ACC Late Game Shooting Volume and Efficiency

    After diving into why Cole Anthony couldn't save North Carolina from a disappointing season, we remain in the ACC for this next article. This time, we'll look at the conference as a whole regarding players' end of game shooting volume and efficiency. 

   We'll first look into shooting volume which I thought would be interesting because teams face a tough choice in crunch time: who do they trust with the ball to go out and win them the game? Some opt for a committee approach, whereas others almost exclusively count on a few go-to guys. The graph below focuses on the top 20 ACC players who took the highest volume of shots in the final 2 minutes of the game this past season. Free throws were excluded, so only field goal attempts count. This will give us indication of each team's end of game scoring load distribution. Also, if you count closely, there's actually 22 players represented instead of 20. This is because there was a three-way tie for 20th on the list.

Note: I used Luke Benz's ncaahoopR package in R to compile the data.

Before moving into any analysis, I made the following list to also include each player's school for more clarity:

Andrien White - Wake Forest

Brandon Childress - Wake Forest

Chris Lykes - Miami

Cole Anthony - North Carolina

Dane Goodwin - Notre Dame

David Johnson - Louisville

Elijah Hughes - Syracuse

Garrison Brooks - North Carolina

Harlond Beverly - Miami

Isaiah Wong - Miami

Jalen Cone - Virginia Tech

Jay Heath - Boston College

Jose Alvarado - Georgia Tech

Kihei Clark - Virginia

Landers Nolley II - Virginia Tech

Markell Johnson - NC State

Michael Devoe - Georgia Tech

Prentiss Hubb - Notre Dame

Tevin Mack - Clemson

Tre Jones - Duke

Trent Forrest - Florida State

Trey McGowens - Pittsburgh

   

     At first glance, Brandon Childress' total jumps out at you as he took 44 of his team's shots in the final 2 minutes, good for almost 1.5 per game. Wake Forest clearly had faith in their senior to get it done. They also trusted Andrien White, another senior, despite him only averaging 9 points per game. He joins David Johnson (6.3 ppg), Harlond Beverly (7.2 ppg), Isaiah Wong (7.7 ppg), and Jalen Cone (8.0 ppg) as players on this list averaging under 10 points per game. Clearly, these guys weren't necessarily always the top scorers for their teams. 

    

    Aside from Wake Forest, another interesting team to look at is Miami. They have three players (Chris Lykes, Harlond Beverly, and Isaiah Wong) on the list indicating that not many other players got a look besides these three guys. It's also interesting that they trusted not one, but two freshman in Beverly and Wong when it mattered most. The three player dominance contrasts with a team like Louisville who had David Johnson barely on the list, leaving plenty of clutch shot opportunities to be spread out among several other players.

    

    The above graph was good for telling us which players were entrusted to deliver in the final minutes of the game, but it doesn't tell us if they were effective in doing so. After all, shooting 50 shots at say 20% is far less valuable as 30 shots at 55%. We again turn to effective field goal percentage to help us determine efficiency. Effective field goal percentage is more telling than the traditional field goal percentage because it weighs three-pointers more heavily as they're worth more points. This is especially true in late-game moments. The graph below takes these same 22 players and compares their end of game shot volumes to their eFG% in the final 2 minutes.

    Brandon Childress stood out in the earlier graph, and he sure stands out in this one too. Despite shooting far more than his ACC compatriots, he was able to maintain an above average effective field goal percentage. While Wake Forest's season was one to forget, Childress was able to salvage something out of his senior season in Winston-Salem with his fantastic end of game efficiency. Shooting volume aside, Jay Heath and Michael Devoe stand out in the efficiency department. Heath's high percentage was driven by his 8 made three-pointers, while Devoe had a more balanced attack with 4 made three-pointers, 4 jump shots, and 4 layups. It'll be interesting to see if their respective teams take note of this trend, and give them more of the ball in the final 2 minutes next season. It seems like a smart move given that their effective field goal percentages can afford to take a hit with the increased volume and they'd still remain effective late-game players. Another thing to look out for is if the high late-game efficiency translates to better overall shooting efficiency for both players next season. Devoe already saw a 7.2% increase in eFG% between his freshman and sophomore seasons, and will look to continue the increase. Heath, a freshman, will be trying to make an efficiency jump similar to Devoe's heading into his sophomore year.

