Futurity · Creativity & design

New Model Improves Baseball Rankings

Statisticians developed Bayesian Multivariate Rank Regression (BMRR) to analyze complex sports rankings, accounting for disagreements, incomplete data, and uncertainty.

Rankings from different sports outlets vary, making consensus difficult. BMRR combines rankings from multiple sources and time points, addressing differing player counts and changing opinions.

Researchers tested BMRR on MLB player rankings (2021-2024) from ESPN, CBS, Bleacher Report, Yahoo Sports, and MLB, focusing on 55 consistently ranked players.

BMRR treats rankings as evidence of player value, using a Bayesian framework. It handles incomplete lists and ties, quantifying consensus certainty.

This is crucial as rankings influence decisions like trades and salaries. BMRR identified factors like WAR, age, and salary associated with higher rankings, suggesting future potential is considered.

The analysis showed player trajectories, like Shohei Ohtani's rise. BMRR can also assess the probability of one player ranking above another.

BMRR can evaluate rankers. MLB's and ESPN's rankings aligned most with the consensus, while others showed different tendencies.

BMRR has broad applications beyond baseball, applicable to any field with complex ranking data. The code is publicly available.

AI-samenvatting op basis van de bron.