I Built a Model to Investigate the NBA's Injury Epidemic — Transcript

Full transcript

1. 0:00And, you know, we just need bodies.
2. 0:05The imaging from yesterday showed an issue
3. 0:07that we want to get addressed quickly.
4. 0:09Our only concern is his long-term health.
5. 0:12You know, Dub is, as everybody knows, I
6. 0:14mean, he's as tough as they come. He's
7. 0:17an elite competitor.
8. 0:18In the space of a week, Anthony Edwards
9. 0:21scored 47 points on a sore right knee
10. 0:23and then needed surgery on it.
11. 0:27And Tyrese Haliburton, in his first
12. 0:29month back from a torn Achilles, went
13. 0:31down with a hamstring strain.
14. 0:33J-Dub broke his leg, Paolo Banchero
15. 0:35sprained his ankle, and we had former
16. 0:38NBA player Eddie Johnson call it an
17. 0:41epidemic.
18. 0:42And last season really was brutal.
19. 0:45The players who had made an All-Star
20. 0:47or All-NBA team the year before missed
21. 0:4939% of their team's games.
22. 0:51So I built a model to find out if NBA
23. 0:54injuries are really getting worse,
24. 0:57and to do it, I went back through every
25. 0:59box score since 1986.
26. 1:02The short answer here is yes, but it's
27. 1:05not the injuries you're thinking of.
28. 1:10The obvious way to answer this is to
29. 1:12count injuries. But that doesn't work,
30. 1:14because the way the NBA reports injuries
31. 1:16keeps changing.
32. 1:18Teams used to stash players on an injured
33. 1:20list. Then came official injury reports.
34. 1:23Then came rules on resting stars and
35. 1:25a 65-game minimum for awards.
36. 1:27If I counted the reports, I'd be partly
37. 1:29counting paperwork here.
38. 1:31So I use the one thing that hasn't
39. 1:33changed since the 1980s, which is the
40. 1:35box score. Either you played in the
41. 1:37game or you didn't.
42. 1:39I pulled almost a million box score
43. 1:41lines from 1986 to this week.
44. 1:44For every rotation player, meaning
45. 1:46anyone who plays about 20 minutes a
46. 1:48night, I checked every game his team
47. 1:50played. And if he wasn't in the box
48. 1:52score, that's a game missed.
49. 1:55There's one catch with this method
50. 1:56though. A box score can't tell me why
51. 1:58a player sat.
52. 1:58It could be a torn ligament, a sore
53. 2:01ankle, a rest day, or a suspension.
54. 2:04So I'll have to pull those apart as
55. 2:06I go here.
56. 2:07So again, version 1 is the simplest
57. 2:09count possible, which is the total
58. 2:10number of games rotation players missed
59. 2:12each season.
60. 2:14From 1990 to 2010, rotation players
61. 2:16missed about 3,363 games a season.
62. 2:20Last season, they missed 6,704.
63. 2:21The most in my whole dataset.
64. 2:25In other words, the numbers basically
65. 2:27doubled.
66. 2:28But that isn't a completely fair comparison.
67. 2:31The league has added three teams since
68. 2:321990, so some seasons were cut short.
69. 2:35And modern coaches spread minutes across
70. 2:37more players. So there are simply more
71. 2:39rotation players to get hurt.
72. 2:41So in version 2 of our model here,
73. 2:43I turned the count into a rate.
74. 2:46For each rotation player, I divided
75. 2:47the games he missed by the games his
76. 2:49team played while he was on the roster.
77. 2:52From 1990 to 2010, rotation players
78. 2:54missed 16.8% of their teams games.
79. 2:56Last season, they missed 27%.
80. 2:58The highest of all 37 seasons.
81. 3:00Put simply, a typical rotation player
82. 3:04used to miss about 1 game in 6.
83. 3:07And now it's missing more than 1 in
84. 3:094.
85. 3:10If I only count absences of two games
86. 3:12or more, last season drops to second,
87. 3:15behind 2021-22, when COVID protocols
88. 3:18kept players out for weeks at a time.
89. 3:20But COVID isn't what's behind the rise.
90. 3:23The rate jumped in 2017-2018 before
91. 3:25the pandemic, and each of the last
92. 3:28four seasons is higher than any season
93. 3:30before 2018.
