NBADivisonSimulation

Introduction

• This Shiny App aims to predict the outcome of the current season in the Atlantic division of the NBA.
• This is essentially a two horse race between Boston and Toronto with other teams realistically out of contention.
• The app works by taking the win probabilities of all remaining Boston and Toronto games and performs a Monte Carlo simulation.
• The simulation calculates how many wins both Boston and Toronto finish the season with and then returns the outcome of the division as a data frame.

How it works

• The user has the option to input the probabilites for all remaining games, however default probailites are provided.
• Once the probabilities are provided the user hits the “run” button and Shinys reactivity feature re-runs the simulation and provides new results.
• Shiny is especially usefull for this application because the win probabilites are constantly changing as injuries occur.
• This reactivity allows a novice user to constantly update the simulation without any programming knowledge.

Explaining the code

-Inputted probabilites are stored in a vector as such:

#Each element in the vector represents the probability of boston winning that particular game
boston <- c(.34,.55,.7,.4,.2,.8,.4,.3,.6,.5)
#We then use these probalities to generate random binomials for each game as follows:
bostonresults <- rbinom(length(boston),1, boston)
bostonresults

##  [1] 0 0 1 0 0 0 1 0 0 0

#Current number of wins for boston
bostoncurrent <- 42
#We sum up our predicted results and add them to the current number of wins
bostonfinal <- sum(bostonresults) + bostoncurrent
#Our final predicted number of wins:
bostonfinal

## [1] 44

The same process is repeated for Toronto. When Boston and Toronto play each other we generate a random binomial, add it to Bostons result and assign the inverse to Toronto.

Explaining the code continued

• The code explained in the previous slide is placed inside a loop and run 10000 times unless the user specifies a different number.
• After each loop the results are stored on a vector so we have the data from all our simulations.
• We then compare how often Toronto has more wins than Boston and vice versa. We also calculate how often the two teams are tied.

Thank You