I get a lot of questions from people about the process I use for making guesses at the future performance of pitchers.  Instead of responding to three or four emails I decided to put down my “process” for all to see.  I will use upcoming free agent Barry Zito as my example.  I’d also like to point out that it is best to look at three years of data as opposed to one year, but I will be using mostly just one year for the purposes of this article.

 

First, I rely on the stats provided by Fangraphs and The Hardball Times.  The first thing I want to do is validate the ERA.  I like to do this by using Left on Base percentage (LOB%).  LOB% is the percentage of baserunners allowed that didn’t score a run.  The league average LOB% is typically about 71% (in fact, as of August 18, the LOB% in both leagues was 71%).  If a pitcher has a LOB% of less than 68%, then luck was not on his side.  A LOB% of greater than 74% tells me the pitcher was lucky in this regard.  Barry Zito’s LOB% is currently 79.7%.  We can expect this number to come down, if not this year, then next.  Be aware that as this number comes down, his ERA will likely go up.  To give you an idea of just how much it might go up, adjusting his runs allowed total until his LOB% reaches a more typical 71% level increases his ERA all the way up to 4.69.  That is quite a difference.

 

(As an aside, the following current Orioles pitchers have been unlucky this year: Byrdak, Ortiz, Loewen, Birkins, and Lopez.  The following have been lucky based on this particular stat: Manon, Ray, and Rleal.)

 

The next thing I want to do is to validate the number of hits allowed.  We can say with a pretty good amount of certainty that a pitcher has a lot of control over whether the batter hits a groundball or a flyball.  Current data supports that pitchers have little control over whether a batter hits a line drive (even though flyball pitchers tend to allow more line drives than groundball pitchers).  Current data also suggests that pitchers don’t have a lot of control over whether a flyball becomes a popup or an outfield flyball (again, flyball pitchers will typically induce a higher percentage of popups).  On average 20.4% of all balls put in play become line drives, and 13.3% of all flyballs become popups. 

 

So how do we use this information?  Barry Zito has a Line Drive percentage (LD%) of 16.6%, and an Infield Flyball percentage (IFFB%) of 11.9%.  Assuming that Zito will follow typical patterns (not necessarily a safe assumption, but one we almost have to make), we can expect his LD% to go up either this year or next.  This is important because 73% of all line drives become hits.  He has allowed 83 line drives this year.  Increasing his LD% from 16.6% up to 20.4% would equate to an additional 19 line drives (and about 14 hits).  As for the expected increase in IFFB%, we can expect an additional 2 or 3 popups.  Unfortunately for Zito such a small increase won’t help him that much.  The bottom line is that we shouldn’t be surprised to see a slight increase in the number of hits allowed by Zito next season – his new home park and defense will obviously play a role here as well.

 

That brings us to his home runs allowed total.  To do this, we’ll look at his HR/FB%.   This stat is determined by dividing the number of home runs by the number of flyballs (both infield and outfield).  The typical league average is about 12%.  This is important because like line drives and popups, a pitcher does not have a lot of control over whether a flyball becomes a home run.  So if you see a Josh Beckett with a HR/FB% of 18%, then it is almost guaranteed that this number will drop.  Just the opposite applies to Joe Blanton’s 6.6% rate.  Zito’s rate is 10.1%, so we could expect maybe a slight increase. 

 

What do we know so far?  Thanks to LOB% we have an idea that he probably should have allowed more runs this year – as many as 20 more runs.  Thanks to LD% and IFFB% we know that his hits allowed total is close, but may go up slightly.  Thanks to his HR/FB% we have an idea that he might be expected to allow more home runs next year.

 

Still not sure what to think?  There’s another tool we can use – xFIP.   Information for this stat can be found on THT’s glossary page (see links at the end of this article).  The difference between FIP ERA and xFIP is that the home run portion of xFIP is adjusted for batted ball data.  Zito has an actual ERA of 3.83 and an xFIP of 5.33.  That’s a big difference.  Based on this and his high LOB%, I’ll go on record predicting that we will see an increase in his ERA next season.  Of course, in fairness to Zito, xFIP minus ERA predicted an increased ERA in each of his past two seasons as well – but he also had “normal” LOB% rates the two previous seasons.

