New York Knicks Analytics Assessment Overview
At a time when salary caps are skyrocketing, and teams are sinking $100s of millions on players who are not adding commensurate value, it’s more important than ever for teams to know that the players they select add value and will do whatever it takes to win and help the team achieve its goals.
The main challenge that is keeping the Knicks and most NBA teams from unlocking the full potential of their data and players is the inability to translate analytics and complex data into David Fizdale’s game plan and in a format that players can easily understand and determine on their own, ways to maximize value and contribute to wins. Data is useless if it can’t be easily analyzed, translated and acted upon by players.
What if your players were adequately rewarded for doing the things David Fizdale needs them to do to achieve team goals? Would they focus less on vanity statistics, such as scoring and 3-point shooting, and instead focus on stats that build team synergy and fulfill team needs, such as hustle stats, assists, floor spacing, close outs and ball movement? What if the Knicks had a turnkey process that integrated their analytics programs with a reward system tied back to their strategy and game plans?
In this analysis of the Knicks I share with you an analytic model for identifying your impact stats and team needs, and a proven framework for getting your players to execute the team’s strategy and adapt their games to fulfill team needs.
With this system, David Fizdale can integrate analytics into his game plans, players can demonstrate their ability to adapt their playing style and execute game plans, and owners can have confidence knowing that the players they acquire will make the greatest contribution towards achieving team goals.
Objectives of this Analysis
In this analysis for the Knicks, I deliver a performance strategy that ensures your players are adapting their playing styles to achieve team goals and achieve synergy. I also share an analytic and mathematical model for:
- identifying the Impact Stats that contribute to team winning
- building a Team Needs Profile that outlines the optimal set of stats the Knicks needed this year in order to achieve 50, 60, and 70 wins…and what it will take next season
- developing an Agility Scoring Portfolio, which provides the point values for rewarding player actions on the court.
- determining which players across the league statistically can most adapt their games to support the Knicks’ specific team needs.
In the following sections, I’ll explain the Agile Sports Analytics Framework and the approach and formulas behind the analytic models used in my analysis. This includes:
- A process for translating analytics into player results
- Machine learning models for measuring stat impact, determining team needs and recommending the statistics needed to achieve team goals, and determining which players around the league are most able to adapt their playing style to help the Knicks achieve their target metrics.
- A math approach which calculates the estimated amount of Agility Points that should be assigned for each player action that is executed on the court.
For this research, I chose to utilize publicly available data from stats.nba.com. I utilized all available game statistics from the previous four NBA seasons (2014-2015 – 2017-2018), plus statistics up to March 8th of the 2018-2019. This came out to 5,895 games in total and included all team stats and player stats across all the available stat tables (traditional, advanced, miscellaneous, scoring, usage, opponent, defense, shooting and tracking). In total, 431 statistics were investigated. In addition, these statistics were collected for the host and guest teams and for each of the players. Out of these 431 stats, I selected 17 which were used to build a predictor for match results. To reduce the stat set to the 17 selected for this analysis, I removed stats that strongly correlate with game results, such as final scores and points scored per quarter. I then built a machine learning model to classify wins and losses, taking into account host and guest team stats. I tuned this model and received an auc-roc score of 96%. Auc-roc represents the area under the ROC curve, and a score of 90-100% is considered to have excellent accuracy. For creating a model, I used gradient boosting classifier which has a parameter called feature importance. It shows how frequently a parameter has been used for building a decision tree. The more an attribute is used to make key decisions with decision trees, the higher its relative importance. Using this method, I selected the top-100 parameters and applied a logical process for reducing them, which included removing duplicate information and stats that included multiple parameters. For example, there were parameters for rebounds, offensive rebounds and defensive rebounds. Applying this logic, if we use offensive and defensive rebounds parameters, it’s redundant to use the total rebounds parameter, since its value always equals the sum of offensive and defensive rebounds. An example of statistics with multiple parameters is player efficiency rating (PER), which uses FGM, FGA, 3PM and a fixed integer to calculate.
Agile Sports Analytic Model
The ASA Model supports the processes, tools, roles and events that drive the ASA Framework. It’s comprised of two models and several supporting procedures. This section outlines the approach, decisions, analytic models and algorithms used in my analysis to build the Knicks’ Agility Points Portfolio. The model predicts team and player performance and the ability for players to adapt their playing style, fulfill team needs and contribute to team goals.
