Nate Oats discussed the integration of analytics in his philosophy, game planning and in game adjustments in the elite eight post game conference.
A few days later Kirk Golsberry released this image which generated a reaction from Kevin Wildes.
The interview of Nate Oats and the marketing plots created by Kirk Goldsberry explain the evolution that has taken place in basketball, especially in the NBA.
To generate the plots, Goldsberry has to take shot chart data and generate shot charts by placing shots into bins. The methodology will be explained in Golsberry book but to provide context, this is what a shot chart looks like in the 1990’s.
And this is what a short chart looked like last year.
These are end of year shot charts for the Chicago Bulls for two different seasons. The raw data is plotted and differentiated between wins and losses, and we can see that there is more diversity in shots taken for both eras but there has been a decrease in shot locations in today’s basketball than in the past.
If we were to place the data into buckets, the denser areas will remain and the least dense areas will fade out.
Nate Oats’ discussion about expected value however does generate a raw shot chart where there is a moat of white space between the restricted area and the three-point line. The diagram on the right created by Kirk Goldsberry could be an actual shot chart for every Alabama Crimson Tide game under Nate Oats.
To analyze how Alabama plays and what Alabama needs to do to beat Connecticut we have to explain what expected value is.
Expected Value
The simplest way to explain expected value is to think of currency. Money has a value, and the value determines what a costumer can purchase.
Expected value gives shots taken from different locations their value. When coaches use the term efficiency in the context of shot selection, what they mean is taking a shot with the highest value.
In the 1990’s the shot with the highest value was a wide-open shot or a shot close to the basket because the probability of making those shots is greater.
In the era of analytics, the shot with the highest value is a shot that with the highest point per possession or (PPP).
An efficient shot is a shot with an expected value of 1.0 or greater. A shot that is not efficient is a shot with an expected value less than 1.0.
Coaches who do not believe in analytics, if given a choice between a wide-open midrange shot by a player that shoots the shot at 49% or a wide open three-point shot by a player who shoots the shot at 34%, would prefer the midrange shot at 49%. The shot is closer, and it is shot at a higher percentage. It’s an ideal shot. A great shot.
When we apply expected value to a midrange shot at 49% and a three-point shot at 34% the expected value is greater for the three-point shot than the midrange shot. The three-point shot has an expected value of 1.02 and the midrange shot has an expected value of .90.
To calculate the expected value the formulas are as follows:
Two-point shot: 2* percentage of shot in decimal form.
Three-point shot: 3 * percentage of shot in decimal form.
If the midrange shot is made, a team scores two points. The maximum value of the shot is two-points. 2 * .49 = .90
If the three-point shot is made, a team scores three-points. The maximum value of the shot is three-points. 3 * .34 = 1.02.
Over the long run, if you take shots with a higher expected value then shots with a lower expected value, you will score more points.
EXAMPLES OF HOW EXPECTED VALUE PLAYED OUT IN VARIOUS NCAA TOURNAMENT GAMES
The table breaks down Alabama’s shooting in their elite eight game against Clemson.
Alabama’s lowest expected value against Clemson was 1.0. The 1.0 value is from shots taken at the rim. Nate Oats allows one player to shoot a midrange shot but everybody else either shoots a three-point shot or an at the rim two-point shot (dunk or layup).
The free-throw line had the highest expected value, and we will talk about how the free-throw line fits into Nate Oats’ strategy but the shot that damaged Clemson was the three-point shot.
The table is Arizona’s expected value of their sweet sixteen game against Clemson.
Arizona shot 5 of 28 from the three-point line for an expected value of 0.54. Arizona shot themselves out of the game against Clemson in the sweet sixteen.
Arizona had a 1.36 expected value at the free-throw line which is good, and Arizona’s three-point shooting is more problematic because Arizona was getting to the free-throw line more than Clemson. Arizona shot 26 free-throws and was getting layups at the rim.
Arizona’s failure in the sweet sixteen is never adjusting based on the data.
Nate Oats uses a third-party analytics firm which keeps data and feeds analysis to the coaching staff. I’m not sure if Arizona does the same but if you are shooting the three-point shot at an expected value that is not close to 1.0, it is the coaches’ duty to adjust and demand for different shot selection.
Patterns in the Data
Three-point shooting will be key if Alabama wants to beat Connecticut and FTA will be a key indicator of how Nate Oats is adjusting to what is happening on the floor.
The bar chart is a high-level view Alabama’s three-point shooting. The chart is not a distribution chart but what we can see by the layout is that there are more crimson color bars to the right than on the left and there are more gray bars to the left than the right.
This bar chart compares the average number of free-throws overall versus the average number of free-throws in wins and losses.
The dashed line marks the location on the plot showing the overall average. When the line is extended the only bar that surpasses the dashed line is the average for losses.
The takeaway from both plots is that an adjustment that Alabama makes when the game is not flowing towards a positive outcome, the adjustment is to be more aggressive and get to the free-throw line.
