PHONE FAX WEB
QUANTIFYING PLAYER DECISION MAKING
By Norman de Silva
Data Tracking Credit: Kevin Owens, Ford Higgins & Trevor Bergwall
April 6, 2015
Quantifying Player Decision Making
BACKGROUND
“The most difficult thing in all of basketball is knowing when to shoot and when to pass.”
-‐ Oscar Robertson
DEFINING DECISION PARAMETERS: GOALS, URGENCY & LOCATION
This study is aimed at finding measurable and meaningful methods which can help improve the most important factors in successful basketball: Effective Field Goal Percentage, Field Goal Percentage at the Rim and Assisted 3 Point Percentage.
All decision making by any player on the floor can be thought of as having a different level of urgency. If you have the ball, the urgency is usually higher than when you don’t. The urgency level usually defines the amount of time that a player has before a decision needs to be made (much like how urgency level increases while a 5 second count approaches during an inbound situation). In less urgent situations, the offensive player is afforded more time before he must decide what to do.
It can be accepted as a general rule that the closer the ball is to the goal or rim, the more dangerous the threat of a score is to the defense. The closer the ball is to the rim, the more defensive help is required to lessen the threat of a score by the offense. This is usually the most common way to put the defense at an immediate disadvantage. Coaches consistently reinforce this truth by preaching and stressing the importance of not only penetrating deep into the defense and putting “pressure on the rim”. No matter how you look at it, the urgency of all decisions ramps up as you get closer to the rim because you are requiring more immediate defensive attention and the player has less time before facing a negative outcome.
One of the most statistically important factors in both offensive and defensive efficiency is converting (FG%) at the rim.
*If you aim to have a high field goal percentage at the rim, one often overlooked factor is: knowing when NOT to shoot can be just as important as how well you finish at the rim. This is often never accounted for in analysis.*
Before we can measure the effectiveness of shooting or not shooting we must define a measurable decision area. The area where it can be seen that this heightened decision making process really intensifies is a common location on the court known as “the nail” or at the center of the floor, 15 feet from the front of the rim. Inside this 15 foot region can easily be referred to as a “paint touch” for practicality. Therefore this imaginary arch is the threshold where we begin to measure the players decisions based on his actions once in the “paint touch region”.
The other type of meaningful movement involved in decision-‐making is horizontal movement, which is measured in “thirds” of the court. We can then begin to measure not only how deep the ball has gotten or not gotten, but also how much the defense has had to move from side to side before the possession is used.
Both vertical and horizontal movement will help us derive conclusions about both the team and its players’ decisions. We will aim to find optimal decision times, measured by both metrics, as they apply to the team. We will also aim to find specific results and optimal levels for each player. This type of data is a much more usable form which coaches can
Quantifying Player Decision Making
MEASUREMENT METRICS
THIRDS & PAINT
When deciding how to measure ball movement at the beginning of the season, the process caused me to be cognizant of what is and isn’t “relevant” when it comes to moving the ball. Some ball movement can be purposeful while other movement can be meaningless. It is near impossible to categorize all ball movement into a category that is either good or bad but there are some things that are present and recognizable in most useful ball movement.
One of the purposes of ball movement is to force the defense to move from their desirable and current position, to some other position where they currently are not. Forcing defenders to move causes openings, gaps, and other opportunities that were once unavailable when the defense was in their desired position.
Some of the most dangerous and valuable ball movement is vertical movement of the ball (penetration). We were charting this through what we called “paint touches.” This is when the ball is advanced to within 15 feet of the rim or within an imaginary 15-‐ foot arch, which is helpfully visualized by the paint. What this doesn’t account for is horizontal movement of the basketball. Coaches all over the world preach ball movement and specifically “ball reversals” in order to “turn the defense over.” In sticking with our organizational terminology, I chose to measure the floor in “Thirds”. There is a “Middle Third” which is in between the lane lines extended up to half court and then there are the two “Outside Thirds” which are between the lane line and sideline, again extended up to half court.
When quantifying the horizontal ball movement, I decided to value each possession based on how many thirds of the court the ball entered into on any given possession. This score could be unlimited within the amount of movement physically possible, given the 24-‐second shot clock.
-‐ If the ball was brought over half court on an outside third, and never left that third, the possession was scored for having 1 third.
-‐ If the ball was brought over half court on an outside third and was reversed through the middle to the opposite third and the possession was then used, this would be considered 3 thirds that the ball existed in during that possession. -‐ If the ball was brought over half court on an outside third, was reversed to the opposite outside third and then back
again to the original third, this would be considered a possession with 5 thirds. This is because the ball entered 5 “new” thirds during the possession with the initial third being the first.
