This is Part III of the A Hard Look at Jose Castillo series. Part I is here [link] and Part II is here [link].

A follow up article after Castillo lost his starting job is here: [link].

One of the comments made by reader "rick" about my evaluation in Part II was this:

"As far as zone rating goes, my point isn’t that Castillo’s got worse last year. My point is that Castillo’s is consistently bad. And I disagree with your analysis of zone rating. It isn’t perfect, but it’s good."

It’s a great point and one I want to address. One of the best discussions ever held on the Internet about the limitations of STATS, Inc’s Zone Rating (ZR) was in 2005 at The Baseball Think Factory [link]. Not only was ZR explained in detail, many of the web’s most talented, and well known, sabermetric analysts talked openly about its limitations.

Over the last few years, many statheads have drifted to John Dewan’s Fielding Bible as a better ZR source because he eliminates as many of the limitations as possible. He does such a good job, almost all of the teams now rely on his work as a primary source of information outside of their own in-house statistics programs, and Bill James incorporated them into his Handbook starting in 2006.

As such, how did Castillo fare in ZR under Dewan’s system in 2005 and 2006? He actually improved.. a .773 ZR in 2005 to a .777 ZR in 2006 [link to Hardball Times]. The median ZR for all 2B in MLB with 300+ innings at the position from 2004 – 2006 is .815, so Castillo was -.038 in 2006.

What does that -.038 mean? It stands for 3.8% fewer balls in play into Castillo’s zone being turned into an out than the average MLB second baseman. Or, the equivalent of 12 fewer outs (309 Balls in zone * .038) than the league average 2B.

However, Dewan’s ZR doesn’t give credit for double plays and you need to know that Castillo’s 69 double plays turned (DPT) in 2006 was not only the third highest number turned between 2004 and 2006 [link], it was 25 above the 2004 – 2006 median second baseman. Setting that off even more is the fact Castillo only saw 30 more balls in the zone than the average second baseman over the three years.

That’s extremely impressive.

The 25 extra double plays turned converts a much higher out total than the 12 lost above by his slower foot work in the zone. I hear you, why couldn’t he have both? Few players are that perfect, but remember, Castillo only has 2.3 MLB years experience (games played divided by 162). Give him time to develop.

When we revisit the suggestion made by the Pittsburgh Post-Gazette’s Stat Geek [link] the other day that Castillo was seemingly the problem why Jack Wilson didn’t turn as many double plays in 2006, we have to look at what opportunities Castillo had to start a double play to Wilson in 2006.

We saw in the Hardball Times link above that Castillo started 7 more double plays in 2006 than 2005, but played about 400 more innings. First, the 2004-2006 league average second baseman started a double play one time for every 10.2 balls in his zone. Castillo started one every 7.9 in 2006, and 7.8 in 2005.

The fact Castillo was 22% better than the 2004-2006 league average second baseman starting double plays to begin with, makes it almost ridiculous to even wonder why Wilson had fewer double plays turned. But I dug anyway just to see what I could find out just because the Stat Geek was wondering.

Wilson had 62 double plays started in 2005 and just 34 in 2006.. a decline of 28. But he also played 280 innings less. Normalized by the number of opportunities Castillo had to start a double play (ie: less than 2 outs, man at 1B, ground ball hit to Castillo, and Wilson playing SS), Jack Wilson turned 9 fewer double plays in 2006 than in 2005 with Castillo.

So what happened?

Castillo had 102 double play opportunities with Wilson in 2006 and they turned 62.5% successfully, and 76 opportunities in 2005 and turned 57.5%.

However, Castillo did have 9% fewer 4-6 double play starts with Wilson in 2006 as a percentage of his double play opportunities. Part of that 9% was in 4% more outs obtained by Castillo throwing to 1B instead of to Wilson to start a double play.. one time every 2.4 DP opportunity in 2006 vs one every 2.6 in 2005. The mean difference was 5 more.

The other 5% lost was in more singles obtained on ground balls hit to Castillo.. one every 15.2 DP opportunity in 2005 vs one every 8.5 in 2006. The mean difference was 6 more singles allowed.

That 9% shift represented 11 fewer double play starts to Wilson which pretty much explains why Wilson saw his 9 less double plays turned with Castillo. The question now is, why these two events occurred.

Castillo also saw an 11% drop in the number of DP opportunities with a right hand pitcher on the mound in 2006. That’s important because our right hand pitchers have been primarily power oriented pitchers, as compared to our soft tossing southpaws. When I looked at the success rate in MLB of a second baseman starting a double play in 2006 with a righty on the mound, the rate soared above southpaws almost 23%. The obvious reason for that is that balls put in play off a power pitcher are typically hit crisper than from finesse pitchers.

While researching that tidbit I came across Freddy Sanchez’s double play started rate in 2006 and saw it was 25% better than Castillo’s. I quickly found out the reason – Jim Tracy put Freddy at second with a righty on the mound 60% of the time and he benefited from that with a 33% better double play started rate. So when you compare Castillo to Sanchez turning double plays in 2006, you won’t be comparing apples to apples.

The higher single rate allowed by Castillo seemed to be a direct result of balls in play from southpaws because Castillo had 12 total singles allowed in 2006 and 9 of them were from left hand pitchers. In 2005, only 1 of the 5 singles were. So perhaps they were from lite grounders or balls in play up deep up the middle Castillo got a glove on? I don’t know because I didn’t research all 9 to see.

