THE BOOK--Playing The Percentages In Baseball

Wednesday, July 23, 2008

Mo, Bill James, and me

Bill James posted Mariano’s seasonal Win Shares - Loss Shares, whereby he gives him a total of 136 Win Shares and 24 Loss Shares.  I replied:

Mariano’s 136-24 WS/LS means that he has 45.3 wins and 8 losses.  That gives him 53.3 game slices.  If you take Mariano’s IP, divided by the Yanks team IP, multiply by 0.35 times the number of team games played, Mariano gets 37.4 game slices.  We need to leverage those.  The typical closer gets his peformance to affect the game twice as much as the standard pitcher, “pound per pound”.  He had time as a starter and setup guy, so let’s figure he should have about 65 game slices.  From this little viewpoint, it looks like WS/LS doesn’t give closers the right balance.

45.3 wins and 8 losses means that Mariano is +18.7 wins above average (a .500 pitcher would have gone 26.6 - 26.6).

According to Fangraphs.com: http://www.fangraphs.com/statss.aspx?playerid=844&position=P
Using the change in Win Expectancy states (Win Probability Added), Mariano is +43 wins.

Even unleveraged, if you look at Baseball-Reference.com, he has a 2.30 ERA in a league/park of 4.54 with 1000 innings (including 2008).  A simple runs above average gives you that he is +2.24 runs per 9 inning above average, which over 1000 innings is +249 runs, or close to +25 wins.

His WPA/LI at Fangraphs (which is Situational Wins, without the extra effect of the leverage) is +26 wins.

At this point, it looks to me like WS/LS doesn’t fairly value the effect of Mariano Rivera’s regular season.