    

    In taking a step back from these highly efficient scorers and moving to the more inefficient ones, we see evidence that could partially explain why both Notre Dame and North Carolina failed to meet expectations this season. This is because they both had two players on this list shooting below 35% which isn't good enough to consistently win tight games. For North Carolina, Garrison Brooks shot 33.33% on 18 shots, and Cole Anthony was 28.30% on 23 shots. I was a bit surprised that Anthony was able to put up 23 shots given how he missed a significant portion of the season as mentioned in my last article. It really emphasizes how much North Carolina relied on him to be "the guy" and how freshman volatility hampered his efficiency and prevented that from happening more consistently. Notre Dame didn't have as disappointing of a season as North Carolina's, but they weren't in the NCAA Tournament picture like many hoped they would be. Late game inefficiency by Prentiss Hubb and Dane Goodwin certainly didn't help. Poor Goodwin ended with by far the worst effective field goal percentage out of any players on this list with 5.88%. He went 1-17 in the final 2 minutes across the season, including 0-12 on three-pointers. His 17 shots is a small sample size, and unluckiness certainly could've played a part. However, it's definitely an area for him to improve upon going into his junior season as he tries to help get Notre Dame back into the upper echelon of the league.

    While the small sample sizes from these players make definitive conclusions difficult, it gives us an indication of which players are tasked with making the big plays at the end of games and how efficient they are at this task. Down the road, I might look into these same areas for some of the other 5 major conferences and possibly for the NCAA as a whole.

Why Wasn't Cole Anthony the Savior for UNC?

    The 2019-2020 season was certainly one to forget for Tarheel fans. The team tied for last place in the ACC with a 6-14 conference record and 14-19 record overall. This was their worst season by far since 2001-2002. 

    However, it didn't always seem as though the season would turn out this way. Early on, the Tarheels picked up nice wins against Alabama and #11 Oregon in the Battle 4 Atlantis. Coming out of Thanksgiving break, they were sitting at 6-1 and tracking towards another dominant season with a nice postseason run. This promise seemingly disappeared with a blowout loss to #6 Ohio State in which the Tarheels shot 27.4% from the field and another poor offensive performance against an excellent Virginia defense. To make matters worse, leading scorer Cole Anthony suffered a knee injury before the next game against Wofford. The Tarheels proceeded to suffer an embarrassing home defeat to the Terriers and it appeared like they desperately needed Anthony back. There seemed to be a popular narrative that if the team could just tread water until Anthony came back, they could make a late-season NCAA Tournament push. Yet, when he finally returned to the court on February 1st, the team still struggled. Why wasn't he the savior that he was expected to be?

    The most basic place to start looking is at points.

    In this graph as with all following graphs, the light blue line represents North Carolina's entire team statistics, the black line is the team average, and the red line represents Anthony's solo statistics. Games 9 through 21 are specifically marked because game 9 is the last game that Anthony played in before his injury and game 21 is the first game he played in after recovering. As we can see, when healthy, Anthony played a large role in the Tarheel's scoring with above or near 25 points on several occasions. However, we also see that as a team, North Carolina's scoring didn't take a dramatic hit when Anthony was out. There was just as much scoring volatility as when Anthony played, and they scored over their season average of 72.2 points per game on several occasions. If Anthony was as crucial to this team as many made him out to be, we would expect a lot more below average scoring performances from the team. Late in the season and well after his return from injury, the Tarheels had several consecutive games where they scored above their average. However, Anthony was relatively consistent in his point totals after he came back and wasn't the sole cause for the team's scoring boost.

    While scoring wasn't affected too much by Anthony's absence, we'll next look at its impact on the team's shooting efficiency by analyzing effective field goal percentage over time. Effective field goal percentage gives an added value to three point shots because they give a team more points than a two-point field goal if made.