Stat Impact Analysis Model
For analyzing stats that most impact team wins and losses, I created a machine learning model. I utilized stats of the host and guest teams to predict with high precision which team will win games. I utilized stats from all 30 NBA teams over the past four seasons. I built a stat impact model using Gradient Boosting Classifier from sklearn library. For tuning hyper parameters of the model I used the GridSearchCV method, which found parameters that showed a high roc-auc score of 96%. Gradient Boost helps to analyze importance of features.
Applying this method, I discovered which features are important for prediction and removed the non-important predictors. I also removed features that contained duplication of information. In the end, I narrowed the stat set to the following 17 stats. Below are the 17 stats that most impact winning, and the Knicks’ ranking among the other 29 teams in the NBA. See table below:
Knicks’ Impact Stats/Coefficients
|STAT NAME||COEFFICIENT||Knicks Ranking||STAT NAME||COEFFICIENT||Knicks Ranking|
|FT%||0.077844744||18||PTS OFF TO||0.022482112||25|
This part of the investigation gives us a set of actions that predefines results of the game, and which we will reward or penalize players. The three actions for which we are going to penalize players are blocked shots against, turnovers and personal fouls, because they have a negative coefficient. This means that these stats negatively impact wins.
The Knicks rank highest in 2nd chance points and offensive rebounds. These stats will yield less Agility Points, as they are statistics the team naturally performs well in.
The Knicks rank lowest in field goal percentage, blocks against and assists. These stats will yield more Agility Points, as they are statistics the team requires more effort and focus in order to perform at a level that meets the team’s needs and achieve their goals.
In the Team Needs Analysis Model I answer two questions:
- “what statistics do the Knicks need to deliver to win N games during the season”, and
- “what statistics do the Knicks need to achieve to beat a specific opponent”
Needless to say, this question has many answers. But not all answers have the same practical meaning. For example, it’s harder to change team style drastically than it is to slightly increase values of stats that have positive impact, and slightly decrease values of stats that have negative impact.
So, to answer both questions I determined a number p, where increasing values of all the stats with positive impact by p% and decreasing values of stats with negative impact by p% will give desired results. The approach for these two problems was similar with the only difference in objective function. For the first case I calculated estimated amount of games won during the season using calendar 2017–2018 statistics and 2018 – 2019 statistics up to March 8th. For the second case I used estimation of the matches between the Knicks and their opponents, and the probability of Win as a parameter.
To find a proper p I used binary search algorithm on selected function. It was appropriate to apply this algorithm, since both functions are monotonic, because the team with better stats will have higher results. And the higher the p the better stats the Knicks have, the higher the value of the selected function.
The other approach to this problem is to analyze which stats each player on the Knicks can achieve to contribute most to team success. To do this, I applied the League-Wide Player Agility Assessment model (see section, League-Wide Player Agility Assessment Model).
With this model we can find optimal stats for each player, the optimal set of players for each opponent/game and their recommended minutes played. In this case team stats can be defined as a combination of stats of the chosen players. This model gives us an understanding of what is the best possible performance the Knicks can achieve without acquiring any new players. It also can determine which players and combination of players can affect team results most, and allows us to estimate player performance/cost values.
The table on the following page displays the set of impact statistics identified in the previous section, and the statistical performance needed for the Knicks to win 50, 60 and 70 games this season. As you can see, very small, achievable improvements in the key impact areas can yields significant results.
Knicks’ Stats Needed to Win 50, 60 and 70 Games
|PTS OFF TOV||16.31||16.62||16.72|
As you can see… small, achievable improvements across high impact stats can yield significant results.
Agility Points Portfolio Model
For building the Agility Points Portfolio and the predictor model for the Knicks I analyzed each stat separately. I calculated “cap” value for a given stat. “Cap” value in this case refers to the minimal value of the stats, which gives the best results during the season for the Knicks. There was one exception, that the value of the analyzed stat be equal to selected value. “Cap” value was found using binary search method.
After I found the “cap” value I calculated the score of the team with average stats and “cap” value in selected stat, and the score of the team with average stats and 0 in selected stat. Then I assigned Agility Points (AP) for the stat as a difference between the two values divided by the “cap” value of the stat.
Recommended Agility Points Algorithm
In this section I offer a method for applying mathematics to determine the Agility Scoring System. The idea of this approach is to calculate how many points each action will gain for the Knicks players using average team stats and average opponent stats.
