Krishna Narsu
@knarsu3.bsky.social
@bball-index.bsky.social, LEBRON creator
Twitter: https://twitter.com/knarsu3
Twitter: https://twitter.com/knarsu3
Yeah I just fixed that. I forgot to update the predictive LEBRON sheet with the correct minutes but fixed it for the team ratings.
October 6, 2025 at 8:33 PM
Yeah I just fixed that. I forgot to update the predictive LEBRON sheet with the correct minutes but fixed it for the team ratings.
Also I made that exact mistake earlier so I know I fixed it lol.
October 6, 2025 at 8:16 PM
Also I made that exact mistake earlier so I know I fixed it lol.
No I'm using the minutes projections for the 25-26 season. I just didn't include the rookies that didn't have a predictive LEBRON. But they are included in the team ratings. Some of the columns in the predictive LEBRON are messed up tho- I forgot to include the correct Team for this year.
October 6, 2025 at 8:16 PM
No I'm using the minutes projections for the 25-26 season. I just didn't include the rookies that didn't have a predictive LEBRON. But they are included in the team ratings. Some of the columns in the predictive LEBRON are messed up tho- I forgot to include the correct Team for this year.
Reposted by Krishna Narsu
Check it out: www.sports-reference.com/blog/2025/08...
Massive Expansion of Play-by-Play and Stats on Pro Football Reference | Sports-Reference.com
Massive Expansion of Play-by-Play and Stats on Pro Football Reference
www.sports-reference.com
August 7, 2025 at 5:09 PM
Check it out: www.sports-reference.com/blog/2025/08...
The accuracy on the testing data set for the predict_series_ridge model was 74% which is ok but I think can be better (the model was kinda put together too quickly). Accuracy on predicting the exact series score was 20.6% which is tough to do.
April 19, 2025 at 4:43 PM
The accuracy on the testing data set for the predict_series_ridge model was 74% which is ok but I think can be better (the model was kinda put together too quickly). Accuracy on predicting the exact series score was 20.6% which is tough to do.
My personal preference among all the models is the predict_series_ridge model. It includes all of DARKO, predictive EPM and LEBRON and it does not rely on exact minutes projections (which is where OKC gets underrated cause the minute projections have Chet at 16 mpg).
April 19, 2025 at 4:43 PM
My personal preference among all the models is the predict_series_ridge model. It includes all of DARKO, predictive EPM and LEBRON and it does not rely on exact minutes projections (which is where OKC gets underrated cause the minute projections have Chet at 16 mpg).