THE BOOK--Playing The Percentages In Baseball

Stephen Strasburg’s 1st start vs. the Pirates was actually a xFIP negative perfect game if that is possible. He had a -0.50 xFIP. I don’t think many players, if any in history accomplished that.

What an interesting question, and a simple one to answer, thanks to BaseballReference’s Game Finder:

http://www.baseball-reference.com/play-index/share.cgi?id=EyJZi

12 Games since 1919, where the pitcher threw a complete game, no HR/BB/HBP, and FIP <= 0 (as defined above).

The most amazing things of those 12? Clemens threw 3 of them, and Gooden threw his two in back-to-back starts.

Prior to 1993, the constant was 3.0.  Can you redo and show those results?

Thanks for linking to my piece. I must agree with you entirely that it was a big omission to exclude HBP. I like the idea re: an FIP of 0.00. The A’s don’t have any then, even their actual perfect games don’t meet that threshold!

Since the minimum Bill James game score for a perfect game is 87, how about a game with zero BBs, zero HBPs, zero HRs and enough Ks to get a FIP game score over 87, I don’t know the IPvalue so I can’t calculate FIP game score.

Okay, I found out the IPvalue is 3, that means 14 Ks, that’s very hard to do.

I used retrosheet/sql and also found 12 games meeting the original criteria with a CG.

Interestingly Kid K’s 20 strikeout game is left off because of the HBP to Biggio. Sam Mcdowell also did it on May 1st of 1968 but hit the first batter of the game! He was also the only pitcher of the group to surrender a run.

5 more players then meet the pre-1993 criteria of at least 14 K/9 IP. Additionally, Terry Mullholland struck out 14 with no BB/HBP/HR in a CG in 1993.

Vida Blue did it in 11 innings. Frank Tanana gave up 9 hits.

Sam Jones was the first, 1960; Eric Bedard the most recent in 2007.

Tango #3 -

This is basically an update of Mike G’s list in #2 for the pre-1993 FIP.  I’ve got 14 complete games with at least 14 Ks (we’re going to exclude some 8IP CG losses with 13Ks) and no BBs, HRs, or HBPs.

There are 3 8IP losses, one by Dwight Gooden, one by Fergie Jenkins and one by Terry Mulholland (this was IN 1993; did the FIP constant change after 1992, or after 1993?).  The game scores range from Luis Tiant’s 10 IP masterpiece with a GS of 99 to Urban Shocker’s 4R, 8H win, with a GS of 69.  Maybe Shocker knew how to pitch to the score.

Here’s the list:
http://bbref.com/pi/shareit/sQ7Nv

Here’s the 1993-2011 list (5 games):
http://bbref.com/pi/shareit/WsIvc

I love how everyone has ideas bouncing off everyone else.  Great job to all!

http://uzrillusion2.blogspot.com/2012/01/more-on-perfect-fip-games.html

A problem with this is that someone who gives up hits faces more batters, thus gets more chances to strike someone out.  In a perfect game that was also an xFIP perfect game the pitcher would strikeout 14 (or 15) of 27 batters, or a little over half the batters.  Thus for every two hits the pitcher should need an extra 1+ strikeouts - otherwise someone who pitched a perfect game with 13 strikeouts would have really pitched a much more perfect game (in terms of thing he could control) than someone who gave up 7 hits and struck out 14 (Hello Fergie Jenkins 1971-07-24).

Looked at through this lens, most of the xFIP perfect games become less than perfect.  The Koufax perfect game is still there, of course, but of the others only Van Mungo, Tiant, and Clemens’ last two still qualify.  Tiant qualifies even if we include in the extra plate appearances for both the hits and the extra inning.

Excellent point.  I have a more accurate FIP version that uses PA, not IP.  IP after all is only used for its ubiquity, and works very well the seasonal-level.

For what we are doing here, the more accurate version with PA should be used.

In the FIP equation, you basically change “IP” to “PA/4.3”.  It’s obviously not as elegant as the traditional FIP formula, but this is the better way to do this.

So, for a traditional perfect game where the pitcher faces 27 batters, he would need to have 3.2 = 2*K/(PA/4.3) = 2*4.3*K/PA

Which means:
K = 3.2 * PA / 8.6 = 0.37 * PA

Facing 27 batters means you only need 10 K.

Thank you very much for making this point, and reminding me that the better FIP equation is more appropriate here.

If I have the numbers right, then tango #11 actually makes a bunch more games qualify (assumption: BF according to B-R = PA).

Click my name for a google doc that has 105 games that I believe qualify.

I made a google doc too (in username), except I returned a total of 109 games. This includes 8 ‘traditional’ Perfect Games, 3 Losses, 3 3-run games, and two ‘no-hitters’ with 28 BF.

I wonder which 4 I included that weren’t in yours.

woops.

this link should work.