[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
THE BOOK cover
The Unwritten Book is Finally Written!
An in-depth analysis of: The sacrifice bunt, batter/pitcher matchups, the intentional base on balls, optimizing a batting lineup, hot and cold streaks, clutch performance, platooning strategies, and much more.
Read Excerpts & Customer Reviews

Buy The Book from Amazon


2013 Bill James Handbook

Advanced


THE BOOK--Playing The Percentages In Baseball

A blog about baseball, hockey, life, and whatever else there is.

Monday, April 20, 2009

Dual-roled pitchers

By .(JavaScript must be enabled to view this email address), 12:41 PM

Rally does the work I’ve been meaning to do.  He looks at how pitchers perform when they pitch as both starters and relievers in the same season.  He seems to do a “matched pair” whereby he makes sure to use the same number of PA(?) in both roles (by using the minimum of the two?), for each pitcher.  Anyway, as shown earlier, it’s the K rate that is the biggest differentiator.  That bumps up by 15%.  If you remove the K and BB, the HR per contacted PA also drops, probably by around 10%.  The H per BIP also drops substantially by 5%. 

In “per 700 PA” numbers, the number of K goes up by 19 (impact of 4 runs), the walks goes down 0.6 (negligible impact), the HR goes down by 2 (impact of 3 runs), and the Hits per BIP drops by 12 (impact of 9 runs).  All in all, that’s a 16 run impact per 700 PA.  700 PA is 162 IP.  As you can see, that’s 0.1 runs per IP, or 0.9 runs per game.

Pretty much exactly the numbers we’ve been talking about.  (Note that part of this may be due to the recent discovery elsewhere on this blog that it could have been that relievers pitch during the colder part of the night.)

In any case, note that the change in hits per BIP was “substantial” but only 5%.  This is because the difference between something being a hit or out is 0.75 runs or so and 75% of all PA are BIP.  So, the little impact on a per PA basis adds up alot.  The signal (per PA) may be small, but when you add it up, it becomes substantial. 

Consider this a nail in the DIPS-coffin if you will.  DIPS gave us a wonderful way of looking at the data.  It just means we have to look way harder to find the signal.

Anyway, then Rally does it, and looks at it by era.  He does the 1977-1992 pitchers and the 1953-1976 pitchers.  And he finds an almost identical effect!  Fantastic!

Now we know.  Just give a blanket 1 run per game adjustment.  The question will be why.  And the answer will probably be a 0.7 run difference based on the “pacing” versus “going all out” and a 0.3 run difference based on pitching in a colder part of the day as a reliever.

(47) Comments • 2009/04/22 SabermetricsPitchers