Quantifying Player Decision Making
PAINT
SUCCESS LEVELS WITH PAINT TOUCHES
Each possession this year was documented in binary fashion with either a 1 or a 0 as having a paint touch or not having a paint touch. With this very simple yes or no tracking statistic, the results between the two categories were drastic. It can be believed that vertical penetration closer to the goal is more valuable when used correctly than horizontal ball movement on the whole. In our tracking, we were able to sort between half court possessions and transition possessions. The following numbers include both all possessions and then also specifically just half court possessions against a set defense. As can be expected our transition possessions, against an unset defense, were more effective and brought up the overall numbers despite transition only accounting for 27% of our possessions.
TOTAL
SUCCESS
SCORE RATE
PTS
PPP
PAINT TOUCHES
3145
1872
59.52%
3865
1.2289
NO PAINT TOUCH
1730
486
28.09%
1258
0.7272
HALF COURT PAINT TOUCH
2267
1267
55.89%
2670
1.1778
HALF COURT NO PAINT TOUCH
1360
377
27.72%
963
0.7081
Quantifying Player Decision Making
THIRDS
SUCCESS LEVELS IN CORRESPONDING THIRDS
When looking at the data compiled throughout the season, we were able to pull the results for every possession and assign those results to the number of thirds that the ball entered in to for each of those possessions. From there, we were able to analyze the level of success we had in each number of thirds. Below is a table showing the results corresponding with how many thirds the ball moved in to. We are able to see the following:
-‐ Number of Thirds.
-‐ How many possession we achieved that exact number of thirds.
-‐ The number of times we were successful when the ball reached exactly that number of thirds. -‐ The rate of success in that number of thirds.
-‐ How many points we scored when reaching that number of thirds. -‐ The points per possession when reaching that number of thirds
THIRDS
TOTAL
SUCCESS
RATE
PTS
PPP
1
293
75
25.60%
85
0.2901
2
1086
478
44.01%
617
0.5681
3
1007
450
45%
659
0.6544
4
707
356
50.35%
505
0.7143
5
315
158
50.16%
235
0.7460
6
133
79
59.40%
107
0.8045
7
59
31
52.54%
44
0.7458
8
15
10
66.67%
16
1.0667
9
5
4
80.00%
5
1.0000
10
4
1
25.00%
3
0.7500
At first glance, the data seems to be somewhat obvious. The more ball movement we got, the better our final result in that possession was. What starts to become apparent is that there is a tipping point where the more and more ball movement we get, we eventually come to a point where we fall off a “cliff”. What this means is that ball movement is great but if you aren’t taking advantage of it or if you are unable to couple it with meaningful penetration into the defense, then it soon becomes meaningless and self-‐perpetuating. The question then becomes, “are you moving the ball with purpose or just moving the ball?”
The more you move the ball side to side, the more you are moving closer to a shot-‐clock violation or some other type of potential turnover. That aside, your likeliness of finding a better shot than the first available, is also increasing. Your risk-‐reward for that possession is increasing over time. It is certainly a line worth walking but the question also becomes, “to what point?”
Quantifying Player Decision Making
The previous data and the “Possession Cliff” is further visualized below.
It can be seen that once we begin to enter 7 thirds, the rates take a small dip and the amount of data for thirds 8 and up is not enough for us to assume the results are stable. There seems to be a sweet spot forming which we will discuss later.
Quantifying Player Decision Making
COMBINING BALL MOVEMENT & PENETRATION
PAINT & THIRDS IN RELATION TO EACH OTHER
Now that we have established the relative importance of ball movement and penetration (thirds and paint touches), we can now move a step forward and talk about the results when we do both. We can then compare that to the results of when we only do one of the two or none at all. Below is a chart encompassing the data from the entire season that shows how often we score when we get different levels of ball movement along with or without a paint touch.
NO PAINT TOUCH:
PAINT TOUCH:
THIRDS &
PAINT P & SUC. TOT. T&P T&P SC. RT T&P SUC. RT
1 25 46 54.35% 0.54 2 378 690 54.78% 0.55 3 333 604 55.13% 0.55 4 304 544 55.88% 0.56 5 127 220 57.73% 0.58 6 63 101 62.38% 0.62 7 23 41 56.10% 0.56 8 8 12 66.67% 0.67 9 4 4 100.00% 1.00 10 1 4 25.00% 0.25
It soon becomes obvious that both of these metrics are powerful in their own right, but when used together in concert with one another, the success rates are significantly higher. At different points during the season, Coach Young would encourage me to put these numbers on the board due to their visual power and the obvious message they send. An example of how we would display this data to the players in a simple fashion is given below. A ball reversal is any possession with at least 5 thirds, signifying
that the ball changed sides of the floor and then came all the way back to where it started.
