Basketball players who make several baskets in succession are described as having a hot hand. Fans and players have long believed in the hot hand phenomenon, which refutes the assumption that each shot is independent of the next. However, a 1985 paper by Gilovich, Vallone, and Tversky collected evidence that contradicted this belief and showed that successive shots are independent events. This paper started a great controversy that continues to this day, as you can see by Googling hot hand basketball.
We do not expect to resolve this controversy today. However, in this lab we'll apply one approach to answering questions like this. The goals for this lab are to (1) think about the effects of independent and dependent events, (2) learn how to simulate shooting streaks in R, and (3) to compare a simulation to actual data in order to determine if the hot hand phenomenon appears to be real.
In this lab we will explore the data using the dplyr package and visualize it
using the ggplot2 package for data visualization. The data can be found in the
companion package for this course, statsr.
Let's load the packages.
In [1]:
library(statsr)
library(dplyr)
library(ggplot2)
Our investigation will focus on the performance of one player: Kobe Bryant of the Los Angeles Lakers. His performance against the Orlando Magic in the 2009 NBA finals earned him the title Most Valuable Player and many spectators commented on how he appeared to show a hot hand. Let's load some necessary files that we will need for this lab.
In [2]:
data(kobe_basket)
In [4]:
str(kobe_basket)
In [3]:
summary(kobe_basket)
Out[3]:
In [5]:
head(kobe_basket, 8)
Out[5]:
This data frame contains 133 observations and 6 variables, where every
row records a shot taken by Kobe Bryant. The shot variable in this dataset
indicates whether the shot was a hit (H) or a miss (M).
Just looking at the string of hits and misses, it can be difficult to gauge whether or not it seems like Kobe was shooting with a hot hand. One way we can approach this is by considering the belief that hot hand shooters tend to go on shooting streaks. For this lab, we define the length of a shooting streak to be the number of consecutive baskets made until a miss occurs.
For example, in Game 1 Kobe had the following sequence of hits and misses from his nine shot attempts in the first quarter:
$$ \textrm{H M | M | H H M | M | M | M} $$You can verify this by viewing the first 8 rows of the data in the data viewer.
Within the nine shot attempts, there are six streaks, which are separated by a "|" above. Their lengths are one, zero, two, zero, zero, zero (in order of occurrence).
1. Fill in the blank: A streak length of 1 means one ___ followed by one miss.
2. Fill in the blank: A streak length of 0 means one ___ which must occur after a miss that ended the preceeding streak.
Counting streak lengths manually for all 133 shots would get tedious, so we'll
use the custom function calc_streak to calculate them, and store the results
in a data frame called kobe_streak as the length variable.
In [6]:
kobe_streak <- calc_streak(kobe_basket$shot)
We can then take a look at the distribution of these streak lengths.
In [7]:
ggplot(data = kobe_streak, aes(x = length)) +
geom_histogram(binwidth = 1)
We've shown that Kobe had some long shooting streaks, but are they long enough to support the belief that he had hot hands? What can we compare them to?
To answer these questions, let's return to the idea of independence. Two processes are independent if the outcome of one process doesn't affect the outcome of the second. If each shot that a player takes is an independent process, having made or missed your first shot will not affect the probability that you will make or miss your second shot.
A shooter with a hot hand will have shots that are not independent of one another. Specifically, if the shooter makes his first shot, the hot hand model says he will have a higher probability of making his second shot.
Let's suppose for a moment that the hot hand model is valid for Kobe. During his career, the percentage of time Kobe makes a basket (i.e. his shooting percentage) is about 45%, or in probability notation,
$$ P(\textrm{shot 1 = H}) = 0.45 $$If he makes the first shot and has a hot hand (not independent shots), then the probability that he makes his second shot would go up to, let's say, 60%,
$$ P(\textrm{shot 2 = H} \, | \, \textrm{shot 1 = H}) = 0.60 $$As a result of these increased probabilites, you'd expect Kobe to have longer streaks. Compare this to the skeptical perspective where Kobe does not have a hot hand, where each shot is independent of the next. If he hit his first shot, the probability that he makes the second is still 0.45.
$$ P(\textrm{shot 2 = H} \, | \, \textrm{shot 1 = H}) = 0.45 $$In other words, making the first shot did nothing to affect the probability that he'd make his second shot. If Kobe's shots are independent, then he'd have the same probability of hitting every shot regardless of his past shots: 45%.
Now that we've phrased the situation in terms of independent shots, let's return to the question: how do we tell if Kobe's shooting streaks are long enough to indicate that he has hot hands? We can compare his streak lengths to someone without hot hands: an independent shooter.
While we don't have any data from a shooter we know to have independent shots, that sort of data is very easy to simulate in R. In a simulation, you set the ground rules of a random process and then the computer uses random numbers to generate an outcome that adheres to those rules. As a simple example, you can simulate flipping a fair coin with the following.
