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


SABR101 required reading if you enter this site. Check out the Sabermetric Wiki. And interesting baseball books.
MOST RECENT ARTICLES
MAIL : You ask | We say

Advanced


THE BOOK--Playing The Percentages In Baseball

<< Back to main

Thursday, September 23, 2010

Felix Game Score

By Tangotiger, 03:37 PM

Felix had an interesting game.  “He” lost by giving up one run (HR), walking 4, with 5 K, and a single.  He had groundouts coming out of his butt, including 2 DP.  It was a complete game, 109 pitches. 

I introduced 4 Game Scores the other day.  Yeah, I know, 4?  Like, how can the saber market handle 4, all coming from the same guy?  Anyway, they are all well-constructed, and focusing on particular aspects of pitching.  The first one was simply based on runs and innings:
GameScore = 6.7 * IP - 10 * R + 40

Felix had 8 IP, 1 R, giving him a GameScore of 84.  Basically, if he always gives up 1 run in 8 innings, he’ll win 84% of the time.  It’s actually closer to 90% of the time, but this is a crude metric as most linear metrics are, and they’ll break at the extremes.

The second one was based on K, BB, and IP:
GameScore = 0.5 * IP + 3 * (SO-BB) + 40

That gives Felix a Game Score of 47.  Basically, focusing only on these stats, he pitched an average game.

The third one was based on the FIP categories:
GameScore = 2.7 * IP - 13 * HR - 3 * BB + 2 * SO + 40

That’s also 47.  Again, an average start.

The last one focuses on all the outcomes (H, HR, BB) and IP:
GameScore = 8.6 * IP - 5 * H - 8 * HR - 3 * BB + 40

That’s a Game Score of 79.  We’d expect him to win 79% of these games.  That sounds about right.  If you type in his batting line in my Markov calculator, which presumes random sequencing (but no outs on base, meaning no extra benefit from all his GB and possible DP), he should have allowed two runs.  And 4.5 runs scored of support against 2 runs allowed, using pythagenpat gives us a win% of 80%.

So, based on how you want to view Felix’s performance, this is his game scores:
84 - looking at runs allowed (includes sequencing)
79 - looking at batting outcomes (no sequencing)
47 - looking at FIP components
47 - looking at K, BB

We can discuss how much we should weight each number, but, basically, let’s say we do what I said in that thread:

For a single start, I would do 35% version 2 (K, BB), 30% version 3 (FIP), 20% version 4 (component runs), and 15% version 1 (runs allowed).

That gives us a Game Score of 59. 

A simple average of the above 4 is 64.  The simple average looks like this:
GameScore
= 40
+ 4.5 * IP
+ 1 * SO
- 1 * H
- 2 * BB
- 2 * R
- 5 * HR

Anyway, you need to play around with a few of these extreme games to see the resulting game scores and decide the right balance for you.  If we can come to an agreement, all the better.


#1    brent      (see all posts) 2010/09/24 (Fri) @ 00:59

I wonder if anyone could take us (the not so great sabr minds here) slowly through how to figure out what kind of contract would be fair to extend Jose Bautista through his final arb year and maybe two or three free agent years?

Thanks for the help anyone who would do this.


#2    Tangotiger      (see all posts) 2010/09/24 (Fri) @ 01:59

Brent, no hijacking.  You can find another more appropriate thread.


#3          (see all posts) 2010/09/24 (Fri) @ 03:09

Unofficial Leaderboard for Average Game Score:

Column #1 = Pitcher
Column #2 = Average Game Score
Column #3 = Cumulative Game Score

Roy Halladay 63.17 2021
Adam Wainwright 62.51 2000
Felix Hernandez 62.51 2000
Jered Weaver 59.62 1908
Josh Johnson 59.45 1665
Cliff Lee 59.36 1543
Roy Oswalt 59.29 1838
Ubaldo Jimenez 58.87 1825
CC Sabathia 58.56 1874


#4    Brent      (see all posts) 2010/09/24 (Fri) @ 03:12

Sorry, Tango. I didn’t see any thread that fit on the first page. Does anybody check the older pages?


#5    brent      (see all posts) 2010/09/24 (Fri) @ 03:23

You’re gonna make me go back to this thread? http://www.insidethebook.com/ee/index.php/site/comments/jose_bautista/

Cut me some slack- that thread’s over 3 weeks old. :p I can barely remember yesterday. I said I wasn’t a genius ^^


#6          (see all posts) 2010/09/24 (Fri) @ 10:20

I think the ratio is how the numbers will be used.

If it for how the pitcher pitched (results based - looking in to the past) I would go with (looking at the results, runs allowed and component runs, and a little bump for Ks and BBs.):

20% runs allowed
20% K and BB
60% Component runs

Now if the data is to be used in any predictive measure, like MGL was talking about in his recent post on consistent and inconsistent pitchers, I think a formula like the following should be used (basically some combination of of these three formulas, minus runs allowed):

30% K and BB
30% FIPS
40% Component runs


#7          (see all posts) 2010/09/25 (Sat) @ 21:15

I presume it would be preferable to use (BB+HBP-IBB) for BB?

One other question—James’ formula gave a bonus for longer outings.  Is there any logic to this?  I could see, for instance, that the difference between, say, a 7 inning outing and a 5 inning outing would be less than the difference between a CG and a 7 inning game, because you avoid more marginal relievers.


Page 1 of 1 pages


Name (required)
E-Mail (optional; WILL be published)
Website (optional)

<< Back to main


Latest...

COMMENTS

May 25 06:43
Largest demonstration in Canadian history?

May 25 06:39
Lack of hustle during a game

May 25 05:00
Help needed with sticky issue…

May 25 02:38
NFLPA lawsuit against collusion

May 25 01:43
Neal Huntington’s best moves

May 24 23:50
Rooting for laundry

May 24 17:04
Firefox, IE, or Chrome?

May 24 12:07
How to beat the shift

May 24 11:11
Incredible story

May 24 09:41
Racial bias in card collecting: not the collectors, but the players on the cards