NO REVERSAL &
NO PAINT TOUCH
25.53 %
BALL REVERSAL &
NO PAINT TOUCH
34.94 %
NO REVERSAL &
PAINT TOUCH
54.93 %
BALL REVERSAL &
PAINT TOUCH
61.11 %
THIRDS &
NO PAINT NP & SUC. T&NP TOT. SC. RT T&NP SUC. RT T&NP
1 50 247 20.24% 0.20 2 100 396 25.25% 0.25 3 117 403 29.03% 0.29 4 52 163 31.90% 0.32 5 31 95 32.63% 0.33 6 16 32 50.00% 0.50 7 8 18 44.44% 0.44 8 2 3 66.67% 0.67 9 0 1 0.00% 0.00 10 0 0 #DIV/0! #DIV/0!
Quantifying Player Decision Making
Below is another visualization of the same data.
Quantifying Player Decision Making
DECISION OPTIONS
The closer the ball gets to the rim, the less time that player has to decide between a number of options. The player could shoot, pass, get fouled, turn the ball over, dribble it out, etc. As mentioned earlier, it is just as important to know when not to shoot as when to shoot when related to field goal percentage. Because of this, I have categorized all the options into what I believe to be the 2 most important possible outcomes. These two outcomes would be “using” the possession for yourself or “not using” the possession for yourself. A shot, turnover, dribble out or drawing a foul would be a use of the possession; a pass would be considered a non-‐use.
It is through these high-‐pressure moments with less reaction time that we can observe and learn how a player will
continually react over time and with what level of efficiency his decisions render. We can begin to quantify what percent of the time the player decides the best outcome is to use the possession for himself and when they decide the best possible outcome was best served by not using the paint touch for himself.
POSSIBLE DECISION OUTCOMES
Once the player has decided to use the possession or not use the possession, we can then begin to evaluate the efficiency rendered by that player’s decisions. There are a number of possible outcomes that we chose to observe. They include the following:
Points Scored
No Points
Scored
Good Shot
Apempt
Shot Apempt
Bad or No
Shot Locaqon
PAINT TOUCH
Quantifying Player Decision Making
METRICS USED
-‐ Points were measured on a basic scale of 0 to 4 possible points for each possession.
-‐ Shot attempts were graded by basketball staff members in real time who judged whether the shot attempt was a good shot or bad shot based on things like personnel, location, rhythm, level of defensive contest and other factors on a simple binary scale of 0 for a bad shot and 1 for a good shot.
-‐ Shot location was observed with regions of the court broken down into Restricted Area, Paint, Mid-‐range, Corner 3, and Above the Break 3 as well as whether or not a foul was drawn on the shot.
Some of the valuable statistics we were able to derive from this data that helps to quantify the decision making of a player and the team are as follows:
Paint Touch Score Rate = PTSc/PT Paint Touch Good Shot Rate = PTGs/PT Paint Touch Usage Percentage= PTU/PT Paint Touch Non-‐Usage Percentage = PTNU/PT
Paint Touch Usage Score Rate = PTUSc/PTU Paint Touch Usage Points = PTP – PTNUP Paint Touch Non-‐Usage Score Rate = PTUSc/PTNU
Paint Touch Non-‐Usage Points = PTP – PTUP Paint Touch Points Per Possession = PTP/PT Paint Touch Points Per Usage = PTUP/PTU Paint Touch Points Per Non-‐Usage = PTNUP/PTNU
The above metrics can be used to see many things: For instance, how often a player chooses to shoot in the paint versus deciding to kick the ball out to a teammate. Depending on the physical attributes and skills of a given player, the optimal balance between usage and non-‐usage can vary (we will calculate this optimization later on). Some players shoot too much on their possessions in the paint and their percentages from the field and at the rim suffer as a result. As defenses adjust to the tendencies of certain players this can become even more problematic. Some players will pass too much on their paint possessions. This can cause them a strong opportunity cost of potential high percentage shots. Also as defenses adjust, they may stay home when these players reach the paint and passing opportunities will become less available and the paint touch will render fewer open shots.
Quantifying Player Decision Making
Other things we can learn from these metrics are things like Success Rates and Points Per Possession on both Uses and Non-‐ Uses. Some players will show to be poor finishers if they have low results on their uses. On the flip side, this could mean they are drawing extra attention at the rim and are missing out on higher percentage open shots by their teammates if they were to look to kick the ball out more. Other players who have high results on their paint uses could prove to be good finishers or rather they aren’t shooting enough when getting in the paint and have high percentages because they are too selective.