However, it should be noted that there are several approaches you can use to develop a scoring system. A combination of mathematics and coach’s expert judgment yields the best results.
Each shooting successive action will give as many points as its point cost:
- 3PM = 3 Agility Points
- 2FGM = 2 Agility Points
- FTM = 1 Agility Point
In addition, a point value based on a set of rules was determined for the stats that directly correlate with shooting stats:
- 2nd PTS = .25 Agility Points
- PTS OFF TO = .25 Agility Points
- FBPs = .25 Agility Points
- PITP = .25 Agility Points
Note: This model is adjustable. Any existing analytic model which the Knicks uses can be integrated into the Agile Sports Framework.
For all non-shooting stats I used a coefficient, where scoring a field goal cost more than the non-scoring stats. Each coefficient depends on how many actions the team collects during one game, on average, in comparison with the other teams. If the number of actions the Knicks collect compared to the rest of the league is:
- in the 80%-100%, then the coefficient of 0.7 was applied
- in the 60%-80%, then the coefficient of 0.75 was applied
- in the 40%-60%, then the coefficient of 0.8 was applied
- in the 20%-40%, then the coefficient of 0.85 was applied
- in the 0% to 20%, then the coefficient of 0.9 was applied
For each action, I calculated the point contribution of the action.
Offensive rebounds gives a team the ability to earn 2nd points, so its costs equals to average 2nd points per offensive rebound:
OREB = OREB_COEFF * team_2nd_PTS’] / team_OREB
Defensive rebounds prevent opposing teams from scoring 2nd points, so it’s calculated in a similar manner, except all the stats used will be from the stats of the opponent
DREB = DREB_COEFF * opp_team_2nd_PTS / opp_team_OREB
Cost of assists are calculated as average points created with assists divided by the average amount of assists per game:
AST = AST_COEFF * team_AST_PTS_Created / team_AST
To calculate cost of personal fouls and personal fouls drawn I needed to count how many free-throws made happen per 1 foul of the team, so
PFD = PFD_COEFF * team_FTM / team_PFD
PF = -PF_COEFF * opp_team_FTM / opp_team_PFD
To calculate Agility Points for turnovers I needed to understand how many points each turnover gained for the opposing team. For doing this I took the average amount of points scored for turnovers by the opposing team and divided it by the average amount of turnovers opponent makes vs opponent team
TOV = -TOV_COEFF * opp_team_PTS_OF_TO / TOV_vs_opp_team
Steals prevent teams from successfully completing a play. Average points scored during one attack of the opponent team can be calculated using the average amount of points per one possession:
STL = STL_COEFF * opp_team_PTS / opp_team_poss
To calculate the contribution of blocked shots we take into consideration two things:
- Blocked shots prevent the opponent from scoring a goal. The average amount of field goals scored per one possession.
- After blocked shots one of the teams gets the rebound. If the shooting team rebounds, there’s an opportunity to score 2nd chance points. In this case I increased APs for 2nd pts per offensive rebound multiplied by the probability to get offensive rebound. I calculated this probability as average offensive rebounds made by the team divided by sum of the defensive team rebounds and offensive rebounds of the offensive team:
BLK = BLK_COEFF * [(opp_team_PTS – opp_team_FTM) / opp_team_poss – (opp_team_2nd_PTS / opp_team_OREB) * opp_team_OREB / (team_DREB + opp_team_OREB)]
We can simplify this using the following formula:
BLK = BLK_COEFF * [(opp_team_PTS – opp_team_FTM) / opp_team_poss – opp_team_2nd_PTS / (team_DREB + opp_team_OREB)]
In the same manner we can calculate Agility Points for BLKA:
BLKA = – BLKA_COEFF * [(team_PTS – team_FT) / team_poss – team_2nd_PTS / (opp_team_DREB + team_OREB)]
One adjustment that I considered in this model is with personal fouls (PF), which is best applied as an average, instead of through this model.
This set of coefficients allows us to quantify contribution of each player towards team success in numbers. This set of coefficients also shows which stats players should focus on to gain the highest Agility Points score.
The table below provides an Agility Points Scoring model for the Knicks based on this approach.
Knicks Team Agility Scoring Model
|2FGM||2nd PTS||3PM||AST||BLK||BLKA||DREB||FBPs||FTM||OREB||PF||PFD||PITP||PTS OFF TO||STL||TOV|
This table outlines the points rewarded (or deducted) based on player actions during the game.