The bottom line is, while Wilson did turn 9 less less double plays in 2006 than in 2005, you have to be hard pressed to even be worrying about a trivial 9 fewer Jack Wilson double play starts when Castillo’s double play start rate in 2006 was 22% better than league average to begin with. But if you want to blame Castillo, be sure to also consider the change in the distribution of balls in play from southpaws.

As for reader "rick" suggesting Castillo is ‘consistently bad’, we’ll just have to agree to disagree. Bad is too vague a term for me. I mean, if 11 fewer DP starts to Wilson is bad, so be it. If Castillo seems ‘aloof’ or ‘lazy’ to the fans, it’s pretty obvious by his statistics in this three-part series he is performing above average in many fielding categories, so what relevance is there to the observation? If you want to complain about 18 errors, then you haven’t read the series. And let’s face it, fielding percentage is almost a worthless stat.

Wrapping up this series I’ll just say that Jose Castillo is far from the best overall fielding second baseman in the game. However, he is one of the best pivot man in the game despite his limited 2.3 years of MLB experience. While his footwork in the zone might be a tad slow, the increased productivity from his cannon arm and pivot far exceeds that lost from his slower ground work.

Now all we can do is pray he finds some more plate discipline, learns better pitch recognition, and continues to mature as a hitter. My bet is that he is about a year or two from breaking out into becoming a solid .280 hitter or better with 20 home runs and above-average defense.

This is Part II of the Hard Look at Jose Castillo series. Part I is here: [link].

The Pittsburgh Post-Gazette ran an article today where they speculated whether or not the 2007 team would be better defensively than last year’s team [link]. One of the statements made in the article by the Pirates beat reporter was:

"Assuming Jose Castillo keeps his job at second base, it is hard to imagine him making 18 errors again, including, amazingly for someone at his position, 15 on throws."

First, the reporter’s statement is factually incorrect — Castillo had three throwing errors last year. I assume the reporter just read his notes wrong. Jack Wilson actually had 3 times as many throwing errors as Jose Castillo did.

Secondly, while we know Castillo made 18 errors last year, nobody seems to know what kind of errors they were. And more importantly, if the errors were the type that any second baseman should have made or not.

Only five second baseman in MLB had 700 or more total chances last year and they averaged 16 errors each, with just one player having less than 14.

Castillo had 8 ground ball errors of the 18 he made — that’s 44%, which ranked him in the 46th percentile of all qualified second baseman (# of Error Type/# Errors Overall). So, in fact, he wasn’t among the worst second basemen fielding ground balls in baseball last year as had been speculated by one writer [link].

Of the eight ground ball fielding errors Castillo made, five were routine grounders he should have handled, two were semi-tough in that the direction of the ball changed within five feet of him, and one took a crazy hop that very few players on planet Earth could have made a play on.

Castillo also had six dropped ball errors. Three of the six were routine plays he should have made.

Two of dropped ball errors were initially charged to other players and later in the year given to Castillo – one was charged to Paulino for bouncing a throw in the dirt to Castillo while covering second on an attempted base steal, and the other was a throw by Sanchez in the dirt while Castillo covered 1B.

The last dropped ball error was charged to Castillo but should have been charged to Jack Wilson as a throwing error, as the throw to Castillo was at his shoestrings while he was covering 2B for a force out.

Then Castillo was charged with three throwing errors on plays where other defenders failed to catch the ball thrown to them. If Castillo should be charged with all six dropped ball errors above from poor throws to him, then I don’t know how he could be charged with throwing catchable balls to other players who failed to at least block the ball.

The first throwing error was a one hopper to Craig Wilson covering first on a nice stop by Castillo who had to throw off balance. But the ball bounced and took a long hop – five foot or more – and came to Wilson knee high or better, but it went off the heel of his glove and Castillo was given the error.

The second throwing error was on a relay throw to the plate and Paulino didn’t even attempt to block the ball that hit right near his feet. While the throw might not have been catchable before the bounce, it was easily a blockable ball and the error should have been on Paulino, not Castillo.

The last throwing error there is no video of and I don’t know what happened except hearing the announcers state that the ball bounced past Sanchez, so I have to assume this was Castillo’s error.

The very last error Castillo made for the year was on a pop up to short right field where Castillo was backing up trying to catch the ball, Burnitz was in a position to catch it but didn’t call Castillo off and Castillo couldn’t see him, and the ball came down hitting Castillo’s mitt and bounced out. It’s a legitimate error called by the official scorer but it’s a tough error for him to have to eat.

A total of five runs directly scored from Castillo’s 18 errors. One from Castillo’s relay throw to Sanchez covering 3B, one run from the failed catch by Craig Wilson, one run when Castillo covered 1B and Sanchez threw the ball in the dirt, and two runs, one from each on the ground ball plays where the direction of the ball changed in the last five feet.

There’s no question that Jose Castillo made a lot of seemingly silly errors. The fact 9 of his 18 errors were made on days after the team had a day off  suggests that Castillo might not be able to get his concentration level up to speed after an off day. Further suggesting a concentration issue is that Castillo commited two errors in one game three times, for 33% of his total errors.

At the same time, we know Castillo started the year off out of shape after Dave Littlefield asked him not to work out over the winter since he had a torn MCL he was healing. And we also know Littlefield rejected Castillo’s plea to play in the World Baseball Classic to help him get in shape.

To that end, 66% of Castillo’s errors were made in the first half of the season and just one of the six errors in the second half were of the routine variety. So there’s no question Castillo tightened his game in the second half.

I have included video below showing every error Castillo was charged with so you can make your own decisions. Here is the official scorers rules on errors from the MLB rule book [link].

Every second baseman gets judged by the same rules, so in that regard Castillo received equal treatment. Yet, I don’t care who you are, and I don’t care how conservatively you view the errors he was charged with watching the video, you’ll be hard pressed to find 18 legitimate errors. 14 maybe.. but 12 or less is probably correct.

After you decide how many errors he deserved, remember that the other five players in MLB last year who had 700 or more total chances averaged 15.5 errors each.

As you listen to the video, listen closely to Gregg Brown and Lanny Frattare as they talk about Castillo and compare their attitude to the tone Castillo receives from the local beat reporters. Then listen to the tone of ex-players John Wehner and Bob Walk in comparison.

Then also remember that at least two (possibly as many as four) errors were originally charged to other players by official scorers and then changed later in the year and given to Castillo, along with the knowledge that the front office of every team has the right to appeal to MLB any official scorer decision.

It’s enough to make you start wondering if someone has it out for Castillo, and why. Here’s Castillo’s error reel:

Alan Schwarz penned a nice article in the New York Times the other day about a high school student from Bloomington, Minnesota, named Victor Wang [link].

The sixteen year-old determined that OPS was misleading when used to try and determine why a team scored as many runs as they did in any given year. He concluded that a weighted OPS system was better and found that 1.8 times OBP plus SLG correlated more closely in determining which players helped their team the most to score runs.

In the book Moneyball, OBP was said to be worth 3 times as much as SLG, so this discovery is somewhat new wave.

When you take Wang’s weighted OBP(*1.8) + SLG result and then divide by 4, you get a new stat sweeping sabermetric land called GPA, or Gross Production Average. The end result of GPA looks much like batting average — .360 is very good, .265 is about average, and .200 is horrible.

The true value of GPA is significantly more important than OPS because it is more accurate in equally comparing players team-to-team on their ability to contribute to runs scored.

Here are a few of the 2006 Pirates GPA’s for 2006:

Bay	0.310		Wilson	0.234
Sanchez	0.288		Burnitz	0.232
Paulino	0.260		Castillo	0.229
Nady	0.259		Doumit	0.223
Bautista	0.246		McLouth	0.223
Randa	0.241		Duffy	0.218

As you can see, we only had two players above average – Bay and Sanchez. Adam LaRoche had a .300 GPA last year so you can see how valuable he is to the Pirates if he can maintain his production. But that still only makes three players above average on the team.

To give you some context, here are the Brewers top 12 — again, of batters with more than 200 at bats:

Hall, Bill	0.294		Jenkins, Geoff	0.264
Gross, Gabe	0.290		Graffanino, Tony	0.254
Koskie, Corey	0.274		Weeks, Rickie	0.251
Fielder, Prince	0.273		Bell, David	0.251
Cirillo, Jeff	0.270		Mench, Kevin	0.245
Hart, Corey	0.266		Miller, Damian	0.241

They had six players above the .265 average scale, and a seventh just at it.

The Reds had five players above .265, the Cubs had four players above, and the Cardinals and Astros both had five with two above .300 each.

And here are the top 16 from the National League with 200 or more at bats in 2006:

Pujols, Albert	STL	0.361		Atkins, Garrett	COL	0.322
Howard, Ryan	PHI	0.354		Johnson, Nick	WAS	0.318
Berkman, Lance	HOU	0.345		McCann, Brian	ATL	0.318
Scott, Luke	HOU	0.343		Helms, Wes	FLA	0.316
Bonds, Barry	SF	0.336		Holliday, Matt	COL	0.315
Jones, Chipper	ATL	0.334		Bard, Josh	SD	0.313
Cabrera, Miguel	FLA	0.333		Bay, Jason	PIT	0.310
Beltran, Carlos	NYM	0.323		Duncan, Chris	STL	0.310

While the top players closely resembled the OPS leader board last year, when you get below .300 it becomes a lot more obvious who was marginal and who helped their team to put up runs.

It’s not just power you think of with GPA, although it will seem that way. It is the ability of a player to get on base times 1.8, coupled with power. But that’s not all encompassing.

For instance, Chris Duffy looks pretty weak with his .218 GPA above, and he was weak by the GPA standard. But guess who had a higher average in runs scored per game last year – Duffy or Sanchez? Right – Duffy at .548 to Sanchez’s .541.

So GPA has to be looked at with all other available stats to make a final determination on who helped the most, although GPA goes a long way toward being more meaningful to the senses than OPS does, as Schwarz mentioned.

Over the last few days I had the wonderful opportunity to spend some time with a long time NL front office exec. At one point the conversation turned to how well the Pirates did in the second half last year and I stated I thought it was primarily because of better defense behind better pitching. He agreed defense was the primary reason.

But then he told me there was another explanation – one that was just as important because it created a lot more luck for the Pirates than they were able to put together in the first half. It was a reason I hadn’t heard before, and one that took me by surprise.

Bench match ups.

Taken aback, I wanted to know more. He said, do you think it was a coincidence that the team started winning more after Xavier Nady became a Pirate and Chris Duffy came back? I said I thought it was. In 2 minutes he showed me why it might not have been such a coincidence, and I’m going to try and explain those two minutes to you today.