In [8]:
coin_outcomes <- c("heads", "tails")
sample(coin_outcomes, size = 1, replace = TRUE)
Out[8]:
The vector outcomes can be thought of as a hat with two slips of paper in it:
one slip says heads and the other says tails. The function sample draws
one slip from the hat and tells us if it was a head or a tail.
Run the second command listed above several times. Just like when flipping a coin, sometimes you'll get a heads, sometimes you'll get a tails, but in the long run, you'd expect to get roughly equal numbers of each.
If you wanted to simulate flipping a fair coin 100 times, you could either run
the function 100 times or, more simply, adjust the size argument, which
governs how many samples to draw (the replace = TRUE argument indicates we put
the slip of paper back in the hat before drawing again). Save the resulting
vector of heads and tails in a new object called sim_fair_coin.
In [9]:
sim_fair_coin <- sample(coin_outcomes, size = 100, replace = TRUE)
To view the results of this simulation, type the name of the object and then use
table to count up the number of heads and tails.
In [10]:
sim_fair_coin
table(sim_fair_coin)
Out[10]:
Out[10]:
Since there are only two elements in outcomes, the probability that we "flip"
a coin and it lands heads is 0.5. Say we're trying to simulate an unfair coin
that we know only lands heads 20% of the time. We can adjust for this by adding
an argument called prob, which provides a vector of two probability weights.
In [11]:
sim_unfair_coin <- sample(coin_outcomes, size = 100, replace = TRUE,
prob = c(0.2, 0.8))
In [12]:
table(sim_unfair_coin)
Out[12]:
prob = c(0.2, 0.8) indicates that for the two elements in the outcomes vector,
we want to select the first one, heads, with probability 0.2 and the second
one, tails with probability 0.8. Another way of thinking about this is to
think of the outcome space as a bag of 10 chips, where 2 chips are labeled
"head" and 8 chips "tail". Therefore at each draw, the probability of drawing a
chip that says "head"" is 20%, and "tail" is 80%.
In a sense, we've shrunken the size of the slip of paper that says "heads",
making it less likely to be drawn and we've increased the size of the slip of
paper saying "tails", making it more likely to be drawn. When we simulated the
fair coin, both slips of paper were the same size. This happens by default if
you don't provide a prob argument; all elements in the outcomes vector have
an equal probability of being drawn.
If you want to learn more about sample or any other function, recall that you
can always check out its help file with ?sample.
Simulating a basketball player who has independent shots uses the same mechanism that we use to simulate a coin flip. To simulate a single shot from an independent shooter with a shooting percentage of 50% we type,
In [19]:
shot_outcomes <- c("H", "M")
sim_basket <- sample(shot_outcomes, size = 133, replace = TRUE)
In [20]:
table(sim_basket)
Out[20]:
To make a valid comparison between Kobe and our simulated independent shooter, we need to align both their shooting percentage and the number of attempted shots.
Note that we've named the new vector sim_basket, the same name that we gave to
the previous vector reflecting a shooting percentage of 50%. In this situation,
R overwrites the old object with the new one, so always make sure that you don't
need the information in an old vector before reassigning its name.
With the results of the simulation saved as sim_basket, we have the data
necessary to compare Kobe to our independent shooter.
Both data sets represent the results of 133 shot attempts, each with the same shooting percentage of 45%. We know that our simulated data is from a shooter that has independent shots. That is, we know the simulated shooter does not have a hot hand.
In [21]:
sim_streak <- calc_streak(sim_basket)
In [22]:
ggplot(data = sim_streak, aes(x = length)) +
geom_histogram(binwidth = 1)
In [25]:
pnorm(1800, mean = 1500, sd = 300)
Out[25]:
In [26]:
pnorm(24, mean = 21, sd = 5)
Out[26]:
In [27]:
qnorm(0.9, mean = 1500, sd = 300)
Out[27]:
In [29]:
qnorm(0.1, mean = 21, sd = 5)
Out[29]:
In [46]:
1- pnorm(34,24,4)
Out[46]:
In [49]:
1- dbinom(0, 10, 0.07)
Out[49]:
In [54]:
sum(dbinom(50:160, 160, 0.28)) # direct add binomal
Out[54]:
In [56]:
1- pnorm(50, 44.8, 5.68)
Out[56]:
In [33]:
# mean:
245*0.25
Out[33]:
In [36]:
# SD:
sqrt(245*0.25*(1-0.25))
Out[36]:
In [31]:
1- pnorm(70, 61.25, 6.78) # Normal distribution approximation
Out[31]:
In [32]:
1- pnorm(70-0.5, 61.25, 6.78) # with 0.5 adjustment, normal approximaiton
Out[32]:
In [37]:
sum(dbinom(70:245, 245, 0.25)) # direct add binomal
Out[37]:
In [57]:
str(kobe_streak)
In [ ]: