Simulations
1 Simulations
To give an idea of the distribution of the max streak length of a simulated player shooting at 50% make rate I simulated samples of size 100 on 20 “players” below
I replicate this data with boxplots here for future reference
2 Star NBA Players vs Simulated Player
The previous graphs in the DataClean.rmd file are obviously not the best demonstration of the frequency at which certain streak lengths occur, so now I create some graphics to first demonstrate several star players frequencies, then divide down into streaks of different shot types using boxplots and violin plots (Violin plots are a combination of a Box plot and a Density Plot that is rotated and placed on each side, to display the distribution shape of the data). I test each demonstration against a simulated player who makes 50% of his shots streak lengths.
Most of the players selected here are simply because they are stars, but then I decided to filter for the players with the highest field goal percentage in the NBA who are almost entirely centers since they mostly dunk. DeAndre Jordan and Rudy Gobert have the top 2 all time leading field goal percentages in the NBA and are consequently closest to achieving the “hot hand.” Centers and LeBron (always an outlier) also appear to be one of the few types of players whose modal streak length is not 1. Seth Curry is selected for reasons that will become apparent later.
3 Different Shot Types
Here I begin to group by shot type and then simulate the three actively playing highest and lowest players in shooting percentage for each type vs a simulated player shooting at a 50% make/miss rate. I avoid selecting players with a low number of total shots where Jordan Silbert has a (very impressive) 100% shot make percentage on 1 shot by filtering for at least 100 shots made.
| Top 3 and Bottom 3 in 2 Point Shooting %1 | |||
|---|---|---|---|
| Player | 2015-2020 Shooting % | Position | |
| Mitchell Robinson | 71.99 | ![]() |
C |
| Lucas Nogueira | 69.51 | ![]() |
C |
| Gary Clark | 68.57 | ![]() |
PF |
| Sasha Vujacic | 37.24 | ![]() |
SG |
| Rashad Vaughn | 36.62 | ![]() |
SG |
| Jevon Carter | 33.89 | ![]() |
PG |
| source: basketball-reference.com via nbastatR | |||
|
1
Players with at least 100 shots in 2015-2020
|
|||
Now looking at streak length based on these players we get some very interesting results, despite being one of the worst shooters in league history, Sasha Vujacic is also very streaky - his miss streaks must also be impressive to yield these results.
I repeat the above processes for 3 point field goals here.
| Top 3 and Bottom 3 in 3 Point Shooting1 | |||
|---|---|---|---|
| Player | 2015-2020 Shooting % | Position | |
| Seth Curry | 44.28 | ![]() |
PG |
| Duncan Robinson | 43.68 | ![]() |
SF |
| JJ Redick | 43.00 | ![]() |
SG |
| Josh Okogie | 27.37 | ![]() |
SG |
| Corey Brewer | 26.92 | ![]() |
SF |
| Evan Turner | 26.67 | ![]() |
SG |
| source: basketball-reference.com via nbastatR | |||
|
1
Players with at least 100 shots in 2015-2020
|
|||
Once again graphing we get some interesting results, Seth Curry when shooting 3s reaches the same levels as the simulated player fairly consistently, and no other shooter comes close.
4 Shot Distance
Dunks are likely obfuscating the data here as they are practically guaranteed so I also exclude those. The NBA has a 22-foot 3-point line in the corners and a 23-foot, 9-inch line making things pretty weird for determining shot distance around the line. I filtered up to 22 feet, and then selected by 3 point shots.
A work by Duncan Gates
gatesdu@oregonstate.edu