He started off by telling me that as of June 29th last year, the Pirates lead off batter had the 6th best OBP in the National League, and 14th best in baseball. Chris Duffy had left about May 1st and Jose Bautista had been leading off. But the team had only won 27 games of the first 80 they had played.

That was a curious stat, he said, because the winning percentage of the 13 teams with OBP higher than the Pirates was .535, or .519 in the NL only. One reason the Pirates lost so many of those games was because our lead off batter had unproductive OBP.

Unproductive OBP, I muttered?

Sure, he said, the Pirates had the 5th highest number of lead off batter strikeouts in baseball and the 4th fewest number of walks. While the Pirates did had the 13th best lead off batter OPS on June 29th, it was useless OPS because most of the production came with few men on base.

Oh, I said, as a little light went off in my head.

In other words, he declared, the Pirates were spinning their wheels at the top of the order. After the break, that changed and he showed me how.

ESPN has a great statistic they keep that is often over looked – how batters do against certain pitcher types. For instance, if you follow this link and scroll down to the "By Pitcher" area, you’ll see that Jason Bay has a career .913 OPS against power pitcher types. At the very bottom of that same page, ESPN shows you how they define each type.

When Nady was obtained and Duffy came back on to the roster, the exec went on to say, Tracy had the ability to better match up his available players against the type of pitcher the team was facing that day. While this availability didn’t provide more runs scored than runs allowed over the second half, it played a huge role in their ability to win.

Before Nady was obtained, the only other person the Pirates had on their roster that could hit a finesse pitcher was Jose Bautista. When you compare Bautista’s June production when he faced a lot of ground ball/finesse type pitchers, to his August when he faced more power pitchers, you’ll see the difference real quick.. a .639 OPS vs a .910 OPS.

I said I had thought all along that the reason for most of the ups and downs in the Pirates production was because of streakiness of the players. He agreed it was, to some degree. But there is usually a reason for the streakiness, he declared. One of those reasons is feeling comfortable in the box against who you are facing, especially for younger players.

There you go.. one more person’s take on why the Pirates did so well in the second half. So I took his idea one step further – I broke down most of the players on the roster against each of the main four pitcher types – power, finesse, ground ball, and fly ball, and here is the result:

FB	GB	Finesse	Power
Bautista -	Bautista +	Bautista +	Bautista -
McLouth +	McLouth -	McLouth -	McLouth +
Castillo +	Castillo -	Castillo -	Castillo -
Nady +	Nady -	Nady +	Nady -
LaRoche -	LaRoche slg+ rest-	LaRoche -	LaRoche slg+ rest-
Wilson -	Wilson +	Wilson +	Wilson +
Bay slg+ rest+/-	Bay ++	Bay +/-	Bay ++
Doumit slg+ rest-	Doumit -	Doumit +	Doumit BA- rest+
Hernandez -	Hernandez -	Hernandez +	Hernandez -
Duffy +	Duffy -	Duffy -	Duffy +

A plus sign means the player hits that pitcher type better than not, a minus is just the opposite. The "+/-" sign means a neutral ability and you’ll see that some players either hit the ball out of the park, or don’t do anything against some pitchers. That is marked with a "slg+ rest-" sign.

I then put together proposed lineups based on the pitcher type we might face on any given day:

FB	GB	Finesse	Power	Others
Duffy	Duffy	Duffy	Duffy	Duffy
Wilson	Wilson	Wilson	Wilson	Wilson
Sanchez	Sanchez	Sanchez	Sanchez	Sanchez
Bay	Bay	Bay	Bay	LaRoche
LaRoche	LaRoche	Nady	LaRoche	Bay
Paulino	Bautista	Bautista	Paulino	Nady
Castillo	Paulino	Paulino	Castillo	Paulino
McLouth	Castillo	Hernandez	McLouth	Castillo

Notice that LaRoche gets his days off against finesse pitching, with Nady taking 1B and Bautista playing RF. Hernandez also starts at 2B that day. Unfortunately, I wasn’t able to get Castillo out enough, like against power pitchers, because there wasn’t a better replacement for him, offensively or defensively. And Doumit never played a game because he simply lacks a position to play.

I also have Bay batting 4th more often than not. You can be sure Jim Tracy won’t do that but he might consider it.

The end result won’t be that the Pirates score a ton of runs. Instead, this proposal suggests it might create more luck than not, and luck is what propels us each year.

For those that want to create a cheat sheet, I have prepared a list of each of the five NLCD team starters and rated them based on the type pitcher they are. For instance, Jeff Suppan is rated a 9 in finesse. I simply divide 9 by the 411 finesse pitchers in the pool and subtract the result from 1.0 to get the percentile he is in.

In Suppan’s case, that is the 98th percentile and a sure bet Jim Tracy is going to look hard at matching up his players to face him. At about the 60th percentile, you might start seeing less of a concern from Tracy on matching up his players and instead worrying about who is seeing the ball better that day.

Notice I didn’t list fly ball pitchers. That is because so few rated out as a FB pitcher using the five year requirement ESPN uses to qualify them.

