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
If you are a media member and would like a review copy of The Book, please contact Kevin Cuddihy of Potomac Books.

Buy The Book from Amazon

MOST RECENT ARTICLES
MAIL : You ask | We say

Advanced


THE BOOK--Playing The Percentages In Baseball

<< Back to main

Monday, July 24, 2006

More WPA

By Tangotiger, 07:35 AM

Studes takes another look at WPA over at Hardball Times.

I would also suggest to newcomers to read the previous blog entry on this topic here

Now, as for the specific mechanics…


1. BaseRuns is appropriate for pitchers, not for batters.  Linear Weights is appropriate for batters, not for pitchers.  But, in a pinch, it really doesn’t matter.  They’re both close enough.

2. The clutch calculation should be WPA - (Runs*RPW*LI).  It doesn’t look like Studes used the LI part (but I can’t tell).  The RPW should be close to 11 or so.  You could make the argument that you want WPA/LI - Runs*RPW.  It may look weird, but that first term is really sum(WPA) / average(LI).  So, a guy can get an LI of 1.00 on average, but if he can pump up his good outings in high-leverage situations, he’ll get the benefit of his clutch performance.  The second equation makes things fairer, especially when you include relievers.

3. He keeps saying things like 17 WPA points (or 34 wins).  In fact, if you are talking about WPA points, which is win points above AVERAGE, the you need to talk about wins above AVERAGE.  Saying “34 wins” really is misleading, because you don’t know what the basis is.

4. To get wins and losses out of WPA, you have to understand that WPA is itself the result of win advancement (WA) minus loss advancement (LA).  I would much prefer that everyone who produces WPA-type charts to include the win and loss advancements.  The Mills Brothers’ Player Win Average for example, did win points / (win + loss points).  It’s a very nice display, and I would show this in conjunction to win points minus loss points (or WPA).

Anyway, to reverse engineer WPA into WA and LA, you should understand that an LA is caused by an out.  An out, on average, is worth around .026 or .027 wins.  More specifically, it’s .027 * LI.  So, if you have 15,643 starter innings, at an LI of .96, and their overall WPA is +4 wins, then we get this:
LA = 15643 * 3 * .027 * .96 = 1216
WA = LA + WPA = 1220

You will also realize that for a team, the WA and LA will have no bearing on their actual wins and losses, but that their WA minus LA will exactly equally the team W minus L.

In the typical case, if you have a team that has 91 W and 71 L, their WA and LA will be around 123 and 103.

(5) Comments • 2006/07/24 • SabermetricsRun_Win_Expectancy
Page 1 of 1 pages

<< Back to main


Latest...

COMMENTS

Nov 21 17:29
Sabermetric Moves of the 2009 Pre-Season

Nov 21 20:13
Runs Produced

Nov 21 19:27
Marcel 2009 is here

Nov 21 17:50
The New Triple Crown

Nov 21 16:43
Nate Silver: hero to interviewers

Nov 21 10:57
New BBTN

Nov 20 20:34
ABSO-lutely… not!

Nov 20 19:23
R.I.P. Tom Boswell, sabermetrician; P.A.L.L.(*) Tom Boswell, human being

Nov 20 18:06
Top Free Agent Pitchers

Nov 20 17:45
NBA’s Marcel