Saturday, June 15, 2013

The Greatness Number - a slight modification Japanese style

By PATRICK GORDON | Managing Editor
@Philabaseball

A few months back I introduced the Greatness Number, a metric that uses three variables to rank baseball clubs across different eras.

[Read about the Greatness Number here.]

I created the formula as an objective way to rank the greatest (and worst) baseball teams in major league history. 

At the time, I contended the formula was the most accurate and advanced means to truly identify the best and worst baseball teams dating from 1900 to the present.

Makoto Kozuru hit 51 homers and collected 160 RBIs for the 1950 Shochiku Robins, the second best team in Japanese history.I still hold that belief, but some additional research has led me to tinker a bit with the statistic.

Earlier this week I concluded a study of professional baseball teams in Japan dating to 1950. Using  statistics from Baseball Reference, I imported data on more than 750 teams and allowed the Greatness Number formula to rank Japanese clubs from best to worst, just as I had with MLB data.

The original Greatness Number formula based on MLB data from 1900 to 2012: (.503+(OPS*.097)+(RDiff*.006)+(WHIP*-.053).

For those wondering,  OPS, Run Differential, and WHIP are variables in the equation because they are statistically significant to winning percentage [as seen here.]  

There were two somewhat significant glitches I encountered while analyzing the baseball data from Japan. 

First, the number of games played each season varied, so a dominant team could have a significantly more positive run differential because they played more games and perhaps beat up on lesser competition. The opposite could be said for a weaker club, who could have a larger negative run differential because they had to suffer through more games in a given season than their counterpart the year before or after.

The differences in scheduling and games played is not something that I thought of while completing my original study on MLB clubs - likely because the number in MLB has been a near constant at 162 games since 1961.

Secondly, for those unaware, ties exist in Japanese baseball. Yes, you read that right. Typically, a game is declared a tie if the score is even after 12 innings. Unfortunately, the data I obtained from Baseball Reference  does not include information from games that finished as a tie. Not a major issue, but one that limits my data from being being 100 percent accurate. On average, teams tie between four to six games a season.     

To compensate for the difference in games played per season I swapped run differential from the Greatness Number model and replaced it with run differential per game. The change keeps the concept of run differential in the equation, but now it evens things out for clubs that may have played a different amount games each season.

The original (using run differential) Greatness Number formula based on Japanese data from 1950 to 2012: (.489+(OPS*.183)+(RDiff*.000)+(WHIP*-.092). For those wondering, the actual coefficient for run differential is 0.000669.

The modified Greatness Number formula using run differential per game based on Japanese data from 1950 to 2012: (.492+(OPS*.151)+(RDiffPG*.088)+(WHIP*-.077).

As noted in my previous work with the Greatness Number, this metric is meant to be viewed as a winning percentage based purely on statistical ability. It evens the field for every club and allows for a fair and unbiased ranking.

I will release the Japanese rankings in their entirety later this summer, but here are the top 10 clubs in Japanese baseball history from 1950 to present according to the modified Greatness Number (CL = Central League, PL = Pacfic League): 

League YR Team JAPmGN
CL 1951 Yomiuri Giants 0.786652
CL 1950 Shochiku Robins 0.772801
CL 1955 Yomiuri Giants 0.723353
CL 1953 Yomiuri Giants 0.717566
CL 1952 Yomiuri Giants 0.717118
PL 2005 Chiba Marines 0.695593
PL 1978 Hankyu Braves 0.67936
PL 1971 Hankyu Braves 0.674115
CL 1954 Yomiuri Giants 0.673551
PL 1951 Nankai Hawks 0.673186
- Patrick Gordon is Managing Editor of the Philadelphia Baseball Review.  Follow him on Twitter @Philabaseball or contact him at pgordon@philadelphiabaseballreview.com

0 comments: