Comparing NCAA Conf. and International Leagues for Draft

[SeanD]

I am trying to normalize data for a draft model. Obviously comparing and adjusting stats for international prospects based on league is important, but is there any evidence that adjusting for NCAA conference is worthwhile too? Or would it be better to adjust for each opponent since non-conference opponents make up a significant percentage of the schedule?

Andrew Johnson, on his blog 'Counting the Baskets', talks about calculating his European league adjustments based on the performance differences of players who go straight from the NCAA to Europe. I couldn't find any other mention of HOW people are normalizing international prospects' stats. Is there an obvious common method that I'm missing? Do you have any ideas or suggestions about how best to do this?

[Crow]

To comment on part of your questions, I think adjusting for opponents (using SRS for simplicity or opponent adjusted offensive and defensive ratings would be even better) is essential. I don't have analysis to support that; it is a guess based on logic that quality of opponent matters when you are projecting performance against an even bigger step up in opposing talent.

On Euro projecting, people may not be willing to give up secrets that are getting them paid, hope to be paid or at least published and prominent. May have more luck in private and if you have secrets / interesting ideas / data to exchange.

[nbacouchside]

I am trying to normalize data for a draft model. Obviously comparing and adjusting stats for international prospects based on league is important, but is there any evidence that adjusting for NCAA conference is worthwhile too? Or would it be better to adjust for each opponent since non-conference opponents make up a significant percentage of the schedule?

Andrew Johnson, on his blog 'Counting the Baskets', talks about calculating his European league adjustments based on the performance differences of players who go straight from the NCAA to Europe. I couldn't find any other mention of HOW people are normalizing international prospects' stats. Is there an obvious common method that I'm missing? Do you have any ideas or suggestions about how best to do this?

Short answer: Yes. Layne Vashro did a bunch of work on this before getting scooped by a team. http://nyloncalculus.com/2015/11/06/dee ... the-world/

[Dr Positivity]

For me adjusting for conference is critical in the scoring stats, but I haven't seen much evidence that adjusting for conference in non scoring production reb, ast, stl, blk helps more than it hurts. One sign of that is if you look back at the non-scoring stats in those categories of some recent high profile mid majors like Lillard, McCollum, George, Stuckey, etc., it's about what you'd expect from them if they had been in a major conference

[SeanD]

Thank guys. The Layne Vashro article on Nylon was really helpful. And Dr. Positivity, I'll look into your comments. I suspected that not all metrics should be adjusted. I'm interested to see if there are any others besides points that are competition dependent as well. Kevin Pelton, on the most recent Mid-Range Theory Podcast episode, talked about his own underestimation of mid-major players because of an over adjustment of strength of schedule. His idea, which I'm excited to look at, is that after a top tier of talent, most players are pretty similar, whether they play for Syracuse or Southern Mississippi.

Dr. Positivity, any idea why scoring is more effected by normalizing? I would expect to see stats like stls and blks, that are often more indicative of athleticism, show a need for adjustment too.

[Dr Positivity]

I can't really saw with confidence, my best guess would be that scoring has more to do with the talent of their teammates and coaching on their own team than other stats do. Of two players with equal talent but one in a weak conference and one in a major, for the one on a lower conference maybe the coach is running a lot of isos around his skillset, sets up multiple screens for him in a set, puts players on the floor just to compliment him such as floor space. Meanwhile the same talent player who's 4th best on Kentucky doesn't have the offense built around him as much. Maybe he's taking a lot of spot up shots and cutting when he has the skills to iso off the dribble and be pnr playmaker. In general it's probably related to how scoring doesn't translate well from NCAA to NBA, and is also pretty vulnerable to age in college - scoring changes a lot when faced with different competition, different rules, different systems, different age of players