Players with high rates and points per possession on their non-‐uses could prove to be great passers in these high pressure situations. They could also alternatively prove to be not passing the ball out enough, meaning they pass so infrequently that teams don’t bother to stay home on the perimeter and all assume the player is going to shoot. Players with low results on their non-‐uses could either be poor passers under duress or they could be over-‐passing when they get in the paint and teams don’t help on them as much.
All of these potential conclusions about a certain player have to be used in the coexistence of the eye test. Any of these individual stats alone may not tell you much of anything but when they are looked at together and relative to each other, they can be very helpful in identifying issues and strengths. Later in this study, we will show examples of this analysis in real-‐life use during the 2014-‐15 Delaware 87ers season.
Quantifying Player Decision Making
DECISION MAKING ANALYSIS
MEASURING OVERALL SUCCESS AND FAILURE
Defining success and failure in these cases is really a matter of relativity to the measurements of other players. The sample size of this study is only 1 team throughout the course of 1 full season. The down side is you are limited to a sample of players who can add to or retract from each other’s success or failure and are not independent of one another. The upside is that we have a full 50 game season of data on these players which is a significant amount of time to observe trends and accumulate reliable data on those players.
Generally speaking, Points Per Paint Touch could be considered one of the best measures to use because it encompasses the most variables while still maintaining a level of objectivity. As in most cases, anything over 1.0 would be considered good. That being said, when you are successful at getting the ball into the most dangerous regions of the defense, you would hope to convert at a higher rate than other times, making 1.0 on the low end of what you would hope for as a result from your average paint touch.
Since we do not have data on this topic from any other teams, we will have to use these numbers as the baseline for what the averages are across the board. From here we can start to evaluate where individual players fall on the respective spectrums of usage and efficiency. Below you can observe the season totals for each player. Green cells recognize an above average result whereas red cells designate a below average result. The color intensity represents how far each result is from the mean.
2014-‐15 DELAWARE 87ERS TEAM AVERAGES
POINTS PER PAINT TOUCH: 1.0922 GOOD SHOT SUCCESS RATE: 68.31% PAINT TOUCH USAGE %: 60.52% PAINT TOUCH USAGE POINTS PER POSSESSION: 1.073 PAINT TOUCH NON-‐USAGE %: 39.48% PAINT TOUCH NON-‐USAGE POINTS PER POSSESSION: 1.1217
Quantifying Player Decision Making
EXAMPLE OBSERVATIONS:
-‐ Excluding players with small sample sizes, the players with the highest rates on their paint touches this season were Jordan McRae, Drew Gordon, D.J. Seeley, Ron Roberts and Jamal Jones. It is possible that you could make the argument that these were our 5 best overall players this season.
-‐ Kenny Hall uses 90% of his paint touches. This is something that confirms the eye-‐test in the fact that he rarely ever kicks the ball out once he gets it.
-‐ Malcolm Lee and Malik Wayns have the highest Non-‐Usage percentages. This makes sense that our point guards would be kicking the ball out the most for others. While this makes sense for Lee, since he only scores a team low 0.75 points per possession on his uses, at the same time Wayns could potentially be using more of these possessions for himself since he is at 1.11 points per possession on his finishes.
-‐ Joonas Caven is a surprisingly very, efficient decision maker. He has a small sample size because he thinks of himself simply as a shooter and prefers not to drive. He was also 100% on his 5 kick-‐outs for a total of 15 points and 3.0 points per possession. He could be coached into driving the ball more often as well as using his underrated passing ability. -‐ As might be expected, Ron Roberts was our best finisher in the paint, scoring 1.32 points per possession on his paint
touches.
