During the 2023 NBA playoffs I saw a lot of chatter about the Golden State Warriors’ roster and if their reported “two timelines” approach would work as they attempted to repeat as NBA champions. Of course, they did not repeat as champions as the Denver Nuggets ascended to the throne and claimed their first NBA title. However, it got me thinking about the makeup of contenders’ rosters and particularly how they allocate roster spots and minutes to younger players. Conventional wisdom is that contenders rely on star veterans for the main contributions, but use late career vets looking for a ring over a payday and cheap first contract players to balance out their rosters and offset the higher salaries of their stars. The goal was determining to what degree that is true, how it differs across subsets of teams, and if it has changed over time.

I gathered regular season and playoff data on rosters and minutes played from the 2000-01 season onwards as the basis for my analysis. I calculated each player’s age and years in the league for the corresponding season before adding categories for “under 23” and “first four years in the league”. Under 23 was chosen as a qualifier for young players because 23 is commonly the last age for any sort of youth tournament in worldwide sports and is around the age of many college seniors, while the first four years in the league matches up with the contract length for first round picks (and some second round picks). I chose to try both definitions of a young player to capture any potential differences with how teams treated players entering the league based on their prior experience.

From there I moved to categorizing teams as non-playoff, playoff, and contender, where contender is simply the final four teams and the playoff teams group excludes contenders. I opted for that instead of measuring by regular season net rating or betting odds mainly for simplicity. Therefore each player-team-season combination were also labeled as being regular season or playoff data and non-playoff, playoff, or contender. I rolled up the data by season type (regular season or playoff) and contender status and took the mean number of minutes and roster spots allocated to players for both the under 23 and first four years definitions. I initially tried splitting the graphs out by season as well as season type and contender status, but found them to be spiky and not display any sort of trend outside of the KD / Westbrook / Harden Oklahoma City teams being true outliers.

First Four Years

Players in their first four years are clearly given less opportunities as teams get better, hardly a surprising result. Non-playoff teams are allocating over half of their roster spots and about half their minutes to these players while the difference between the roster and minutes percentages are larger for better teams. The drops in roster percentages from regular season to playoffs for playoff and contending teams is likely due to cycling through younger players at the end of their roster. Similarly, the drop in minutes percentage from first two round teams to contenders in the playoffs could be due to changes in rotations being cut shorter and shorter further into the playoffs and is an area for investigation.

Under 23

These charts show how relatively fewer opportunities are given to players under 23 (u23) compared to those in their first four years. Some nuance is missing there as players often move from in the u23 group to out of it while still in their first four years, whereas it’s rare for players to be in their fifth year while still under 23. However, we still see non-playoff teams devoting more resources to u23 players with about one fifth of their roster spots given to these players and, reflecting these players’ likely status as lottery picks, a bit more than their share of minutes.

In an interesting contrast to the first four year graphs, roster percentage increases in the playoffs for u23 players for both contenders and first two round teams. My theory for this seeming discrepancy between the two groups is that the end of roster that I posited as the cause for the drop-off in first four year players is mostly concentrated in end of bench players that are less likely to be tied to the team through draft capital.

Once again, teams that reach the playoffs give u23 players less than their “fair share” of minutes, though there is only a small drop in minutes from regular season to playoffs for contenders and none at all for first two round teams. I think this is likely due to these players already earning their coach’s trust during the regular season and carving out a niche in the rotation, though it would be interesting to see how those minutes progress by round in the playoffs.

Future Work

There is plenty of room for future work in this vein as I largely confirmed existing knowledge around young players and contending, though highlighting how this works in practice is still valuable. Potential avenues for further research is breaking out the playoff statistics by round instead of contender / first two round teams, checking on players over 35, and breaking out the ages more granularly instead of relying on these broad buckets. It’s also interesting to think about how legitimate of a strategy Sam Hinkie actually had - considering the youth typically associated with top draft picks and the time it takes players to reach a top tier of championship equity, does this strategy improve the marginal chances of contending or does the long runway invite too many ways for the process to be disrupted?

Conclusion

This work mostly confirms existing thoughts around roster construction and contending. One thing missing from this high level aggregated type of analysis is the allowance for outliers such as the KD / Russ / Harden OKC teams. It’s also worth mentioning that the 2022 NBA Champion edition of the Golden State Warriors did rely heavily on the contributions of Jordan Poole, who was under 23 and in his first four years, while rostering Moody, Kuminga, and Wiseman, though they relied on these players and others qualifying as u23 or first four years less than the 2023 team.