Knicks League-Wide Player Agility Assessment Model
For the Player Agility Assessment Model, I attempt to answer what is the best possible Agility Point score a player can likely achieve, based on their past stats. I was not interested in all possible player performances, but only on those that could be shown with some fixed probability (for example, with probability of 80%).
To calculate this, I used player historical performance from the 2016-17 and 2017-18 and 65 games into the 2018-19 season. I normalized stats of the players on a per-second basis. I also removed performances when players spent less than 25 minutes on the floor, to avoid results that include players who play mostly when the game results are already determined, such as with blow outs or final minutes.
For the Player Agility Assessment Model, I took into consideration the fact that a player can change their stats by intentionally changing their playing style (provided they are rewarded for their adjusted efforts). On the other side, as it pertains to this analysis, I wanted to have evidence that the player is capable of achieving given or similar stats, based on their past performance. I used the hypothesis that the higher a player showed some value of the stat the higher the probability that the player can demonstrate those performances again. The hypothesis that if a player has shown stats X = (x1, x2… x17) in one game and stats Y = (y1, y2… y17) in another game, then all the stats alpha * X + beta * Y, where alpha + beta <= 1 and alpha >= 0, beta >= 0 also can be achieved.
For given set of stats e =(s1, s2, …, sk) the probability to achieve such set of stats was calculated as a fraction X / (n * (n – 1) / 2), where n is the total number of historical player performances and X is a number of pairs of performances e1 = (s11, s12, …, s1k) and e2 = (s21, s22, …, s2k) such that there’s a linear combination alpha * e1 + beta * e2 exist which dominates vector e in each of its components, such that alpha >= 0, beta >= 0 and alpha + beta <= 1.
By applying these optimal stats into the Agility Scoring System we can determine which players on the team make the greatest contribution to the Knicks’ success, can contribute most to the teams’ needs and whether their predicted contribution is worth their salary.
The table below provides the recommended Agility Points Scoring Model for the Knicks players based on this approach.
Knicks’ Player Agility Points Scoring Model
|Player||2 FGM||2nd PTS||3PM||AST||BLK||BLKA||DREB||FBPs||FTM||OREB||PF||PFD||PITP||PTS OFF TO||STL||TOV||AP SCORE|
|Dennis Smith Jr.||12.65||0.51||6.40||18.67||0.85||-0.87||4.65||1.00||3.22||0.87||-2.86||2.68||2.63||1.06||3.82||-8.22||47.07|
We can take is a step further. By applying this logic for every player in the league, and applying their optimal stats to the Knicks’ Agility Scoring System, I can determine which players across the NBA are most likely able to adapt their games to help the Knicks achieve their goals. This list focuses only on players who averaged over 24 minutes per game. When cross-referenced with player salaries and including players will a little less average playing time, we can determine the best trades for the team based on your budget.
The table below identifies the top 50 players around the league, who based on past performance, can make the greatest contribution to helping the Knicks win next year.
Top Prospects Based on the Knicks’ Agility Scoring System
|Player||Team||2 FGM||2nd PTS||3PM||AST||BLK||BLKA||D REB||FBPs||FTM||O REB||PF||PFD||PITP||PTS OFF TO||STL||TOV||AP SCORE|