PLAYER PT SUCCESS PTS RATE PPP PTU PTU% PTU
PTS PTUPPP PTNU PTNU% PTNU PTS PTNUPPP
MALCOLM LEE 150 101 171 0.6733 1.1400 68 45.33% 51 0.7500 82 54.67% 120 1.4634 NOLAN SMITH 51 30 56 0.5882 1.0980 32 62.75% 41 1.2813 19 37.25% 15 0.7895 VICTOR RUDD 254 171 268 0.6732 1.0551 164 64.57% 181 1.1037 90 35.43% 87 0.9667 JAMAL JONES 108 80 110 0.7407 1.0185 71 65.74% 72 1.0141 37 34.26% 38 1.0270 D.J. SEELEY 437 330 543 0.7551 1.2426 242 55.38% 276 1.1405 195 44.62% 267 1.3692 GIDEON GAMBLE 25 16 27 0.6400 1.0800 19 76.00% 20 1.0526 6 24.00% 7 1.1667
MELVIN JOHNSON III 33 23 37 0.6970 1.1212 25 75.76% 26 1.0400 8 24.24% 11 1.3750
RAHLIR HOLLIS-‐ JEFFERSON 130 93 153 0.7154 1.1769 70 53.85% 83 1.1857 60 46.15% 70 1.1667 KENNY HALL 100 64 100 0.6400 1.0000 90 90.00% 88 0.9778 10 10.00% 12 1.2000 DREW GORDON 218 167 253 0.7661 1.1606 148 67.89% 159 1.0743 70 32.11% 94 1.3429 RONALD ROBERTS 97 72 128 0.7423 1.3196 78 80.41% 103 1.3205 19 19.59% 25 1.3158 JARED CUNNINGHAM 84 60 104 0.7143 1.2381 62 73.81% 67 1.0806 22 26.19% 37 1.6818 SEAN KILPATRICK 81 57 103 0.7037 1.2716 47 58.02% 47 1.0000 34 41.98% 56 1.6471 JOONAS CAVEN 18 15 28 0.8333 1.5556 13 72.22% 13 1.0000 5 27.78% 15 3.0000 TINY GALLON 44 26 44 0.5909 1.0000 29 65.91% 24 0.8276 15 34.09% 20 1.3333 LAQUINTEN MILES 52 33 59 0.6346 1.1346 34 65.38% 30 0.8824 18 34.62% 29 1.6111 MALIK WAYNS 177 129 219 0.7288 1.2373 73 41.24% 81 1.1096 104 58.76% 138 1.3269 JORDAN MCRAE 82 64 105 0.7805 1.2805 47 57.32% 54 1.1489 35 42.68% 51 1.4571 JOEL WRIGHT 49 33 59 0.6735 1.2041 37 75.51% 38 1.0270 12 24.49% 21 1.7500 NORVEL PELLE 5 3 6 0.6000 1.2000 4 80.00% 6 1.5000 1 20.00% 0 0.0000 TEAM 1603 1147 1846 0.7155 1.1516 1007 62.82% 1100 1.0924 596 37.18% 746 1.2517
Quantifying Player Decision Making
PLAYER PATIENCE LEVEL
DATA CAPTURING
In order to get a more encompassing measure of a player’s level of patience, we need to extrapolate from additional data, which we tracked over the course of the season. For each possession of the season, in addition to penetration via paint touches, we also tracked ball movement via “Thirds”. Thirds is a term used in our organization to break the court into three vertical segments. The middle third being between the lane lines and extended up to half court. In addition, there are 2 outside thirds extending from each lane line to the sideline, again extended up to half court. In order to track the movement of the ball, we counted the number of new thirds the ball entered in to on every possession.
Since we have established that paint touches are good, we must note that we have also established in previous reports that ball movement is a good thing, to a point, before reaching the “possession cliff”. In knowing that these two things are good for our team’s chances of success on a given half-‐court possession, the measure then becomes, how willing is the player to wait on these things to be accomplished before choosing to use the possession.
It should be noted that in this analysis, I have thrown out all possessions deemed as “transition” because in a transition situation, a player is usually better served in using the possession quickly before the defense is set because transition opportunities usually render higher percentage outcomes. The data being used moving forward will only be data pulled from half court
possessions against a set defense.
NUMERICAL PATIENCE VALUE
In determining patience, there are two variables now involved: ball movement and penetration. Unless a great opportunity presents itself unexpectedly, it is wiser to not use the possession until the offense has accomplished a quality level of ball movement or penetration which we already know increases the chance that your usage will render success.