 

One final thing I like to look at is the pitcher’s trend rates – specifically command and strikeout rates.  Maybe the best predictor of future success or failure for a pitcher is Command Rate (K to BB Ratio).  Zito’s Command Rate trend of 2.01 – 1.92 – 1.63 just isn’t very promising.  Another good indicator is K-Rate.  Zito’s K-Rate trend of 6.89 – 6.74 – 6.29 is another not so promising trend for Zito.  If I’m a team with money to spend, I make the decision to spend it on someone not named Barry Zito.

 

How about a couple of other high profile free agents?

 

	Schmidt	Mulder	Lilly	Mussina	Marquis
LOB%	76.1%	67.6%	76.6%	71.3%	67.2%
LD%	18.6%	20.8%	19.4%	17.3%	15.8%
IFFB%	9.7%	1.4%	14.5%	13.0%	10.0%
HR/FB%	6.0%	26.1%	12.1%	9.9%	13.1%
xFIP - ERA	1.43	-1.74	0.52	0.31	0.09
Command	3.26 1.94 2.38	1.69 1.59 1.71	1.89 1.66 1.84	3.30 3.04 4.63	1.97 1.45 1.38
K-Rate	10.04 8.63 7.57	5.58 4.87 4.87	7.66 6.84 7.60	7.21 7.11 8.06	6.17 4.35 4.30
	Improve	Decline	Same

 

The green boxes indicate that we can expect improvement, yellow indicates an upcoming decline, and white indicates “same old, same old.”

 

Jason Schmidt: we can expect to see an increase in ERA, triggered primarily by an increase in home runs allowed.  A declining K-Rate is a bit of a concern.  In addition, if you go back to 2003 and take out an injury plagued 2005 his Command Rate is in decline as well.  Having said that, he’d still be the ace of the Orioles’ staff because the starting point for his ERA is so low.

 

Mark Mulder: we can expect to see an improved ERA, and fewer home runs and hits allowed.  However, the starting point for his ERA is so high there is no way I’d give him a big deal – especially when looking at those K-Rates.

 

Ted Lilly: LOB% tells us his ERA is artificially low.  Otherwise, I’ll be the first to admit to just not being excited at all by his numbers – either in a good or a bad way.

 

Mike Mussina: we can maybe expect to see a few more hits, but his overall numbers suggest that his 3.54 ERA isn’t a fluke.  Still looks like an ace to me.

 

Jason Marquis: he’s a little like Mulder in that his ERA figures to go down, but that it is starting at such a high point, it still doesn’t figure to be all that impressive.  You also have to cringe when looking at his Command and K-Rate trends.

 

So of these five and Zito, if I were a GM in search of a starting pitcher I’d be going after Jason Schmidt and Mike Mussina.

 

So what about the current Orioles starters?

 

	Bedard	Cabrera	Benson	Loewen	Lopez
LOB%	73.2%	70.7%	72.4%	62.7%	65.9%
LD%	19.2%	23.0%	18.6%	22.1%	21.7%
IFFB%	7.0%	10.6%	7.5%	11.5%	8.0%
HR/FB%	9.2%	7.7%	13.4%	9.6%	14.9%
xFIP - ERA	0.33	0.13	0.67	-1.16	-1.43
Command	1.70 2.19 2.40	.85 1.80 1.33	2.20 1.94 1.45	1.35	2.24 1.87 2.41
K-Rate	7.93 7.94 7.57	4.63 8.76 9.37	6.02 4.90 4.20	8.07	6.38 5.07 6.21
	Improve	Decline	Same

 

Lopez and Loewen figure to show the most improvement based on this analysis.  Improved Command would make all the difference for Cabrera.  Kris Benson figures to be the least likely to improve next season.

 

NOTE 1: Please don’t take this article to imply that this is a foolproof method.  Barry Zito could very easily have a career year next season while Schmidt and Mussina could each easily fall flat on his face.  This is intended as strictly a quick and (hopefully) easy method of guessing how well a pitcher will perform next year based on his current stats.

 

NOTE 2: The THT Statistics Glossary contains definitions for the stats referenced in this article. 

 

NOTE 3: I will write a companion (read more stats oriented) article and post it on my blog.