著者

鳥海 崇 綿田 博人

出版者

慶應義塾大学体育研究所

雑誌

体育研究所紀要 = Bulletin of the Institute of Physical Education, Keio University (ISSN:02866951)

巻号頁・発行日

vol.56, no.1, pp.45-53, 2017-01

---

In this study, we calculate the team strengths of the Tokyo Big6 Baseball League teams from 2001 to 2015 seasons by using five mathematical and statistical methods; which are ① Keener's model using points in the standings, ② Keener's model using wins and losses, ③ Bradley-Terry model using wins and losses, ④ Offence-Defense methods using scores, and ⑤ Keener's model using a parameter derived from points and winning rate. Based on the 20 results whose standings of the points and the winning rates are on the same order, we calculate the Kendall rank correlation coefficient between both rankings derived from strengths and from points in each method. The average coefficients from 2001 to 2015 seasons show that Keener's model using wins and losses, Keener's model using a parameter, and Bradley-Terry model have high correlations (more than 0.9). On the other hand, Keener's model using points in the standings and Offence-Defense model have a low correlation (less than 0.9). We also calculate the strengths in the case that the team ranked 1st in the standings is ranked 2nd in terms of the winning rate. The three methods with high correlations show that the strength of the team ranked highest in terms of wining rate is greater than the team ranked second highest. This implies that the winning rate is a better indicator of a team's strength than its points in the Tokyo Big6 Baseball League standings from 2001 to 2015.