    Effective field goal percentage is something that the Tarheels struggled with as they finished the season shooting 46.4%, good for below 300th in the country. Anthony himself was slightly below this team average at 45.1%. We see in the graph that he had early season volatility followed by some struggles immediately after his return. Yet, he certainly flashed promise shooting well over 50% for a several game span towards the end of the season. This hot streak also coincided with 3 of North Carolina's 6 conference wins and the uptick in the team's total points scored. While he was out, his team remained right around their average with the exception of a great performance against Miami (more on that later). Since Anthony's cumulative EFG% was so close to the team's, his absence neither helped nor hindered the team's shooting efficiency for the most part. However, his enhanced efficiency for the stretch towards the end of the season certainly helped in his team being able to get a few conference wins.

   So far, it seems as though the team scored at roughly the same volume and efficiency level with and without Anthony. Did the way they scored change at all during this period without him? First, lets look at the volume of three-pointers taken.

    It seems like the Tarheels tried to compensate for Anthony being out by shooting slightly more three-pointers than their average until right before and through Anthony's return. This didn't pay dividends as Brandon Robinson at 36.9% and Christian Keeling at 32% were the only other players with decent three point volumes shooting over 30%. 

    While three-pointers were definitely up a little, I expected a more profound impact to be found on team assists. This is because Anthony is certainly a scoring guard and without their alpha, I thought the Tarheels would've shared the ball more with a balanced scoring load.

    This didn't happen as dramatically as I would've thought. Yes, the team was above their season average for many of the games without Anthony (especially the aforementioned Miami game where the team played very well and got a rare win). However, we see high assist totals later on the season with Anthony back, so there isn't much of an indication that the team completely reworked the offense with him out.

    One more non-graph area to look into is adjusted offensive and defensive efficiency. These metrics measure how many points a team scores per 100 offensive possessions and gives up per 100 defensive possessions. I used Bart Torvik's website for this, and the values were adjusted for opponent quality in these same metrics. North Carolina's adjusted defensive efficiency was 98.3 over the whole season and 98.5 when Anthony was out. Hardly significant. There was a bit more disparity when it came to offensive efficiency with the season-long statistic being 108.2 which then dropped to 106.3 with Anthony injured. However, a 1.9 decrease over a small 9 game sample could very well be attributed to random variation as opposed to the loss of Anthony. Yet again, there was not a completely dramatic drop as may have been expected.

    Ultimately, the reason Anthony wasn't able to come in and rescue his struggling team is because it appears as though the team didn't play that much differently with or without him. They scored about the same number of points and with roughly the same shooting efficiency. If he truly was to be North Carolina's savior, they would've seen a much more precipitous drop in scoring and efficiency than they actually did. It seems as though they were able to break up his scoring volume across several players who cumulatively had about the same efficiency as Anthony. Thus, when he came back and resumed his role as the scoring leader, it resulted in basically the same production just now from one player as opposed to several. 

    Anthony's return wasn't entirely full of disappointment as he did have a hot streak towards the end of the season where he maintained his high scoring volume but did so at a tremendously high shooting efficiency rate. This definitely contributed to his team picking up a couple of wins, and undeniably provided value. However, his efficiency struggles immediately after returning and in the final few games averaged in with the hot streak and resulted in him not being as efficient overall after his return as was needed. He didn't need to shoot quite as efficiently as during the hot streak, but he needed closer to the 50% mark each game if he were to truly put the team on his back like some expected. Of course, this is an extremely tough ask of any player, much less a freshman scoring guard.

    Lastly, the offense surely underwent some minor tweaks to compensate for the absence of Anthony. Yet, from a broad overview, there wasn't much change as the Tarheels only shot marginally more three-pointers and didn't see a huge uptick in assists. Thus, it was impossible for Anthony to come in and be the hero that many expected him to be when his presence didn't consistently change the team from a statistical perspective.

Timing of Turnovers for Top 10 Most Turnover Prone Players 2019-2020

Welcome to the CBB Statistics Blog! For my first article, I decided to look into when exactly the players with the highest total turnovers in the 2019-2020 season coughed up the ball the most. I used Sports Reference to get the following list of players:

1. Josh Sharkey (Samford) - 177 turnovers

2. Jomaru Brown (Eastern Kentucky) - 149 turnovers

3. Colbey Ross (Pepperdine) - 140 turnovers

4. Antoine Davis (Detroit) - 136 turnovers

5. Tajuan Agee (Iona) - 135 turnovers

6. John Crosby (Delaware State) - 129 turnovers

7. A.J. Lawson (McNeese State) -127 turnovers

8. Antonio Daye Jr. (FIU) - 125 turnovers

9 (tie). Ahmad Clark (Albany) - 123 turnovers

9 (tie). Holland Woods (Portland State) - 123 turnovers

Sharkey tops this list and had the highest number of turnovers since the 2007-2008 season which is when Sports Reference seemingly started keeping track. He edged out Ja Morant's 2018-2019 total by 7. The fact that a guy like Morant found himself on this list despite his pedigree emphasizes that many of the guys aren't on this list because they're bad players, but because they're so depended upon by their team to constantly make plays which inevitably leads to more turnovers.

To get the timing of when each turnover occurred, I used Luke Benz's ncaahoopR package in R. I wasn't able to get data for every game, but in most cases these are nearly complete samples. I then created histograms for each player and made the bins based on eight 5 minute sequences of the game (overtime was excluded). For instance, the first bin will show how many turnovers a player had with 40-35 minutes left in the game (the first 5 minutes of the first half). Without further ado, here are the results along with more information on the players and some brief analysis:

Sharkey led his team in scoring (18.0 ppg), assists (7.2 per game), and steals (2.7 per game) in his senior season. Unfortunately, his dominance in these areas did not translate to much on the court success for the Bulldogs as they finished 8th in a tough SoCon and saw head coach Scott Padgett get fired. The above histogram shows Sharkey was pretty consistent with when his turnovers occurred with an increase coming at the end of games. Given how much his team needed him, this is expected.

Brown led the Colonels with 18.4 ppg and added 2.7 assists per game. The team had a successful season finishing 4th in the Ohio Valley and making it to the conference tournament semifinals before bowing out to eventual champions, Belmont. Brown's turnover distribution is very interesting as a large portion of turnovers comes in the first 10 minutes of the game. There's also a very clear continual decrease as the first half progresses. This could be an area for opponents to exploit by applying additional pressure in the beginning of games before letting off when Brown tends to settle in. It's also likely an area that the Eastern Kentucky staff is targeting to minimize these early turnovers so the team isn't put in a hole at the beginning of games.

Following the trend of these players being high-scorers, Ross was Pepperdine's leading scorer with 20.5 ppg and assist leader with 7.2 per game. He helped Lorenzo Romar's squad have a decent season with a .500 conference record and pushing Saint Mary's to double overtime in the conference tournament. We see Ross had above average turnovers at the beginning of games, right after halftime, and at the end of games.

Son of head coach, Mike Davis, Antoine does it all for the Titans. He was 4th in the country this past year with 24.3 ppg which is actually down from his 26.1 ppg the year prior. He doesn't just score as he led the Titans in assists (4.5 per game) and steals (1.7 per game). He had slightly more turnovers than his average coming out of the half and had a pretty large increase in the final 5 minutes because he's certainly the guy who's getting the ball to close out games.

Iona had a somewhat down year struggling in their conference and without head coach Tim Cluess who later resigned due to health reasons. The Gaels would've snapped their 4 consecutive NCAA tournament berths streak even if COVID-19 hadn't shut down the season as they already lost in the MAAC quarterfinals. Finally we have a player in Agee who didn't lead his team in scoring and who is our first non-guard. However, he was still their second scoring option with 14.7 ppg while being their best rebounder (7.2 per game) and shot blocker (1.2 per game). Yet again we see a slight spike in turnovers after halftime and in the closing minutes.