To explore if the takeaway is correct, the plot compares the correlation between pace and free-throw attempts in wins and losses.
Alabama wants to play at a high pace and to generate open three-point shots, it is easier to get a wide-open look in transition than against a set defense.
The lines in the plot represent the slope between pace and free-throw attempts for wins and losses. The slope is steeper for losses than wins. The caption provides the correlation score between pace and free-throw attempts for win and losses. A correlation of 0.44 is greater than 0.21 which means that the correlation between pace and free-throw attempts is greater during losses than wins.
The higher correlation could mean that the shooting numbers from two and three-point range are not up to par and to get the shooters free, Alabama is emphasizing getting the ball down the ball quicker and attacking the rim to get to the free-throw line.
On Saturday, if Alabama is playing fast and getting to the free-throw line at a high rate it does not mean that Alabama is going to lose. But it does mean that there is a correlation, and it could be a sign that the game is not flowing in a positive direction because of bad shooting from lack of makes close to the rim or the three-point line or both.
The benefits of getting to the free-throw line is that opposing players are accumulating fouls and that hinders their minutes. Fouls on key Connecticut players can impact the outcome of the game. Getting Clingman in foul trouble is beneficial for Alabama but that could be done without a sustained effort to get to the free-throw line.
Patterns with Pace and Field Goal Attempts
The post season is problematic for fast-paced teams because post season games tend to slow down. Alabama has played four games in the post season with two games played above Alabama’s pace average and two games played below Alabama’s pace average.
Alabama has played all games above a pace of 70 which means that Alabama has not been slowed down and the games have not been played at a crawl.
Not many teams have slowed Alabama down in the 2023-2024 season. Connecticut will push the ball, but they are selective and methodical. Their offense is based on ball movement and screening. The ball movement gets a lot of attention but the screening wears on teams as the game moves along. To deliver these body blows, Connecticut must use the shot clock and have movement.
All of this means that Connecticut and Alabama are a contrast in styles. If Alabama can get the game played at a fast pace, then Connecticut is in for a fight because transition three-point shots will be available for Alabama.
Alabama has 11 losses in the 2023-2024 basketball season. The bar plot tracks the two-point expected value for all of Alabama games. Two losses out of the 11 losses took place with Alabama shooting below 1.0 from two-point range.
Alabama does not shoot mid-range shots which means that Alabama missed a lot of close-range shots to have an expected value below 1.0.
The bar chart tracks the three-point expected value for all Alabama games in the 2023-2024 season.
The difference between both bar charts is that Alabama has lost more games where the three-point shot has a lower expected value than 1.0 from three-point range than two-point range. 8 out of 11 Alabama loses took place with Alabama shooting the ball below 33% from the three-point line. 33% is the desired percentage to get 1.0 ppp.
The scatter plot highlights clusters by drawing circles to demonstrate differentials. The clustering technique is a machine learning algorithm that takes data and finds patterns in the data which are then used to make predictions.
We did not apply a clustering algorithm to the data, but we did use the clustering visualization technique to highlight the importance of expected value, pace and their relationship with wins and losses.
In the two-point expected scatterplot we see the two losses where Alabama had an expected value less than 1.0. The plot also shows that Alabama lost a cluster of games where the expected value for two-point shots was at 1.1 or above.
The cluster for three-point expected value captures the majority of the 8 losses within the first circle. One loss is right at the border of 1.0 and another loss is beyond it.
The takeaway is that Alabama needs to hit three-point shots to win. Alabama has very few games where their two-point shot selection has an expected value below 1.0 but despite the two instances, there are plenty of games where they shot well from two-point range and could not win the game.
The visualization adds to the previous plot by providing the slope of the plots for wins and losses.
The correlation scores are provided for wins and losses.
A negative correlation means that the relationship has an inverse affect between the two variables. As one increases the other decreases and vice versa.
In the three-point expected value, as pace increases, three-point expected value decreases. What is interesting is that in wins, the correlation is -0.34 and in losses, the correlation is -0.11. The correlation is greater in wins then losses.
There are a lot more data points for wins than losses. Margin of victory or the variance in scores can play a role in a greater negative correlation. In losses, it’s possible that the pace slowed down because there was a greater emphasis on taking better shots or working the ball more to get a better look. More ball movement slows down the pace.
If we think about this logically, Alabama wants to get up and down. If Alabama is hitting three’s and playing at a high pace, they are likely to continue the pace and the shot selection. If Alabama is not hitting three-point shots, then the pace is likely to slow down as they work to get closer to the rim and get to the free-throw line.
Alabama wants a higher negative correlation because it means that they are playing fast and making shots.
The Problem that Connecticut Centers Present
Replicating this table is Alabama’s greatest fear. The expected value for two-point shots is the problem that the Connecticut centers present. Connecticut has size at the center spot. Donovan Clingan is 7’2” and Samson Johnson is 6’10”. Clingan moves his feet exceptionally well for his size and Johnson has springs at the bottom of his feet. Clingan and Johnson destroy shots at the rim.