Each player has a total number of used possessions. It can then be determined how many of these possessions included a paint touch and also how many thirds were accomplished on these possessions where the player decided he would be the one to use the possession. We can then derive the average amount of thirds are achieved before a certain player uses the possession. We can also measure how likely a player is to wait for a paint touch by himself or others before using the possession. This gives us 2 values:
PAINT TOUCHES
PER USAGE
THIRDS
PER USAGE
Quantifying Player Decision Making
The following is a chart that shows each player’s Paint Touches Per Usage and Thirds Per Usage for our entire season. PLAYER TOTAL POSS USES USAGE THIRDS THIRDS PER USE USES W/ PT PT PER USE
MALCOLM LEE 190 565 2.97 92 0.4842 NOLAN SMITH 77 243 3.16 51 0.6623 VICTOR RUDD 485 1634 3.37 271 0.5588 JAMAL JONES 214 778 3.64 131 0.6121 D.J. SEELEY 563 1680 2.98 337 0.5986 GIDEON GAMBLE 113 354 3.13 49 0.4336
MELVIN JOHNSON III 113 385 3.41 58 0.5133
RAHLIR HOLLIS-‐JEFFERSON 122 392 3.21 101 0.8279 KENNY HALL 192 584 3.04 153 0.7969 DREW GORDON 369 1135 3.08 296 0.8022 RONALD ROBERTS 125 413 3.30 106 0.8480 JARED CUNNINGHAM 176 529 3.01 102 0.5795 SEAN KILPATRICK 182 566 3.11 101 0.5549 JOONAS CAVEN 72 215 2.99 32 0.4444 TINY GALLON 67 205 3.06 44 0.6567 LAQUINTEN MILES 61 178 2.92 39 0.6393 MALIK WAYNS 148 359 2.43 89 0.6014 JORDAN MCRAE 136 419 3.08 81 0.5956 JOEL WRIGHT 57 183 3.21 53 0.9298 NORVEL PELLE 12 23 1.92 9 0.7500 TEAM 3474 10840 3.12 2195 0.6318
When considering only the players with a significant sample size, we can derive a few facts about certain players. (All extrapolations are in comparison to the team averages of 3.12 thirds previous to possession usage and waiting for a paint touch 63.18% of the time before using the possession.)
-‐ Jamal Jones can be considered one of our most patient offensive players, and on average, waits for at least 1 ball reversal before looking to use the possession.
-‐ Malik Wayns can be seen as our most impatient player in terms of ball movement and is willing to use the possession earlier than all others, after only 2.43 thirds have been reached on average.
-‐ Gideon Gamble and Joonas Caven are the most likely to use the possession without the team first achieving a paint touch. This can be sensible at times given their outside shooting ability.
-‐ Our big men, as one would expect, are our more patient players based on waiting for a paint touch, mainly because they don’t use many possessions outside the paint and often must wait for the ball to come to them.
Quantifying Player Decision Making
CASE STUDY: VICTOR RUDD – INCREASING PAINT EFFICIENCY
Through the first 22 games of the season, we began to see a trend as a staff specifically with Victor Rudd’s efficiency on his paint touches despite his exceptional athleticism. The obvious red flag led us to investigate further as to why his efficiency levels were relatively low. Upon review, I began to notice that every time Vic drove to the paint, he was committing himself to a decision before he even got to the paint. His usage percentage on paint touches was very high as Vic continually picked his dribble up before he even got to the paint in a “wind-‐up” type attempt to attack the rim as fast and as hard as possible. Defenses had begun to expect Vic to shoot it each time at the rim or attempt to dunk it. When he would be met at the rim, his finishes became increasingly more difficult as time went on.
As a remedy to this, I began to work with him consistently on taking just 1 extra dribble on all of his drives into the paint. This forced Vic to slow down just enough to realize more options and buy himself more time to make a sound decision. Adopting this new habit of taking an extra dribble and creating more time for him to decide whether to jump into contact and score or use his vision to find a teammate for an open 3, allowed him to increase his efficiency significantly. There are still other times where it is a good thing to attack the basket hard like before and it turns out these opportunities were not sacrificed because his instincts inherently kicked in at the right times. We did not want to sacrifice his usage percentage because 6’8 athlete of his caliber should be thinking about finishing when he gets in the paint, but we rather aimed to increase efficiency.
PLAYER PT SUCCESS PTS RATE PPP PTU PTU% PTU
PTS PTUPPP PTNU PTNU% PTNU PTS PTNUPPP VICTOR RUDD
(GAMES 1-‐22)
92 53 89 0.576 0.9 62 67.4% 60 0.96 30 32.61% 29 0.9
VICTOR RUDD
(GAMES 22-‐50) 152 111 171 0.731 1.2 96 63.2% 115 1.20 56 36.84% 56 1.1
Vic went from converting on 57% of his paint touches in the first 22 games to converting on 73% of his paint touches in the following 28 games after we began this emphasis in his teaching. In using the extra dribble and slowing down just a bit, Vic
managed not to sacrifice his high usage percentage (a good thing considering his personal attributes) and only went from using 67% of his paint touches down to 63%. He increased his points per paint touch overall from 0.9 to 1.2. In addition, when kicking the ball out of the paint to others, the points per possession on his non-‐uses also rose from 0.9 to 1.1 mainly due to increasing the amount of decision making time he has in the paint which allowed him to make better passes and see more of the floor.