|Giannis Antetokounmpo||Milwaukee Bucks||27.67||1.14||2.57||19.66||2.90||-0.98||17.36||1.59||9.47||3.13||-3.11||7.70||6.39||1.47||3.32||-9.10||91.17|
|Anthony Davis||New Orleans Pelicans||25.32||1.64||3.33||13.10||4.51||-0.52||14.95||1.10||9.26||4.20||-2.27||6.79||5.35||1.63||3.66||-4.64||87.40|
|Russell Westbrook||Oklahoma City Thunder||18.94||0.86||5.23||33.02||0.77||-0.60||15.41||1.80||5.76||1.92||-3.03||4.43||3.72||1.69||4.74||-10.74||83.91|
|James Harden||Houston Rockets||14.78||0.96||18.76||22.22||1.31||-0.94||8.53||1.04||12.49||1.22||-2.65||6.72||3.38||1.73||4.57||-11.48||82.64|
|LeBron James||Los Angeles Lakers||21.34||0.86||8.08||23.80||1.13||-0.68||12.20||1.75||6.82||1.37||-1.42||5.22||4.46||1.29||3.21||-8.19||81.22|
|Nikola Jokic||Denver Nuggets||19.69||1.49||5.06||26.40||1.33||-0.71||13.94||0.26||6.19||4.26||-2.93||5.95||4.36||0.77||3.46||-8.45||81.08|
|Joel Embiid||Philadelphia 76ers||21.97||1.55||5.00||11.16||3.72||-0.89||18.32||0.63||11.72||3.45||-3.08||7.66||4.63||1.27||1.40||-8.55||79.97|
|Nikola Vucevic||Orlando Magic||23.55||1.28||4.97||13.82||2.61||-0.57||16.73||0.22||3.04||4.23||-1.91||2.83||4.57||0.86||2.42||-5.22||73.41|
|Karl-Anthony Towns||Minnesota Timberwolves||19.84||1.72||8.14||10.79||3.18||-0.66||15.14||0.26||6.91||4.77||-3.66||5.57||4.29||1.42||2.04||-7.09||72.65|
|DeMarcus Cousins||Golden State Warriors||15.96||0.76||5.22||15.36||3.58||-1.01||15.34||0.91||8.33||2.48||-4.86||8.48||3.89||1.27||4.04||-7.60||72.15|
|Kevin Durant||Golden State Warriors||20.96||0.72||7.88||17.70||2.11||-0.35||9.99||1.66||8.29||0.67||-1.68||4.95||2.96||1.30||1.72||-6.92||71.96|
|Paul George||Oklahoma City Thunder||14.51||0.92||15.28||12.57||0.79||-0.61||10.33||1.55||7.70||1.79||-2.40||4.75||2.43||1.73||4.90||-5.74||70.48|
|Kawhi Leonard||Toronto Raptors||20.36||0.81||7.20||10.42||0.71||-0.52||10.17||1.38||9.19||1.87||-1.28||5.94||3.28||1.59||4.11||-4.80||70.44|
|Kyrie Irving||Boston Celtics||18.53||0.67||11.26||22.18||0.89||-0.50||6.29||1.27||4.50||1.62||-2.21||3.81||3.20||1.27||3.81||-6.47||70.13|
|Stephen Curry||Golden State Warriors||12.20||0.86||21.61||16.66||0.68||-0.39||7.54||1.80||5.82||0.94||-2.19||3.91||2.13||1.40||2.84||-6.62||69.18|
|Ben Simmons||Philadelphia 76ers||18.92||0.77||0.00||25.12||1.42||-0.54||11.51||1.17||4.65||3.04||-2.20||4.82||4.69||0.87||3.03||-8.55||68.74|
|Lou Williams||LA Clippers||17.97||0.40||7.55||22.48||0.34||-0.78||5.08||1.34||10.14||0.84||-1.37||5.86||3.14||0.95||1.91||-7.48||68.37|
|Damian Lillard||Portland Trail Blazers||15.62||0.55||11.61||19.64||0.90||-0.77||6.03||0.98||8.13||1.13||-1.56||4.77||2.72||1.20||2.60||-6.32||67.21|
|Kevin Love||Cleveland Cavaliers||10.32||1.28||14.59||7.73||0.76||-1.09||20.55||0.37||10.40||2.68||-3.03||5.64||2.31||1.59||0.60||-7.94||66.76|
|Blake Griffin||Detroit Pistons||16.26||0.68||10.10||16.41||0.73||-0.68||10.31||0.43||7.56||1.76||-2.40||7.40||3.78||1.10||1.51||-8.35||66.61|
|Jusuf Nurkic||Portland Trail Blazers||20.69||1.68||0.29||11.99||3.24||-0.84||13.94||0.25||6.51||5.92||-4.04||5.14||4.64||0.90||2.87||-6.95||66.23|
|Luka Doncic||Dallas Mavericks||13.03||0.77||10.83||18.99||0.52||-0.51||10.77||1.09||6.79||1.65||-1.79||5.68||2.73||1.07||2.46||-8.60||65.49|
|Julius Randle||New Orleans Pelicans||21.10||1.21||4.07||10.69||1.34||-0.99||12.96||0.84||8.02||3.39||-3.75||5.65||5.06||1.10||1.83||-7.47||65.04|
|Rudy Gobert||Utah Jazz||17.49||1.62||0.00||7.26||4.39||-0.51||16.09||0.19||6.34||5.54||-2.81||6.08||4.37||0.87||2.09||-4.02||64.97|
|John Wall||Washington Wizards||16.57||0.36||6.53||27.26||1.75||-0.78||5.08||1.12||5.25||0.75||-2.12||4.01||3.21||1.20||3.47||-8.97||64.67|
|Domantas Sabonis||Indiana Pacers||21.43||1.54||0.92||11.83||1.26||-0.37||15.02||0.32||5.86||4.78||-4.35||5.61||4.77||1.00||2.22||-7.22||64.61|