<td class="xl24" style="BORDER-LEFT-COLOR: #ece9d8; BORDER-BOTTOM-COLOR: #ece9d8; BORDER-TOP-COLOR: #ece9d8; BACKGROUND-COLOR: transparen

• Posted on February 25, 2007 at 7:48 pm
• Permalink
• 5 Comments
• Filed in: Dailies
• Tags: Soulfood Stats Cafe 2006

CHC	Finse	Pwr	GB		MIL	Finse	Pwr	GB
Zambrano		101	110		Sheets		570
Lilly		90			Capuano		551
Marquis	90		170		Suppan	9		167
Hill		421			Vargas		497
Prior		47			Bush	66		296
Miller		33
Marshall			262		STL
Guzman		207			Carpenter			79
Mateo	159	523	290		Reyes		65
					Wainwright	285	542	205
CIN					Wells	339	161	188
Harang	142	660	332		Franklin	58	391
Arroyo	174				Looper		677	63
Milton

	Runner 1B Only	Other Player’s	Rate
	DP Made	Error Rate	Total
MLB Median 2B	96.3%	0.9%	97.2%
Castillo	95.5%	1.9%	97.4%
difference	-0.8%	1.0%	0.2%
MLB Median SS	54.7%	0.9%	55.6%
Wilson	68.5%	0.0%	68.5%
difference	13.8%	-0.9%	12.9%
	Runner 1B & 2B	Other Player’s	Rate
	DP Made	Error Rate	Total
MLB Median 2B	93.4%	2.5%	95.9%
Castillo	83.3%	8.3%	91.6%
difference	-10.1%	5.8%	-4.3%
MLB Median SS	59.3%	1.2%	60.5%
Wilson	62.5%	0.0%	62.5%
difference	3.2%	-1.2%	2.0%

Castillo showed his weakness when there was a man at 1B and 2B – he ended up with a normalized rate -4.3% below the league average second baseman last year. And that was despite adding back in Wilson’s 8.3% error rate on his plays.

But the value difference between -4.3% and the league average of 0 is just 1/2 an error. That’s right.. 1/2 an error. As it turns out, the only other error made was by Castillo himself.

Jack Wilson, on the other hand, came out smelling like roses turning double plays just as the Stats Geek suggested he might. With a runner at only 1B, he was 12.9% better than a normalized league average shortstop, and 2% better with men on 1B and 2B.

It’s true that Wilson’s INN/DP did drop 19% in 2006 from 2005.. the equivalent of 17 fewer turned. The Geek didn’t tell us why Wilson’s double play rate dropped last year other than to suggest it was Castillo’s fault.

But that simply isn’t true. Here’s a few reasons why:

1) Jack Wilson had a .837 Zone Rating in 2006 versus an .885 Zone Rating in 2005, which was his career high year by a considerable .026 points. That meant Wilson reached 6% fewer balls in his zone last year resulting in 38 more balls in play floating past him to the outfield. Castillo reached 0.8% less.. just 5 fewer than his 2005 season average which is pretty steady gloving year-to-year.

When you consider Wilson turned a DP every 7.4 total chances he had in his zone in 2006, that meant he lost 5 double plays simply because he didn’t stop the ball.

That’s not Castillo’s fault.

2) Wilson committed one error every 63 innings last year versus one every 93 innings in 2005 resulting in 7 additional errors in 2006 when you normalize innings played. Castillo’s error rate was 1 every 69 innings in 2006 versus 1 every 70 innings in 2005. That is also fairly stable work.

But 5 of Wilson’s 18 errors came when he was turning a double play last year versus just 1 in 2005. That’s four more he lost.

That’s not Castillo’s fault either.

While Wilson might have made 17 less double plays in 2006, we’ve identified no less than 9 being his own fault, and that’s more than half. Plus, we haven’t even looked at all the other variables that influenced Wilson’s inability to produce double plays from his 2005 rate.

Yet, consider this — Jack Wilson turned one double play every 12.8 innings last year and, of the 24 qualified shortstops in baseball who played more than 998 innings, Wilson’s INN/DP rate was still 11% above league average [link].

Comparatively, Wilson was 22% above league average turning double plays in 2005. But guess what, if you add the nine we know he missed in 2006 to the 88 he did turn, Wilson’s 2006 DP rate would jump all the way to 19% above league average, just 3% short of his 2005 rate.

But hold on, there’s even more – the league average double play rate for the 24 qualified shortstops in 2006 decreased by 5% from 2005.

No matter how you look at it, Wilson’s DP rate in 2006 was excellent considering his error rate and zone rating. Still, Castillo played steady league average defense both years.

Therefore, the Stats Geek should have been pointing the accusing finger toward Jack Wilson on the left-side of the diamond, not Castillo on the right-side of the diamond, if a finger was raised and pointed any where at all.

Besides, since the Wilson/Castillo team actually produced double plays well above league average rate last year, I’m not so sure I understand why anyone is even pointing a finger at all.

The Stats Geek then said:

"On fielding ground balls, however, Wilson is among the best and Castillo the worst.."

I’ll stop right here because that statement is so overgeneralized it isn’t even worth debating. I love the Geek’s articles, but sometimes. . .

After I looked at Salomon Torres as a potential closer last week [link], I received a lot of email asking me to look harder at Torres in a few areas. Specifically, everyone wanted to know how he had done over the years closing games out as well as when he came into a game starting the inning off.

That made a lot of sense.

I also received a couple of emails from folks in the game suggesting that I look harder at Torres when pitching at home versus away because I had painted Torres as a poor reliever on the road, which none of them believed was true.

This article looks at all three of those suggestions. I used Retrosheet’s wonderful data for the statistics for this article [link].

Let’s begin by looking at Torres when he started an inning off in a game. That chart is below:

Chart Notes: the Earned Runs Allowed column includes runners that were bequest to a reliever and who ultimately scored earned, and the Middle Inn Flag tells us how many times Torres had to be relieved in the middle of an inning.