As mentioned above, I had trouble getting data for certain games for every player, but I certainly had the most issues with Crosby and his Delaware State team. Thus, this isn't a very complete sample and not much should be taken away from it. Crosby brings us back to looking at guards and no surprise in that he also led his team in scoring after transferring from Dayton with 19.7 ppg and assists with 3.3 per game. The Hornets went winless against D1 foes until January 11th and struggled to a 6-26 record.

We head to the Southland conference for our next guy, A.J. Lawson of McNeese State. Lawson was behind forward Sha'markus Kennedy in scoring, but should take on more of the scoring load next year in his senior season with Kennedy having graduated. The Cowboys were gearing up to play in the Southland conference tournament following their 6th place finish before the season ended. We see a decent turnover increase after half and at the beginning of games. There's also a slight increase in the final 5 minutes, but Lawson has the lowest total turnovers in this 5 minute segment of everyone we've looked at so far (including John Crosby with his half-missing data).

Daye Jr. bucks the trend of everyone on this list being either the 1st or 2nd scorer on their teams as he had only the 4th highest ppg on a very balanced scoring FIU team. While he may have lacked in scoring, he made up for it with 4.9 assists per game and 1.8 steals, both of which were team highs. The Panthers were still in the Conference USA tournament and were looking for at least one more win to cement a second consecutive 20 win season. Daye's turnover histogram is certainly unique with slightly higher than average totals in the traditional first 5 minutes and first 5 after half, but the largest increases being at seemingly random segments. The 15-10 minute left interval seems too low to be normal, so this could be indication of a time where he gets his rest before the final stretch.

Clark was the points guy (16.7 ppg), assists guy (4.2 per game), and steals guy (1.6 per game) for the Great Danes. Unfortunately, he wasn't able to lead his team to the NCAA tournament in his final season as Albany was bounced by Stony Brook in the America East quarterfinals. Clark's turnover distribution is a fairly even one with a slight increase at the end of games. Interestingly, his ball security was best for the two 5-minute segments sandwiching halftime.

For the final player of the list, and who's actually tied with the above Ahmad Clark, is Portland State's Holland Woods. Woods was the Vikings' leader in points (17.7 ppg), assists (5.2 per game), and steals (2.1 per game). However, he won't be able to work on trimming down his turnovers for the Vikings as he'll be headed to Tempe to play for Arizona State presumably after sitting out a year. It's a disappointment for the Vikings who enjoyed a nice 4th place finish in the Big Sky. We see Woods with a huge drop in turnovers after the first 5 minutes, almost as dramatic as Eastern Kentucky's Jomaru Brown mentioned earlier. He also had an increase leading up to and after halftime, but finished it off with a rare slight decrease in the final 5 minutes compared to the previous 5 minute interval.

Overall Analysis:

While it was fun to look at these distributions, there's not too much that we can take away from them. After all, most "large" increases or decreases in turnovers between the 5 minute intervals were by about 5-7, which when distributed across a 30+ game season, become not very significant. In fact, the only really significant takeaway is how drastically different Jomaru Brown's turnover distribution was. His early game turnovers is something valuable for his own team to work on as well as for the opposition to prepare for. In addition, we saw the expected trend of total turnovers being higher than the average at the beginning of games, right after halftime, and in the closing minutes. This makes sense given that in the beginning of games, players have to adjust to the opposing defense. They also may struggle to shake off the rust after taking their halftime break and/or have to deal with any defensive adjustments by the opposition. Lastly, the ending of games is always chaotic and given these players' importance to their teams, it's a given that they'll make some mistakes with the ball being almost constantly in their hands at the end. Given that many of these players are "the guy" for their respective teams regardless of how poor they play, an interesting follow up to this analysis would be to look at the distributions for the top 10 power conference turnover leaders. These power conference teams have the depth to pull a guy who's coughing up the ball a lot early, so their distributions may be more skewed to the right.