|Hassan Whiteside||Miami Heat||20.74||1.91||0.30||3.99||4.99||-0.64||18.21||0.25||3.05||7.02||-3.57||4.71||4.92||0.86||2.30||-4.75||64.28|
|Kemba Walker||Charlotte Hornets||14.81||0.66||13.59||17.81||0.81||-1.12||6.03||0.92||6.07||0.74||-1.41||4.77||2.63||1.11||2.85||-6.04||64.23|
|D’Angelo Russell||Brooklyn Nets||16.14||0.50||12.98||23.92||0.59||-0.42||5.89||0.81||3.21||0.91||-1.84||2.32||2.68||1.23||2.92||-8.14||63.69|
|Jrue Holiday||New Orleans Pelicans||16.82||0.47||6.61||23.56||1.42||-0.62||6.09||0.86||4.11||1.52||-1.90||3.32||3.65||0.91||3.59||-6.83||63.57|
|De’Aaron Fox||Sacramento Kings||15.86||0.18||4.90||24.54||1.06||-0.79||5.51||1.69||5.64||0.79||-2.55||5.03||3.21||1.62||4.28||-7.51||63.46|
|Andre Drummond||Detroit Pistons||21.02||1.98||0.23||3.70||3.41||-0.95||16.81||0.43||3.99||7.10||-3.31||4.06||5.18||1.07||4.06||-5.52||63.25|
|DeMar DeRozan||San Antonio Spurs||22.27||0.43||0.46||18.71||0.90||-0.94||8.67||0.45||6.57||0.99||-2.01||4.87||3.59||0.83||2.27||-6.05||62.00|
|Mike Conley||Memphis Grizzlies||13.36||0.45||9.02||20.64||0.66||-0.43||4.70||0.62||6.44||0.86||-1.61||5.03||2.38||1.28||3.33||-4.82||61.90|
|Montrezl Harrell||LA Clippers||23.66||1.57||0.09||7.39||3.46||-0.87||9.26||0.82||5.69||3.99||-3.90||5.71||5.76||0.85||2.59||-4.29||61.77|
|Eric Bledsoe||Milwaukee Bucks||14.85||0.52||7.46||20.40||0.81||-0.49||7.19||1.44||3.50||1.63||-2.22||3.16||3.49||1.46||3.90||-5.92||61.17|
|Devin Booker||Phoenix Suns||17.89||0.54||8.87||20.78||0.37||-0.86||5.47||1.25||7.17||0.71||-2.99||5.18||3.13||1.06||1.79||-9.30||61.07|
|Bradley Beal||Washington Wizards||17.80||0.51||9.89||15.19||1.23||-0.52||5.76||1.23||5.66||1.24||-2.62||3.97||3.21||1.28||3.07||-6.01||60.88|
|Enes Kanter||Portland Trail Blazers||21.30||2.03||0.74||8.08||0.81||-0.93||13.74||0.29||4.53||7.05||-3.03||4.24||5.10||0.87||1.45||-5.76||60.52|
|LaMarcus Aldridge||San Antonio Spurs||23.99||1.18||0.40||8.23||2.43||-0.46||9.88||0.32||6.23||4.30||-2.20||5.14||3.82||0.75||1.24||-4.78||60.49|
|Trae Young||Atlanta Hawks||12.93||0.50||8.38||26.89||0.46||-0.96||4.87||1.09||6.05||1.01||-1.70||5.64||2.84||0.88||2.17||-10.55||60.47|
|Chris Paul||Houston Rockets||9.00||0.61||9.40||26.92||0.40||-0.17||6.66||0.38||4.99||0.71||-2.53||3.77||1.52||0.93||4.73||-7.00||60.33|
|Victor Oladipo||Indiana Pacers||14.73||0.36||9.01||17.30||0.61||-1.02||8.97||1.54||4.10||0.90||-1.95||3.77||2.16||1.22||4.15||-5.74||60.11|
|John Collins||Atlanta Hawks||21.20||1.58||4.50||7.20||0.87||-0.93||11.23||0.76||5.74||5.43||-3.42||4.50||5.07||1.07||0.80||-5.76||59.83|
|Kyle Lowry||Toronto Raptors||6.52||0.42||9.85||28.92||0.98||-0.35||6.18||0.91||3.69||0.89||-2.38||3.98||1.21||0.73||3.20||-6.39||58.38|
|Derrick Rose||Minnesota Timberwolves||20.29||0.52||5.47||17.58||0.59||-0.80||4.42||0.94||4.92||1.09||-1.18||3.01||3.54||0.97||1.69||-4.78||58.29|
|Jimmy Butler||Philadelphia 76ers||15.66||0.60||4.41||13.15||1.16||-0.48||6.06||1.07||6.30||2.38||-1.58||3.91||3.07||0.99||4.43||-3.38||57.74|
|Zach LaVine||Chicago Bulls||18.18||0.76||7.97||13.94||0.84||-1.02||6.62||1.12||6.85||0.82||-2.12||4.02||3.69||1.32||2.01||-7.73||57.27|
|Deandre Ayton||Phoenix Suns||22.10||1.20||0.00||6.80||1.82||-0.47||13.34||0.31||3.12||4.63||-2.92||3.15||4.54||0.83||2.06||-4.58||55.94|
|Marc Gasol||Toronto Raptors||12.63||0.46||5.62||15.84||2.26||-0.33||12.58||0.29||3.75||1.63||-2.73||3.98||2.20||0.71||2.54||-5.64||55.77|
Of course, I understand that optimal stats for each player does not necessarily translate to the players who will adapt their games, but it does give us an idea of who is capable of adjusting their games based on past performance. The real gems are the non stars and mid tier players (not listed above) who can make a big impact for the best bang for the buck.