You can see that the emails I received from those in the game had validity because Torres had a 0.36 / 0.18 split on runs allowed per game in home and away games 2005-2006. That’s huge, just as they suggested.

We can also see that Torres had a 0.06 reduction in runs allowed per game from 2005 to 2006. Just as interesting is the 20% rate Torres was removed in the middle of an inning between 2005-2006.

Now, let’s look at how Torres has done closing out games:

Here we see the same significant split in home and away games between 2005 and 2006. In fact, the rate is nearly three times higher. That is even higher than in the first chart where the split was double.

You will also notice the significant drop in the runs allowed per game between 2005 and 2006 – a 50% decrease primarily because he closed 16% fewer games at home in 2006.

If we compare the year 2006 in both charts, we see Torres allowed 0.25 runs per game in both, so it’s a good bet he’s going to be close to that in 2007 if he closes both home and away games.

With these figures in hand, let’s project how Torres will do closing if he appears in 65 games.. 65 games times 0.25 runs allowed = 16 runs projected for the year. Twelve of the sixteen should be from away games at the 2-1 ratio we found above.

That sounds pretty good until you consider that Mike Gonzalez only gave up 2 earned runs in the 24 games he saved last year (that would be the equivalent of 5.4 runs over 65 games).

As a better example, you might remember cringing everytime you saw Jose Mesa heading toward the mound to close a game in 2004? He only allowed 13 runs over the 65 games he finished that year. But do you remember telling yourself in 2004: if we aren’t winning by more than two runs when Mesa takes the mound, we’ll probably lose?

Many fans did.

Torres has considerably better numbers on the road as we saw above. In 24 games, he gave up just 5 runs last year when finishing a game. But even that rate of runs allowed is 2 1/2 times higher than Gonzalez’s last year.

That’s not good folks. Considering this team figures to play a lot of close games, that’s even more concerning.

No matter how you slice it or dice it, the fans seem to be in for a nail-biting year in tight games with Torres closing for us, especially at home.

There are some variables that could influence how well Torres performs next year, and some of those might be swinging in our favor more so than in previous years — the team could play better defensive behind our closer, we could have more of a lead in the 9th, or we just might find more luck.

Of course, Cam Bonifay was fired as the General Manager in 2001 waiting for those exact same events to occur, and his replacement, Dave Littlefield, has yet to experience those events in his five and one-half year tenure either.

But hey, the team’s slogan is "We Will" so maybe this is the year they finally will? We’ll see. In the meantime, I’m getting my net ready in case I happen to see a pig flying over head.

Happy Valentine’s Day everyone! What could be more appropriate today than to take a hard look at the potential of Salomon Torres to close this year?

The Pittsburgh Pirates traded their star closer Mike Gonzalez to the Atlanta Braves this winter to get Adam LaRoche – the middle of the order power bat they desperately needed. The question now becomes, who closes for the Pirates?

Manager Jim Tracy has already hinted that Salomon Torres will be the closer. This article looks at why Tracy might have made that early decision and who else is on the roster that might be as good, or better, to fill the closer role.

Comparing a closer on one team, to a closer on another team, is like comparing apples to oranges because there are simply too many variables that can influence the final outcome of a game. Park factors, defensive skills of the players behind the closer, age and experience of the closer, umpires, and run support for the closer, are just a few of those variables.

Knowing the limitations of such a comparison, I compared Salomon Torres with the Angels Francisco Rodriguez, who I considered to be the best closer in 2006, Dan Kolb with the Brewers, and Mike Gonzalez, Matt Capps, John Grabow, and Damaso Marte with the Pirates.

I wanted to know two things – how did each pitcher fare under the following critical conditions:

– when entering a game in 2006 that was tied, or there was no more than one run difference either way; and

– when entering a game in 2006 that was tied, or there was no more than two runs difference either way.

It is my belief that when a pitcher with closer blood in his veins enters a meaningful game knowing his performance could be the pivotal factor, they will rise to the occasion whether they are closing or relieving. Simply put, true-blue closer types thrive on the thrill of shutting down the opposing team and their stats will reflect it.

Obviously coming out in the 9th inning with AC/DC blasting on the loudspeakers hypes a closer a lot more than a reliever who enters with two outs in the 8th. But hype doesn’t get results.. pitching gets results. Either you have the ability to shut down batters in critical opportunities, or you don’t. It’s that simple.

Let’s look at the comparison chart. You’ll notice that Torres is listed twice – once showing his full year stats and then I listed just his closing stats starting August 24th when Mike Gonzalez went on the DL. I also posted two years for Kolb – 2006 with the Brewers and 2005 with the Braves.

Notice Rodriguez’s stats in the most critical games when the score was tied or +1 to -1 runs (top table in chart).. he allowed just one run in 30 games while facing 145 batters. That’s just filthy. Mike Gonzalez was right behind him with just two runs allowed in 24 games facing 110 batters. That’s nearly as filthy.

Now look at Torres when closing.. one run allowed in 9 games having faced 50 batters. While that sample size is small, it certainly does trend toward being as filthy as Mike Gonzalez was. One red flag sticks out however.. it took Torres one additional batter per closing opportunity to shut the door than it took Gonzalez. That typically means the closer is either giving up more hits or walks and then getting strikeouts to end the inning, and that is exactly what happened with Torres last September.

