Berg

NBA draft, analytics, Warriors, Kings

[Berg @berg_hoops]

Updated the app with the 2024-2025 season. Compare players' statistical profile at a glance at https://t.co/2cYq7TFKXr

[Berg @berg_hoops]

@thegrantedwards @jessefischer33 Taking the year off. Appreciate you looking out 🙏

[Berg @berg_hoops]

Draft day! Sharing the 2025 model, which ends up tracking pretty well with consensus.

https://t.co/U6BSbglVuO

[Berg @berg_hoops]

@DBCJason The rebounding / usg / ast combination was pretty unique and compelling. I remember some twitter analysts rly high on him for that reason. In retrospect, prob mostly due to playing in a mid conference.

Don't mean to over-index on one player though

[Berg @berg_hoops]

@bballstrategy Top 9 teams cutoff seems cherry picked? Or is there a reason for that cutoff

[Berg @berg_hoops]

@SamQuinnCBS Maybe some financial hits for group B like lower revenue sharing payments, less chance of national TV games.

Only playoffs for group A, of course. And play-in games would carry huge stakes.

[Berg @berg_hoops]

@SamQuinnCBS Promotion + relegation! Like European soccer.

Schedules stay the same, just that your record is tracked against your division - group A (top 20) or group B (bottom 12, assuming expansion)

Bottom 4 of A go down, top 4 of B go up at end of year. Lottery is B only, flat odds.

[Berg @berg_hoops]

@rzbzd_ @SacTownSadNDown @esidery With the Orlando pick yeah

[Berg @berg_hoops]

@MattJacksonUSF Yeah, I really like his fit there too.

Also - Will Riley is an underrated addition to that group of wings, but he'll need time to get stronger

[Berg @berg_hoops]

Biggest reaches for my model:

Niederhauser to Clippers
Tre Johnson to Wizards
Coward to Grizzlies (I love this pick personally FWIW)

[DK @DKPhiSports]

Best and worst values from Round 1. 6 players were picked exactly where I had them ranked 🔮

[Berg @berg_hoops]

@MattJacksonUSF It's a stats model, not my opinion. Model doesn't like him bc he doesn't contribute much besides shooting. Here he is vs Kon for example:

[Berg @berg_hoops]

Overall, 12 picks taken within +/- 3 spots of the model's ranking, pretty good!

(excluded Int'l players since my model is NCAA only)

[Berg @berg_hoops]

Biggest values:

Kasparas to Heat
Jase to Magic
Asa Newell to Hawks

[Berg @berg_hoops]

@mikegrib8 @taylor_snarr @NBACouchside That's the interpretation, yeah.

Specifically, I take each of the first 6 seasons and weight them something like 35 / 25 / 20 / 10 / 5 / 5 from best to worst. So it rewards "consistently good" players a more than just taking the single best season.

[Berg @berg_hoops]

@thegrantedwards @DietBeginsMon On the other hand, Wolf's shooting has apparently been lights out in workouts and he shot well at the combine, neither of which are factored in here. So the model prob underrates him in that respect

[Berg @berg_hoops]

@thegrantedwards @DietBeginsMon Re. Asa's D-RAPM vs Wolf - Wolf is big and knows where to be, but I suspect his DRAPM is inflated a bit from sharing the floor with another big (esp Goldin). These metrics try to tease out these relationships, but there often isn't enough sample size over a single season

[Berg @berg_hoops]

@mikegrib8 @taylor_snarr @NBACouchside If you want a better idea of the input vars, they look similar to the viz on my web app https://t.co/2cYq7TFKXr

[Berg @berg_hoops]

@mikegrib8 @taylor_snarr @NBACouchside Biggest evolutions over the years are:
- Reducing the dimensionality by consolidating inputs (e.g. one “shooting efficiency” metric instead of mid%, FT%, and 3%)
⁃ Taking each player’s input metrics relative to position group, instead of as an absolute value.

[Berg @berg_hoops]

@DietBeginsMon @thegrantedwards I was surprised too considering the limited creation. Model is likes his shooting projection, esp when looking at FT% and midrange %. Along with solid steal and OReb numbers