How the New York Knicks Can Get Their Players to Execute The Game Plan
Agile Sports Analytics Framework
The Agile Sports Analytics (ASA) Framework provides specific processes, roles, events, and tools necessary to translate data and analytics into player results on the court. Performance goals are planned and executed in 10 game increments, called Sprints. Players earn Agility Points, which measure their ability to adapt their playing style to execute team goals. Agility Points are like fantasy points, except the point values are based on the team’s goals, specific needs and game-plan. Each game players earn Agility Points based on their performance against the team’s Agility Scoring System for that sprint. The Agility Scoring System details how many Agility Points each stat generates when the player executes actions on the court. Each Sprint the statistics and Agility Scoring System are reassessed and adjusted, based on the past Sprint performance and the coach’s expert judgment and game plan.
An example of my approach is depicted below, and illustrates the pillars, roles, tools, events and processes of the ASA Framework. To fully understand the processes and how to integrate the Hawks’ existing analytics platform, I urge you to read my book ‘Agile Sports: A Blueprint for Measuring Value, Improving Player IQ and Creating Team Synergy’ , which details the step by step process for executing the framework. Feel free to contact me to discuss the process in detail or for me to provide a complementary sports analytics maturity assessment, measuring 26 best practices and 130 key processes within the Hawks’ organization.
The Agile Sports Framework can be applied to any team’s existing analytics platform to plan and execute team strategy.
The key members of the ASA Framework are the lead analyst, coaches, players, scouts and the agile sports consultant. Each role has specific characteristics and responsibilities critical for building team synergy and ensuring continuous performance improvement.
The lead analyst clearly defines Sprint performance metrics and their associated Agility Scores, prioritizes team and player statistics, manages the Sprint Performance Dashboard, recommends performance goals through Player Agility Cards, and ensures that players understand the Agility Points Portfolio and scoring system, and what they must do in order to add value and achieve team Sprint Goals.
The head coach is the team decision maker and defines the team’s game plan. The coach works closely with the lead analyst to ensure the game plan and team goals are captured in the analytic model. The coach also ensures that the scoring system is consistent with his observations.
The assistant coaches help the players develop their skills, IQ and ability to leverage analytics to add value to reach team goals. Assistant coaches ensure that the feedback captured in the Sprint Planning and Sprint Review meetings are relayed to the players.
The scouts apply analytics to identify the best available talent to contribute to team needs and goals.
The sports analytics consultant assumes the role of advisor to the lead analyst, coaches and players. They are responsible for providing techniques for effective Sprint Performance Dashboard management and Agility Points Portfolio statistics, ensuring the ASA Framework principles and processes are being applied, removing obstacles that impede on player or team progress and recommending changes that enforce the six pillars of the ASA Framework.