94. 3:32Then I split every absence by how long
95. 3:34it lasted. And this is where it got
96. 3:36interesting.
97. 3:37The rate went up by 8 points.
98. 3:40Single-game absences, which is where
99. 3:41rest days live, account for about 2
100. 3:43of them. Absences of 20 games or more
101. 3:46account for just 1.
102. 3:48The other 5 points come from absences
103. 3:51of 2 to 19 games.
104. 3:53Basically, the Rise isn't mostly about
105. 3:55rest days, and it isn't mostly about
106. 3:58season-ending injuries. It's the kind
107. 4:00of absence that lasts 1-3 weeks.
108. 4:03But fans don't notice a backup forward
109. 4:05missing two weeks.
110. 4:06They notice stars.
111. 4:08So version 3 looks at who is missing
112. 4:10the games.
113. 4:12By stars, I mean anyone who made an
114. 4:14All-Star or All-NBA team the season
115. 4:16before.
116. 4:17From 1990 to 2010, All-Star missed
117. 4:20about 11 games a season.
118. 4:21Over the last four seasons, it's been
119. 4:24about 22. And last season, it was almost
120. 4:2732, which adds up to 852 games.
121. 4:31Part of that record was carryover from
122. 4:32the 2025 playoffs, when Tatum, Lillard,
123. 4:35and Haliburton all tore their Achilles.
124. 4:38If I only count absences that started
125. 4:40after opening night, last season ranks
126. 4:43fourth. But it's still about double
127. 4:45the old rate.
128. 4:46And here's the part that surprised
129. 4:48me the most.
130. 4:49From 1990 to 2010, Stars actually missed
131. 4:52fewer games than ordinary rotation
132. 4:55players. 13.8% against 17.1%.
133. 4:58Now it's flipped.
134. 4:59Over the last four seasons, Stars missed
135. 5:0226.9% of their teams games, and everyone
136. 5:04else missed 24.7%.
137. 5:05In plain English, the players fans
138. 5:08pay to see used to be the most available
139. 5:11players in the league, and now they're
140. 5:13the least available.
141. 5:14That's a big reason why this feels
142. 5:16like an epidemic.
143. 5:17But why is it happening?
144. 5:19The explanations I hear most are pace,
145. 5:21age, the schedule, and minutes.
146. 5:24So I tested each one.
147. 5:29Pace first. Teams do play much faster
148. 5:32than in the 2000s, but the early 1990s
149. 5:35were nearly as fast as today, and those
150. 5:38were the healthiest seasons in my data.
151. 5:40Age doesn't work either.
152. 5:42Rotation players are actually younger
153. 5:43now than they were from 1990 to 2010.
154. 5:46And when I adjust for age, the rise
155. 5:48barely moves.
156. 5:50The schedule has gotten easier, not
157. 5:51harder. Teams played about 21 back-to-backs
158. 5:54a season from 1990 to 2010, and now
159. 5:57they play about 14.
160. 5:59And minutes went down.
161. 6:00Rotation players averaged 27 minutes
162. 6:02a game now, compared with 28.5 back
163. 6:04then. Put another way, players are
164. 6:07playing less and resting more, and
165. 6:09they're still missing more games.
166. 6:11A 2024 report the NBA commissioned
167. 6:13found that resting players didn't lower
168. 6:15their risk of getting hurt later in
169. 6:17the season, so rest didn't buy health
170. 6:19according to the NBA.
171. 6:21What does fit is the shape of the absences.
172. 6:24Players now have almost twice as many
173. 6:26absences per game played.
174. 6:28But each one is a little shorter.
175. 6:30That points to teams managing bodies
176. 6:32differently, although box scores can't
177. 6:34prove it.
178. 6:35But short absences don't tell me about
179. 6:37the injuries that end seasons.
180. 6:40So version 4 looks at only the severe
181. 6:42ones.
182. 6:43Absences of 20 games or more did rise,
183. 6:46but only by about 25%.
184. 6:48A much smaller jump than everything
185. 6:50else. By games lost to those long absences,
186. 6:53last season ranks just 5th, behind
187. 6:55seasons like 1996, 1997.
188. 6:59Then there's the Achilles.
189. 7:01In 2024-25, seven players tore an Achilles
190. 7:04during the season or the playoffs.
191. 7:06From 1990 to 2024, the league averaged
192. 7:09fewer than one a season.