Also look at the amount of run support Rodriguez received with the Angels – almost 1/2 run per game. Kolb also received about 1/2 run per game support in 2005 with the Braves in the +2 to -2 runs table. One-half a run per game is a lot of run support for any closer.

Now look at Torres when the Pirates were semi-raking in late August and September – he received less than one-quarter of a run per game in support. Mike Gonzalez, on the other hand, received virtually no run support and now you know why other teams were after him.. he was filthy without support. That’s a plus.

The upper table also clearly shows us why Matt Capps should not be handed the ball to close yet in his young career – he allowed 23 runs in 34 games in the most critical games he entered. But when a little bit of the pressure was taken off him (+2 to -2 runs in the lower table), he excelled, allowing just one run in the 21 games over his +1 to -1 run performances.

Marte and Grabow are pretty much the same animal, with Marte having the edge. Don’t forget that these two were used as specialists from the left side last year so their stats are going to be biased a bit more than the other pitchers. But like I said earlier, results are results and both of these pitchers did an overall better job in critical situations than Torres did last year. In fact, Marte was twice as good as Torres, subject to the bias limitation of being a specialist.

Kolb was limited by the Brewers on when he was handed the ball last year so his results are a bit biased from that as well. But still, he did do an acceptable job in critical games and that is probably why Dave Littlefield is giving him a shot.

Kolb’s success with the Braves in 2005 was fueled by the run support he received in the +2 to -2 games because he was allowing more than one-quarter of a run per game across the board which is two to three times higher than you want to see from your closer.

Jim Tracy is seemingly handing the ball to Torres because of what he accomplished in 23 games from August 24th on last year, but that is just as biased as some of the comparisons here are.

Consider this, half of Torres 12 saves last year were against two deflated teams in September – the Reds and Cubs. Further, in half the games he entered from August 24th on, the Pirates were up by at least two runs. While he did a good job nonetheless, that sample size isn’t enough to override the shockingly high .43 runs allowed per game average he had in critical +1 to -1 games.

Lastly, when you dig into Torres career numbers in +1 to -1 run games since 2003, you’ll find he entered the game 142 times and allowed a total of 93 runs for a .65 average run allowed per game rate. Most relievers get released with numbers like that.

You would also think Torres would fare better at PNC as a ground ball pitcher, but he didn’t. In those same 142 games, 69 were at home were he allowed 50 runs, for a .72 run per game average. That’s off the charts.

Think about that a minute.. 50 runs in 69 games. If he averaged 1 inning per start, that would be a 6.52 ERA and make Shawn Chacon seem like the next Matt Capps.

One thing Jim Tracy better be aware of if he uses Torres is not to bring him in when the Pirates are down by two or three runs in away games because he has historically given up the farm averaging .95 runs allowed per game. If we are ahead by two or three, Torres typically shuts the door better — .31 runs per game.

Nothing in Torres’ resume spells closer material. I can see why Jim Tracy wants to give him a shot based on the few games Torres did do well in, but I have to believe he will be on a very, very short leash.

The obvious choice to close games in Pittsburgh – from the above numbers – is Damaso Marte, then John Grabow.

Pirates General Manager Dave Littlefield has built the 2007 Pittsburgh Pirates around his young pitching staff that ended 2006 with the 15th best ERA in baseball [link].

Every pitcher’s number one goal on the mound is to pitch to contact so they rely heavily on the defense behind them. One way to measure the quality of the team’s defense is their ability to convert balls put in play (BiP) into outs, a measure known as DER – Defensive Efficiency Ratio. Understanding DER is pretty simple.. if 100 balls were put into play on the field and 75% were turned into outs, then the team’s DER would be 75%.

Last year, the Pirates ranked second to last in DER [link]. The biggest defensive deficiency we had, according to the Hardball Times, was when balls were put into play in the air [link]. Out of the three outfield positions, David Pinto’s Probabilistic Model of Range (PMR) suggests that our right fielder’s were the root of our problems [link].

Dave Littlefield has penciled in Xavier Nady to play right field in 2007 and, if you looked at Pinto’s PMR chart above or Nady’s PMR defensive charts [link], you quickly see he is a below average defender in right. Baseball Prospectus confirms Pinto’s work with a 94 career rating in right field for Nady with 100 being league average [link].

The Pirates also had the third worst middle infield combination in baseball last year with Jack Wilson and Jose Castillo as they committed one error every 37 times they touched the ball. That was almost double the league average rate.

And our catcher Ronny Paulino had the third highest number of errors behind the plate, not to mention the eighth highest number of passed balls of all catchers who caught 100 games or more.

So right field isn’t our only question mark heading into 2007 – it’s half the players we are fielding. That’s a genuine concern.

I finally finished my 2007 pitching projections knowing that the fifth starting slot will be taken by Shawn Chacon or Tony Armas, or a platoon combination of the two. For those interested, there is virtually no difference at all between Chacon and Armas when all things are said and done.

Here is my final projection:

The chart assumes Zach Duke, Paul Maholm, and Tom Gorzelanny will pitch a total of 585 innings, and Ian Snell and Shawn Chacon or Tony Armas will contribute 400 innings.

Notice that I am projecting 573 runs allowed over 985 innings for the main five starters. That is a tad higher rate than 2006 overall.

I want you to key on the total number of BiP I am projecting which is 3,081.

That’s important because, in order to determine how much someone like Xavier Nady might hurt us playing right field, we need to know how many opportunities he’ll have.