Prescribed events are used to create consistency and maintain continuous performance improvements throughout the season. All events are time-based to avoid distraction from day to day operations and activities.
The core of the ASA Framework is a Sprint, an incremental scheduled group of 10 games during which a set of specific target metrics and goals are defined, translated into a scoring system and measured.
Sprints contain and consist of the Sprint Planning Session, the 10 scheduled games, Post-Game Reviews, and the Post-Sprint Review/Retrospective meeting. Each of these events serve a purpose; to answer the following:
- What areas can we improve?
- What can the team do better to achieve its Sprint and seasonal goals?
- And how will players deliver the desired result?
After each Sprint, evaluations and necessary adjustments are made, new impact metrics are pulled from the Performance Metrics Log into the Agility Point Portfolio, and the analytic models are rerun learning from historical data and previous Sprint results.
Sprints are used to get players accustomed to hitting target goals and improving upon an area of weakness or other attribute deemed critical to achieving the team’s overall goal. Each Sprint has a definition of the goal to be accomplished, a plan that will be followed, how success will be defined, and the results.
Agile Sports Analytics Tools
Tools for the ASA Framework provide transparency of data and regular opportunities for inspection, adaptation, self-organization and synergy. They include the Sprint Performance Dashboard, Performance Metrics Log, Team Needs Profile, the Agility Points Portfolio and Player Agility Cards.
Sprint Performance Dashboard
The Sprint Performance Dashboard provides views and access into all the tools within the ASA Framework. It offers analysts, coaches, players and scouts views into team performance, strategy and Player Agility Cards.
Performance Metrics Log
The Performance Metrics Log is the repository where all team statistics are stored, before being analyzed, converted to metrics and assigned value and agility points. Stats in the Performance Metrics Log predicted to fulfill a team need and/or have the greatest impact on winning games in the upcoming Sprint are pulled into the Agility Points Portfolio.
Team Needs Profile
The Team Needs Profile provides a view into the optimal performance the players and team should deliver in order to win X number of games. It contains the Impact Stats Analysis and the Team Needs Analysis.
Agility Points Portfolio
The Agility Points Portfolio contains the stats that will be monitored during the Sprint. The Agility Points Portfolio contains the Agility Scoring System and predicted stats the team and individual players need to deliver to win each game within the Sprint. It contains the Player Agility Cards and is where analysts submit Recommended Agility Scores, the coaches align the scoring system to the game plan, and where players plan what they will do in order to achieve the recommended scores.
I applied machine learning algorithms and mathematic formulas to determine the Hawks’ Agility Scoring System (see Agility Points Scoring Model).
Recommended Player Agility Scores are the recommended stats, translated through the Agility Scoring System and defined by the analyst, which players and coaches use as a guide when determining how they will execute the strategy and achieve N number of wins.
Player Agility Cards
Player Agility Cards are the interactive component within the ASA Framework where players can view the statistics they will be measured against (defined by the Lead Analyst) and understand how their actions on the court will translate into Agility Points, and more importantly, wins. Players can adjust their planned statistical output and learn how adjusting their efforts will translate to Agility Points, and help the team achieve its Sprint Goals. By doing this, players are actually planning how they will execute Lloyd Pierce game plan, and doing so in a format they are accustomed to quantifying stat value (fantasy sports).
What’s unique about Agile Sports Analytics is that it gets players to execute team game plans by rewarding points for executing stats that aren’t typically valued in the box scores. It answers for players “what’s in it for me?”. Player Agility Cards allow analysts to recommend player goals, and players to see how each stat impacts their agility scores. This creates player buy-in to the team’s game-plan. Numerous studies show that what gets measured gets done. What good are analytics if they don’t drive player performance? The ASA Framework engages players, enables players to demonstrate they can adapt/create value, helps teams reach their goals, encourages competition within the team to generate the most agility points and guides teams to play with synergy. Most importantly, it integrates with your existing analytics program and is flexible with any statistics and metrics the team uses to measure performance. When implemented into the the culture Agile Sports Analytics can foster a system of G League and European talent ready to make an immediate impact.
Contact Victor Holman
If you’d like to discuss how Victor can help you leverage your analytics to execute game plan and organizational strategy, contact him below (in the footer)
New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics, New York Knicks Analytics,