Dialed In - August 30, 2004

Does Steve Trachsel lull his defense to sleep?

Throw the damn ball!

In one of the Primer discussions, it was muttered that slow pitchers have more errors behind them.  We’ve all heard this: the defense gets bored or distracted or tight.  Fortunately for you, one of the most reputable slowpoke pitchers in the game pitches for the Mets.  Which means I’m interested in this.

One of Bill James’ greatest attributes is his smooth challenging of the accepted baseball conventional wisdom.  James wrote about that – specifically that when an old baseball man asserted something, James would ask “why?” and “Is that really true?”

So someone muttered, and I read, that slow pitchers have more errors made behind them because the defense gets, well, something.  I am not sure what happens, but I believe the defenders are supposed to lose their focus or concentration or some such. 

One of the funny parts to this, for me, is that statheads, when discussing clutch hitting scoff at the idea that hitters suddenly turn on “extra-focus” skills.  The numbers have shown, time and time again, that 99% of players hit about the same “in the clutch”.  Statheads will tell you, “These guys are professionals.  The guys who can’t or don’t focus all the time get weeded out before the majors.”

This same logic slips the mind of statheads on the issue of slow pitchers. 

The question is, can we check this?  We can do several things:

1. Identify slow pitchers.

Certainly - Steve Trachsel. 

I don’t know of any other pitcher that is “accused” of being slow.  Please chime in with pitchers that work slow.  Should I use “Time of Game” as a proxy, or throws to first base?  I’m leaning toward throws to first, because that’s what really takes all the time.  Of course, I don’t know where I am getting “throws to first base” numbers yet, but the Internet has everything somewhere.  Jon Daly informed me that in the 1983 Bill James Abstract, Time of Game for pitchers was listed.  I thought I had that, but I don’t.  If someone gets bored, they can scan it and email it to me.

2. Compare slow pitchers’ statistics against their teammates.

Does Trachsel get errors committed behind him at a greater rate than the other Met starters (and so forth for other slow workers)?

3. Which statistics?

I’m looking at errors committed while pitcher X was on the mound divided by the number of innings pitched.  Is that the correct statistic?  Chip in here.

Mike Emeigh has a nifty database, so to get me off to a good start, he sent me a list of errors committed behind all the pitchers in 2003.  I’m pretty sure he managed it through www.retrosheet.org, a fantabulous site, but Mike has a few sources for play-by-play, so I’m not 100% sure.  Nonetheless, Mike was kind enough to send over the error list.

As soon as I mentioned it to Mike, he said that groundball pitchers would probably be the lead victims.  That makes perfectly good sense.  There are obviously more infield errors than outfield errors.  More groundballs make for more chances at errors.  So off I went in search of groundball/flyball rates.  In reality, I went to MLB.com and looked up groundout/airout ratios (GO/AO).  That’s slightly different, but I believe it will check what we want to use it for.  I also did not park-adjust anything.

Armed with these ideas and an Excel spreadsheet, I have now become just dangerous enough to make some incorrect assumptions, convince myself of something that may be completely wrong and waste your time by writing about it.

Who is in the dataset?

Because of the labor-intensive data gathering, I only dug up GO/AO for the first 100 pitchers in the list (there were some 600).  This covered everyone that qualified for the 2003 ERA title.  I did the sort by Games Started and took everyone with 25 Games Started plus Curt Schilling and Jeff Weaver.  Aaron Sele made the cut with 121.2 IP.  Talk about torturing Angel fans – Sele started 25 games and didn’t average getting through the fifth inning.  That’s frightening.  He was the only one that low.  Colby Lewis was next lowest at 127 IP and then it jumps to 142 IP.  There were 12 of 104 pitchers that were included that did not qualify for the ERA title.  No one that qualified for the ERA title was omitted.  The closest was Johan Santana, who started 18 games for the Minnesota Twins in 2003.

Now before everyone starts jumping up and down about one season, this is the first look.  I know that a single season may not tell you the answers.  More work will be done, but here’s what we found out so far.

I looked at errors per 9 IP, balls in play (BIP) per error, GO/AO and unearned runs (I think that’s the discussion that triggered this).

Basics:

For the 104 pitchers, there were 1429 errors.  That represents about 45% of the total errors.  There were more than 20000 IP and 46000 BIP.  The average E/9IP was 0.641 and the average BIP/E was 32.5.  The average GO/AO (straight average, not weighted) was 1.295.

Of course, I don’t know who all the other slow pitchers are, so I’m not sure how it will all shake out just yet, but Steve Trachsel, in 2003, did not see what one might expect compared to his co-starters.

Pitcher	        Team	W	L	IP	Errors	BIP	E/9IP	GO/AO	BIP/E
Trachsel, Steve	Mets	16	10	204.7	10	503	0.440	0.89	50.30
Glavine, Tom	Mets	9	14	183.3	12	468	0.589	1.33	39.00
Seo, Jae Weong	Mets	9	12	188.3	24	455	1.147	0.77	18.96
Leiter, Al	Mets	15	9	180.7	17	403	0.847	1.49	23.71

Trachsel had a lower error rate, but had a low GO/AO, and that may be expected.  However, Jae Seo had a lower GO/AO and had nearly treble the error rate.  Tom Glavine had a higher than average GO/AO and a lower than average E-rate.  Al Leiter also had a higher GO/AO and had a high E-rate.

The r-squared for GO/AO and E/9IP was 0.05.  That says to me that groundball pitchers don’t generate more errors.  I rechecked GO/AO against BIP/E and got a lower correlation.

Compared to unearned runs (UER), E/9IP had an r-squared of 0.32.  I don’t think that UER would serve as a good proxy for whether or not a pitcher is a slow worker, or whether or not his pace is affecting his defense. 

I had made a cursory investigation into Trachsel’s rate of UER compared to the rest of his team’s throughout his career, but the data was fluctuating around average, so I didn’t think I was looking at the right statistic.

I have more work to do – checking seasonal changes for pitchers on error rates and looking at other slow pitchers.  In any single season, a pitcher can get lucky. 

As Bill James often said at the conclusion of his studies, I think that the null set is the answer – there does not appear to be an effect.

And with that, I’ll open the floor for discussion.