蕭秋銘2026-07-242024-062226-535X10.53106/2226535x2024061302002https://ir.ntus.edu.tw/handle/123456789/131480This study investigates the winning ratio of Major League Baseball (MLB) season games by using a generalized Pythagorean formula. Based on the method of regression analysis, this study verifies the ability of the generalized Pythagorean formula to predict MLB team winning rates. Empirical results show that the multivariate regression model has a highest adjusted R2 (0.9126) than other regression models. In addition, either MSE or MAE of in-sample and out-of-sample, the generalized Pythagorean formula derived from Cobb-Douglas production function has the lowest values than other versions of Pythagorean formula. Based on this generalized Pythagorean formula, team’s managers can acquire potential rookies through drafts or player trades to recruit the players they need to achieve the team’s goals.本文欲藉由廣義畢氏公式法來估計美國職棒大聯盟常規賽的勝率。經由過去10年賽季30支美國職棒大聯盟球隊的攻守紀錄，本文驗證了廣義畢氏公式對於球隊的勝率具有極高的預測能力。實證研究的結果亦顯示，本文的多元迴歸模型相或是二階段迴歸對於Poisson迴歸模型都有較高的預估能力。而且不管是樣本內估計，還是樣本外預測的MSE與MAE，本文所提出的廣義畢氏公式對於勝率的估計都不比其他版本的畢氏公式差。然而，由於本文的廣義畢氏公式是經由Cobb-Douglas產出函數推導而得，因此不僅具備經濟意涵，可以透過計量模型檢驗之，而且此模型更具有彈性(flexibility)與擴充性(extensibility)。再者，球團管理階層亦可透過此公式招募所需的球員，以便提升戰力，提高勝率。棒球畢氏公式Cobb-Douglas產出函數勝敗比Pythagorean Formula for BaseballCobb-Douglas functionWin-loss Ratioarticle