Remember now, Pinto’s work showed us that Nady converted only 97.4% of all BiP to him that a league average defender in right field made an out on in the same play.

I took all BiP from the six above listed starters from 2002 – 2006 to get a feel for the direction the ball went, and then projected 2007′s fielding opportunities from that model.

Here is the result:

As a quick example, Nady expects to see 19.9% of all balls put in play from our right-hand pitchers, giving him 351 fielding opportunities. But when the southpaws take the mound, the number of opportunities drop nearly one-third to 130 despite the fact three southpaws will be in the rotation. Overall he’ll see about 15.6% of all balls in play for the year.

Knowing the defensive holes we had last year, you can quickly determine where we are exposed again this year – right field, based on Nady’s poor fielding ability, and our middle infielders.

I assumed, of course, that Freddy Sanchez will be playing 3B and you can easily see why he is needed there considering the left side of the infield will field one of every four balls in play next year. Jim Tracy simply can’t afford to move Sanchez from 3B and replace him with Jose Bautista who is defensively challenged at the position [link].

Based on the above two charts and my modeling, I am anticipating a 75 win year for the Pirates if Dave Littlefield does not add another player to the mix during the year, and we get 585 innings pitched from the three southpaws and 400 innings from Snell and Chacon or Armas.

The Pirates start the year off with a tough schedule having to go to Houston where they have historically been mauled. Over the first 45 games, they only have four days off and, in one stretch from April 24 – May 20, they play 26 games with just one day off. By May 20th, each starter will have had nine starts and we’ll know by then whether the team is for real or not.

How have Zach Duke and Paul Maholm done the last two years compared to other rookie pitchers in the game over the last fifteen years or so? Both of these pitchers made their debut in 2005 and will be pitching in their third MLB season in 2007.

To answer that question, I pulled every pitcher out of the Lahman database between 1988 and 2006 who had pitched three consecutive years from, and including, the year they made their debut in. Those 805 pitchers became the main overall group.

I then compared the average production of the main group to Maholm, Duke, and Matt Capps because they were the only three pitchers on our staff who made their debut in 2005. My goal was to see where they were at heading into their third year.

To get a better feel for who was producting and who wasn’t in the overall group, I further broke the main group into the sub-groups listed below:

– those pitchers whose year of debut was between 2000 – 2004;

– those pitchers whose year of debut was between 1988 – 1999;

– those pitchers who accumulated 250 or more innings as a right-hand or left-hand pitcher (tending to group the starters together by the hand they threw with);

– those pitchers who switched teams at least once during the three years;

– those pitchers who remained on the same team for the three years, and then that group was broken down further into whether or not the pitcher had attended college.

For those that love to replicate, I used Tango Tiger’s original, raw FIP calculation. For those unsure what Fielding Independent Pitching (FIP) means, Wikipedia has a nice article that will explain it to you. The rest of the abbreviations are as they would seem.

The results are below:

The findings were fairly balanced across the board with the outlier being the different teams sub-group. That wasn’t surprising considering you would expect better pitchers to be kept by their teams.

What did show up as unusual was that the fewer the strikouts obtained (K/9), the lower the ERA generally went, regardless of other factors.

Perhaps crafty soft tossers are the rule?

Also, pitchers making their debut since 2000 seem to be higher strikeout pitchers than in past years. The result from that was a mixed can of corn however because as the strikeouts rose, so did the home runs allowed per 9 innings and ERA.

It was no surprise that southpaw pitchers fared the best over the last 19 years if for no other reason than batters have a harder time picking up the rotation of the ball out of the pitcher’s hand. The sample size of 58 is small but reasonable enough to feel comfortable with.

Coming in right behind the southpaws were pitchers who had attended college and remained with the same team the full three years. This confirmed the long-standing belief that college pitchers are generally more valuable than high school pitchers early in their career. How much difference there really is has been the subject of many a debate [link - subscription required to Baseball America]; [link - Baseball Prospectus].

Now let’s take a look at the 2005 – 2006 stats for Capps, Duke, and Maholm:

Duke and Capps have both performed better than the league average three-year player we found in the ‘overall’ main group column in the first table. We can’t say they are better yet because they still have to pitch their third year, but it does show us their current trend. On the other hand, Maholm is pitching right at the league average rate.

Curiously, Duke’s K/9 and ERA are both low just like we found in the first table. Duke has a lot of other factors going for him too like a low HR/9 rate and a reasonable BB/9 rate, but I thought that relationship was pretty interesting.

Duke’s .060 FIP would put him in the top 12% of all pitchers I looked at. Maholm’s FIP would put him in the top 60%. However, none of the three pitchers would make the top 20% in ERA.

For those who aren’t aware, Duke was drafted in 2001 and Capps in 2002. They are both high school draft products. Maholm was drafted out of college in 2003.

Next year it will be exciting to watch how Duke and Maholm perform coming out of spring training now that their sophomore years are behind them. If history repeats as they say it does, then both of these pitchers should have a very good third year considering their current production trends – especially Duke.

Here are the three-year results from a few other Pirates:

	debut year	+/- FIP Runs
Randy Tomlin	1990	47.0
Jose Silva	1996	8.3
Esteban Loaiza	1995	-0.1
Steve Cooke	1992	-0.2
Jimmy Anderson	1999	-1.9
Ian Snell	2004	-5.9
Josh Fogg	2001	-14.4