hustle plays and securing offensive rebounds that extend possession Josh (and also stop fouling so much).
  • Darrun Hilliard’s ‘clutch’ numbers almost exactly matched his overall numbers, which makes sense for the calmest, suavest man alive.
  • One could give Ryan Arcidiacono the typical lecture about not shooting threes in big moments, but his 3 point shooting percentage actual jumped from the listed 32.5% to a whopping 50% in the last 2 minutes of close games. So he is spared!
  • Among players with high volume, JayVaughn Pinkston easily had the highest eFG%, which substantiates many claims of the need to have a decent sort-of-big man to end long scoring droughts in close games.
  • Daniel Ochefu had very limited minutes in this part of the game, as he played the least amount of crunch-time minutes of any starter in recent memory. Time will tell if this changes this year, but Villanova did just fine without him at the end of games.
  • Lastly, James Bell did an immeasurable amount for Villanova basketball and was a huge part in resurrecting the program from the oil spill caused by the class of 2013. However, there is no denying the stats: Villanova is not exactly losing their most important late game player. We learned this at the worst possible time last year (Seton Hall and UConn), but the writing was always on the wall. The fact Villanova won so often with so little production from its high scorer in close games is remarkable, and a positive sign for future success. The largest contributors with the game on the line all return for the 2015 season, and building on what they did last year is possible.

Ultimately, what made Villanova so successful last season was a combination of all of these factors.

  • Having freshmen (Hart and Jenkins) with basketball wisdom beyond their years and making vital contributions.
  • A player like Arch (at times painfully) never losing his shooter's mentality, and calmly stroking a game-shifting three. Chief throwing down a breakaway dunk to re-energize his team and crowd.
  • James Bell making a three, then another, then another to start the second half.
  • Coach Wright finding the rotations that worked, and for the most part, sticking with them.
  • Hart entering the game at a turning point and diving into the stands.
  • The team breaking out of long scoring droughts on the shoulders of their workhorse JVP.
  • Timely shooting, very few turnovers, and re-discovering ball movement at just the right time.

So many pieces played major roles in what led to Villanova's memorable season, and this shined through in the stats. It's a huge gray area in basketball as to what influence this tangible/intangible balance actually has, but there's no doubting fans would welcome more of the same. It is hard to distinguish between what led to being 7-1 (and 28-3) and what is remembered because they were 7-1 (and 28-3). The fact of the matter is that teams with good records are going to have similarly good records in close games. But the numbers seem as clutch as the team looked at times, so maybe repeating them isn't too far-fetched. Chemistry uses numbers too, sort of.

Few Cinderella seasons have ended with as loud a thud as Villanova's in 2014. When they weren’t blowing through the Big 5, they were winning close conference games by any means necessary. Common knowledge suggests the same close calls that fans began to take for granted last year can not sustain themselves into this season. While probability is indubitably ubiquitous, putting the right players on the floor in crunch time and optimizing what ekes out wins is completely under Villanova’s control.

Building off of the things that won these close games last year, both as a team and individually, is an attainable goal.

Defining an idea as abstract as clutch-ness is impossible, but advanced stats give a peek at what lineups generally work (and suggest why some work better than others). The conclusion is: there is no conclusion. So in order to have similar regular season success in 2015, and extend this success into the postseason, Villanova must continue doing what works best when it matters most.

That would be... clutch.