Quantifying Player Decision Making
CASE STUDY: D.J. SEELEY – MAXIMIZING TEAM PERFORMANCE
Early in the D-‐League season it can be difficult to find out who your team is; specifically who your players are and how they are best used. As a coach, you most likely have never worked with or seen any of them play live before. In the beginning of the 2014-‐15 season, D.J. Seeley, a 2nd round draft choice, was being played as a shooting guard, off the ball. It became clear early on that D.J. was one of our best shooters from the outside and lacked both speed and quickness. Conventional wisdom tells you that with those characteristics, he should play on the wing as a 2.
After game 10 of the season, we took the time to reevaluate our team and our overall strategy. We were 3-‐7 overall and had just suffered a 6 game losing streak. One of the things that became apparent to us was our lack of efficiency in the paint. When evaluating further into which players were efficient there and which ones were not, it became glaringly apparent that D.J. Seeley was one of our best decision makers in the paint. He had the highest success rate on his paint touches of anybody on the team outside of our incredibly athletic and physical Forward/Center who finished most of everything around the rim. D.J. was finding success on 73% of his paint touches, which was second behind the leader at 77%.
The problem was that despite playing large minutes and being incredibly efficient in the paint, D.J. was only accounting for 14% of all our team’s paint touches. The majority of our paint touches (42.1% of them) were coming from our point guards while D.J. spent most possessions spotting up on the perimeter as a shooter.
As a result, the realization became that we would need to find a way to get D.J. more paint touches and more possessions overall where he was being our primary decision maker. In game, 11 we began to start using D.J. at point guard. The reasoning being that Pick & Roll ball handlers often had the best opportunity to not only get to the paint but also be responsible for major decisions on each possession. Our point guards were the players in the most Pick & Rolls by far. This also allowed us to keep the core of our offensive system in tact without changing things just to get our wings in more Pick & Rolls, driving situations or isolations.
By moving D.J. on the ball rather than off the ball, his paint touches as a percentage of our team’s total, doubled from 14.0% to 28.1%.
Quantifying Player Decision Making
Subsequently, right after this change, our team’s output began to skyrocket as a whole and we won 5 of our next 8 games. By moving DJ to the point guard position for the last 40 games of the season, our team’s success rate on paint touches went from 64% up to 70%. He also maintained his strikingly good balance of 55% Usage and 45% Non-‐Usage.
In addition, our team 3 point field goal percentage went from 32% to 39% over that stretch as we accumulated higher percentage shots from 3. Our assists per game went from 17 up to 22. Our Paint Touch Points Per Game went from 45 to 58. And maybe most importantly, our Offensive Efficiency Rating went from 103 to 114.
CASE STUDY: D.J. SEELEY – OPTIMIZING PLAYER WORKLOAD
Identifying the good decision makers on your team and putting them in those situations, is not enough to completely maximize your team’s performance and that player’s performance. When finding something that works, coaches are tempted to overuse that option, eventually forcing it to a point where the option begins to produce diminishing returns. Just as if a good shooter begins to shoot the ball on every possession, the opponent also begins to notice trends and adjust their defense accordingly to minimize the offense’s efficiency and the shooter’s shot results become less and less effective.
Applying this same principle over time, we learned that the higher percentage of our team’s paint touches that went through D.J., the lower his and the team’s efficiency became. The more we “over relied” on D.J. to shoulder the load of more paint touches, the worse our outcomes became. The initial increase helped us immensely but pushing it further became
counterproductive. This could be attributed to fatigue, over saturation, or the player forcing it too much when its not there. Just like in the previous examples, the defense adjusts to minimize the pattern they observe when the repetition becomes too frequent. It does cause an exceeding amount of energy and effort to continually get to the paint while playing high minutes and also trying to keep a sharp level of focus during decisions, which can cause diminishing returns. It could also be true that the player begins to
Quantifying Player Decision Making
The following is a visualization of this effect. The Y-‐axis represents the team’s points per possession on all of his paint touches. The X-‐axis represents his total paint touches as a percentage of our team’s total. As we can see by observing the trend line, the further we move to the right on the graph and increase his workload, it is initially beneficial to the results, but the further we go, the less efficient the results become. Through this analysis, we can see that D.J’s optimal workload as a percentage of the team is just under 25%. This means that our team operates most efficiently through D.J. when he is getting about 1 of every 4 paint touches we get as a team. Putting this into practice can become tricky throughout a game because the sample size needs to be relevant before making a decision. If D.J. gets 2 or 3 paint touches in a row, this shouldn’t be cause for alarm. If over the last 10 possessions, D.J. has had 5 or 6 paint touches (your team doesn’t get a paint touch every possession – more like 70 percent of all possessions), it may be time to get him a few minutes rest or get some other players involved in the offense.
