Internal player development analysis?

[Crow]

Has anyone looked at year to year player performance change by a one metric team by team on average for say 5-10 years? And by player year (average for yr 2 change from yr1, yr 3 from 2, etc.) and position (or role)? Anyone want to? Many GMs talk about the importance of internal development. Who has a record of doing it well? Who is below average? Are the yr to yr curves different (early vs middle or late)? Would help with evaluating organizations, player development to date and still expected. Could help with trade considerations as well.

[jgoldstein34]

I've been thinking of a good way of doing this for a bit, not sure there really is one. Not many organizations where the GM or coach has been the same person for long enough to really get a look at anything outside of noise. If anything, the best measure of this might just be how long the GM or coach has been employed with league turnover being so high. Still may look into this despite that grain of salt, but I think it may lead to some odd results.

I'd be interested to see if Hawk's University is really a thing or if the Spurs really are that much better at developing talent.

[Crow]

Obviously internal development will be affected by draft pick levels. An adjusted analysis might be worthwhile. But a simple version is a starting point. Focusing on guys under 27 would chop the work and put the focus where intended.

I'll probably do a sample team. I may eliminate trivial players below a minutes threshold.

[Crow]

I looked at the Thunder from 2007-8 to 2016-17. I included everyone who was on team for at least 2 years. The yr to yr change when the team was different in the pair was excluded because that is team change as well as internal development. The averages are not minute weighted. But fwiw:

For guys in year 2 of their career, the average Winshares per 48 gain over year 1 was .042. I won't know if that is good, bad or average until I do other teams. For year 3 over 2, up .027. Decelerating but still modest progress. Year 4 over 3, a decline of .008. Year 5 over 4, a bigger decline of .0223. Year 6 over 7 almost unchanged. Year 8 over 7, down again -.01. Year 9 over 8 down .04.

So for Thunder players under 27 after the third year players on average decline or stay the same year to year. That would often be after getting the next contract. Typical? I dunno but I'd guess not. Maybe I have a wrong impression but I would have expected improvement in year 4 and perhaps 5. Regardless, concerning, I think. This is different from straight age curves but I think they went up to 27ish, right? But year 4 might be 25, 26 here, so maybe not that big a difference. If it was 27 or higher, excluded. Who gains or losses gain matters a lot but better to know the averages than not.

[jgoldstein34]

Really interesting to see with the Thunder, but I guess that helps explain a bit how they really don't have that many solid role players. Maybe there isn't much opportunity with 2 big stars at such high usage that impacts development at a certain point. Regardless, would be fascinated to see this compared to other teams.

Typical aging curves normally go to around 27-28, but I'm not certain how many of these curves were created so they may be a bit of a guess.

[jgoldstein34]

I ran similar analysis for the spurs, going from the 2007-08 until 2016-17. I added weighted WS/48 based purely on MP to try and work out any outliers. Also included average weighted age, by minutes played, for each year.

Code: Select all

Year Change  Weighted WS/48  Unweighted WS/48
   1 -> 2         0.0153          0.0398
   2 -> 3          0.026          0.0186
   3 -> 4         -0.011          0.0065
   4 -> 5          0.019          0.0122
   5 -> 6         0.0158          -0.011
   6 -> 7        -0.0358         -0.0083
   7 -> 8         0.0086          0.0167
   8 -> 9         0.0037          0.0003

Year  Avg Weighted Age
   1         22.92
   2         23.59
   3         25.41
   4         25.85
   5         26.97
   6         27.58
   7         27.67
   8         28.87
   9         30.77

I'd suspect that the Spur's avg age is a bit older than most teams, given that they don't draft very highly in the draft. Could be more useful to change this analysis to being run by age instead of by experience in the league. If there's anything I can take away from just the 2 teams we've looked at so far, it would be that players are who they are really by their 3rd year in the league. Small changes for each going forward after that, but pretty steady. Makes me think that we overvalue development in terms of adding to winning at a certain point. Players might add more skills, but they don't necessarily translate to more wins.

[Crow]

Thanks for that.

Weighting does matter.

[Crow]

On unweighted Thunder a bit better gains in year 2 &3, Spurs better in 4&5.

[jgoldstein34]

I haven't looked at the average age for the Thunder players in those years, but I'd wager they are 1-2 years younger. I'd guess the fact that the Spurs do a bit better in later years than the Thunder is a sign of better actual development outside of just young players improving because that's what young players do. Could be that the Spurs are better able to teach a role to a player.

[Crow]

Age is more traditionally used. Early on it is a lot about maturation. Still a form of development.

You are probably right about the average age of Thunder. That would make their early gains a lot about maturation and the drop off a bit later even more about lack of internal development gains from organizational input after year 3.

[Crow]

Same time period, unweighted.

yr 1-2 -.0013
2-3 .0515
3-4 .0225
4-5 -.014
5-6 .0324
6-7 .003
7-8 .006667


Before Coach Bud
yr 1-2 -.0021
2-3 .0625
3-4 .0345
4-5 -.009
5-6 .235
6-7 .003
7-8 .01

Hawks do have the most cumulative positive development (of the yr to yr changes, again unweighted) and least slippage of the 3 teams checked. Thunder and Spurs close on simple unweighted cumulative gains but Thunder had the most slippage.

Rather than focus on these numbers per se or worrying too much about lack of precision / noise, use it to ask what they did big picture most right and least right and examine the specific cases with a more holistic approach.

Coach Bud
yr 1-2 -.008
2-3 .0405
3-4 .0044
4-5 -.0215
5-6 .0324
6-7 .0205

Mostly better internal development of young guys before Coach Bud.

Weighted may be different but I kinda wanted to go unweighted.

I might do more later. But anyone wants to add teams or do the whole thing (this way or their way), go for it.