193. 7:16At that old rate, a seven-tear season
194. 7:18should happen about once in 10,000
195. 7:20seasons. So it wasn't just bad luck.
196. 7:22It was a real data cluster here.
197. 7:26But last season, there was only one.
198. 7:28And if I take out that one record year,
199. 7:31the long-term rise isn't statistically
200. 7:33significant. In other words, it was
201. 7:36a terrible year, but it's not yet a
202. 7:37proven trend.
203. 7:39This season already though had three
204. 7:41Achilles tears before opening night,
205. 7:43so it's definitely worth paying attention
206. 7:45to. ACL tears though have stayed flat
207. 7:48at about two a season.
208. 7:50So, short absences are up a lot, and
209. 7:53severe ones are up a little.
210. 7:56Version 5 asks whether any of that
211. 7:58is bigger than random noise.
212. 8:00To find out, I simulated 10,000 NBA
213. 8:03seasons where nothing changed after
214. 8:052010, building each one from real player
215. 8:08seasons from 1990 to 2010 drawn at
216. 8:11random.
217. 8:12Those fake seasons landed between about
218. 8:1514.5% and 19.5% of games missed, and
219. 8:17not one of the 10,000 reached even
220. 8:19the lowest of the last four real seasons,
221. 8:22which was 22.8%.
222. 8:23Put simply, if injuries had stayed
223. 8:26at their old level, seasons like these
224. 8:29would basically never happen.
225. 8:31Then I ran a backtest.
226. 8:33A backtest is like giving the model
227. 8:34old exams with the answers covered.
228. 8:37For every season since 2000, I made
229. 8:40a prediction using only the seasons
230. 8:41before it, and then I checked it against
231. 8:44what really happened.
232. 8:46The forecast that assumed nothing changes
233. 8:47was too low in 26 of 27 seasons, and
234. 8:51since 2018, it's been off by about
235. 8:545 points a year.
236. 8:55The best forecast simply used last
237. 8:57season's number.
238. 8:59Basically, the league moved to a new
239. 9:01level and stayed there.
240. 9:03So the rise in injuries is real.
241. 9:06And that brings me back to this season.
242. 9:08Through each team's first three games,
243. 9:11rotation players miss 15.5% of their
244. 9:12games. That ranks 16th of 38 seasons,
245. 9:16and it's lower than the 20.4% at this
246. 9:18point last year.
247. 9:20Stars missed even less, 8.6%, which
248. 9:23ranks 25th of 38.
249. 9:26So the opening week was actually normal,
250. 9:28but the big headlines all came in a
251. 9:30few days right after it, and it's not
252. 9:32looking good.
253. 9:33And the first three games is a tiny
254. 9:34sample because in past seasons, the
255. 9:36first week only loosely predicted the
256. 9:38full season.
257. 9:39So are NBA injuries really statistically
258. 9:41getting worse? And the answer is yes,
259. 9:44they are. It's becoming a problem.
260. 9:46Players miss far more games than they
261. 9:49did in the 1990s and 2000s, and last
262. 9:51season was the worst on record.
263. 9:54But they're not the injuries you'd
264. 9:55guess. Most of the rise is absences
265. 9:58of 1-3 weeks.
266. 10:00And so far the Achilles wave of 2024-25
267. 10:02looks like a cluster, not a trend.
268. 10:05But we need to keep an eye on it for
269. 10:07this year.
270. 10:08The biggest change is the stars.
271. 10:10They used to be the most available
272. 10:11players in the league, and now they're
273. 10:14the least available.
274. 10:15And the usual explanations don't fit.
275. 10:18Whatever changed, it changed around
276. 10:192017, and it hasn't gone back.
277. 10:22What do you guys think the NBA should
278. 10:24do about the rise in injuries?
279. 10:25Let me know in the comments below.
280. 10:27And if you want to go deeper, go to
281. 10:28spreadsheetsports.com. It's the most
282. 10:30advanced basketball analytics site
283. 10:31out there, and it's linked in the description
284. 10:33below.
285. 10:34I track things like every team's title
286. 10:36odds and every player's MVP chances,
287. 10:38so you can see how injuries change
288. 10:40the races as the season goes on.
289. 10:43Thank you all for watching, and if
290. 10:44you enjoyed this please like and subscribe.
291. 10:47More content coming for you guys.