There are a lot of variables that can affect this type of data. As I mentioned earlier, data like this can be something that helps to raise flags but is not enough to base decisions upon completely. Things like this need to be coupled with an eye-‐test assessment from someone who knows that team and understands their make-‐up and style. These data points can be affected by who is on the court with that player, who they are playing against, if there were changes made to the roster or to the offense, and even things like health. It took us almost an entire season of data to find a confident level of workload that is indeed optimal for this player. Analysis like this might be best used over a period longer than a year, so that when the player returns, we can feel confident from the beginning of the season the range we should be shooting for in terms of getting that player to the paint versus playing off of other teammates.
Quantifying Player Decision Making
OPTIMAL PAINT USAGE
OVERUSAGE AND UNDERUSAGE ON PAINT TOUCHES
By plotting a player’s usage percentage on paint touches with the corresponding success rate for each game of the season they played in, we can begin to draw some conclusions about their balance between usage and non-‐usage. There is usually a trend that can be seen. If a player has a number of games where their usage percentage was exceptionally high, we might see a trend that their success rates in those games were also either high or low. The same goes for games with low usages. By charting these points and calculating a polynomial line, it will show us when the player’s output is maximized. This optimal level tells us what usage percentage is most likely to render the highest success rates for that player.
Using D.J. Seeley’s chart as an example below, we see that his success rate would be maximized if his usage percentage on paint touches each game were around 62%. Anything over 70% or under 50%, he starts to become decreasingly likely to have a good success rate. In reality, D.J.’s average usage percentage was 55.38%. In his case he was 6.62% away from his optimal usage %, which is really quite good. If anything, he should have used slightly more of his paint touches for himself this season.
Quantifying Player Decision Making
With this data, we can now start to analyze which players either overused their paint touches and should have passed out more, and also those players who underused their paint touches and should have looked to finish more. Players within 5% or 10% of their optimal usage percentage, probably have a good understanding of their own ability and what they can and can’t finish when in the paint. Below is the entire team’s real usage and optimal usage, for the year, as well as how far off they were.
PLAYER SEASON PTU% OPTIMAL PTU% DIFFERENCE
MALCOLM LEE 45.33% 45% -‐0.33% NOLAN SMITH 62.75% 65% 2.25% VICTOR RUDD 64.57% 77% 12.43% JAMAL JONES 65.74% 68% 2.26% D.J. SEELEY 55.38% 62% 6.62% GIDEON GAMBLE 76.00% 79% 3.00%
MELVIN JOHNSON III 75.76% 64% -‐11.76%
RAHLIR HOLLIS-‐JEFFERSON 53.85% 74% 20.15% KENNY HALL 90.00% 65% -‐25.00% DREW GORDON 67.89% 62% -‐5.89% RONALD ROBERTS 80.41% 77% -‐3.41% JARED CUNNINGHAM 73.81% 84% 10.19% SEAN KILPATRICK 58.02% 66% 7.98% JOONAS CAVEN 72.22% 62% -‐10.22% TINY GALLON 65.91% 61% -‐4.91% LAQUINTEN MILES 65.38% 100% 34.62% MALIK WAYNS 41.24% 60% 18.76% JORDAN MCRAE 57.32% 82% 24.68% JOEL WRIGHT 75.51% 100% 24.49% NORVEL PELLE 80.00% 100% 20.00%
Some observations from this data (excluding players with small sample sizes – Miles, Wright, Pelle):
-‐ A positive number represents a player who should have been shooting that much more when getting to the paint based on their results. A negative number represents a player who was probably shooting or turning it over too much when in the paint and should have been passing it out more.
-‐ Kenny Hall, as we know, rarely ever passed the ball out once receiving it in the paint. If he had passed it out more, it would have kept defenses more honest and allowed our team higher resulting rates on his paint touches.
-‐ Melvin Johnson III and Joonas Caven should both have looked to pass it out more. Melvin lacked vision and only drove to score. Joonas actually had great vision but didn’t get to the paint enough to really have a sample size that is reliable. -‐ Jordan McRae, Rahlir Hollis-‐Jefferson, Malik Wayns and even Victor Rudd, all should have looked to finish more when
in the paint. Jordan, Rahlir, and Vic are all good finishers and were unselfish to a fault at times given their physical ability to finish. Malik could have kept teams honest by looking to draw contact or shoot floaters a little more often.