Clingan’s presence at the rim allows UCONN defenders to be aggressive on the perimeter. Aggressive play on the perimeter allows a team to challenge three-point shots and to force three-point shooters off the line.
Illinois kept attacking UCONN at the rim, but Illinois biggest problem was not their stubbornness in challenging Clingan. The problem for Illinois was their stubbornness in attacking Clingan in one motion.
Illinois also failed to use the jump stop and shot fake. Attacking the rim is a violent act and what causes the defense problems is a violent act followed by a stop and a fake. The change in pace makes it difficult for shot blockers to figure out the cadence of a shot. Defenders misjudge the shot and leave their feet. The offensive player is under control and not creating the contact and being under control without initiating contact means that if the defense contacts the offensive player, it will lead to a foul.
Another benefit of the jump stops and shot fake is that it allows the offense to pass the ball back out. When a player attacks with the dribble, defenses tend to collapse. Having a shot blocking center keeps the defense from collapsing but an extended defense is problematic because it opens itself to slot cuts.
A slot cut is when an offensive player cuts to the basket on a dribble penetration from a different angle. If the defense is extended and sticking to three-point shooters in the perimeter, the gap between the paint and the defender is greater which creates cutting angles.
Illinois did not utilize any of these counters. Everything was off one pass or no pass. All challenges to the rim were at the same speed and offensive players jumped from a great distance which meant that the player with the ball was not attacking the rim with altitude in their jump.
Alabama will have the same problems as Illinois because of the UCONN centers. Alabama takes two shots: three-point shots and layups.
If Alabama is not making three-point shots, they will attack the rim and they must have a much greater two-point expected value than Illinois.
Alabama can beat UCONN if they have a three-point expected value greater than 1.0.
Psychological Effects of Three-Point Shot Raining On Your Head
There has been analysis stating that Alabama must shoot over 40 three-point shots to beat UCONN.
Alabama does not need to shoot 40 three-point shots to beat UCONN, but Alabama must make three-point shots at an effective rate.
Efficient three-point shooting puts a lot of pressure on the opposing team’s offense. External pressure on the offensive end causes turnovers and bad plays. Coaches can deal with external pressure. Coaches have solutions for dealing with defensive pressure however, coaches do not have answers for internal pressure.
Internal pressure comes from within. Internal pressure is doubt. When a team hits three-point shots at an efficient rate, it maximizes internal pressure for the opposition. The offense must keep pace and must execute at an elite level.
UCONN executes their offense at an elite level but in most cases, there has been separation between UCONN and their opponent. The separation maximizes the internal pressure on the opposition and decreases the internal pressure on UCONN.
If Alabama shoots the three-point shot efficiently there will be no separation. The lack of separation also creates pressure. Efficient three-point shooting also allows for a faster pace because Alabama can come own and take a three-point shot without worrying about the shot not going in. This means that UCONN must scramble and hustle to get out to Alabama shooters and that opens the floor and allows for slot cuts and backside cuts in areas where Clingan is not at.
Revisit Kirk Goldsberry’s Visualization
If you follow the rabbit hole for Kevin Wildes reaction to Kirk Golsberry’s visualization, a lot of the conclusions are negative.
The general sentiment is that the NBA must do something to change how the game is played. The only realistic change that the NBA can do is increase the level of physicality that defenses are allowed to play with.
Playing with more physicality will bring down the NBA final scores because teams will have to work more for an open shot. Teams will have to move the ball to be able to take shots that are efficient but playing with more physicality will not change the output of shot chart data.
Shooting charts will look the same and Alabama is the prime example of this truth. The college game is more physical than the NBA game. College basketball does not have a three-second defensive rule like the NBA does. College Basketball can play traditional zones where players camp out and guard a spot and do not have to play the tag game with the paint that NBA zones must play.
Despite the increased physicality and the zone, Alabama is in the final four utilizing the analytical philosophy that create the shot chart in Kirk Goldsberry’s visualization. Alabama is creating these shot charts with raw data and without putting the data in bins. Alabama has had remarkable success due to the guidance of analytics.
The only way that Alabama can win UCONN is by replicating Kirk Golsberry’s visualization for modern basketball because it’s the only way that Alabama plays and the sole reason why Alabama is in the final four.
The major problem that Alabama faces is that UCONN is built to beat Alabama’s style. UCONN is also built to beat any style of basketball.
The purpose of this analysis is to figure out if Alabama could beat UCONN by adjusting their game strategy and attack the rim and get to the free-throw line. The answer is no. Alabama cannot win if they do not hit three-point shots and they do not play fast. Alabama could win if they hit threes and play at a slower pace but Alabama’s pathway to victory is to be efficient from the three-point line.
The key word is efficient. An expected value of .54 from the three-point line is not efficient. Alabama cannot beat UCONN if they shoot 40 threes or 20 threes if the expected value of those three-point shots is less than 1.0.
0 Comments