Optimizing Marcel Projection Weights for NPB — Grid Search + Bootstrap Validation

📰 Dev.to · YMori

The conventional Marcel weights (5/4/3) used in npb-prediction aren't optimal for NPB. I ran a 720-combination grid search on 11 years of NPB data and found statistically significant improvements (p=0.003).

Published 7 Mar 2026

Full Article

The conventional Marcel weights (5/4/3) used in npb-prediction aren't optimal for NPB. I ran a 720-combination grid search on 11 years of NPB data and found statistically significant improvements (p=0.003).
Read full article → ← Back to Reads

Related Videos

Watch this before applying for jobs as a developer.
Watch this before applying for jobs as a developer.
Tech With Tim
How to Find a Profitable Claude AI Business Idea (Using Real Data)
How to Find a Profitable Claude AI Business Idea (Using Real Data)
Conor Martin
5 Funny Micro SaaS Ideas
5 Funny Micro SaaS Ideas
Learn With Shopify
Do You Know the Difference Between Investing and Gambling?
Do You Know the Difference Between Investing and Gambling?
Charles Schwab
How to Become More Creative | 5 Practical Steps That Actually Work | #skillsforfuture
How to Become More Creative | 5 Practical Steps That Actually Work | #skillsforfuture
Simerjeet Singh
Coco Gauff's Brand Empire | How the World's Highest-Paid Female Athlete Is Building
Coco Gauff's Brand Empire | How the World's Highest-Paid Female Athlete Is Building
Make More Money Now