As you already know, I am developing a position on pitching value and rotation usage that supports moving to a more traditional 4-man rotation, using a 5th starter only as a swingman, to handle approximately 16 starts and then work out of the bullpen the remainder of his appearances. I support this rotation pattern because (a) there is a shortage of pitchers that are skilled enough to pitch even 100 innings, (b) the number of pitchers able to reach the 100 IP range is historically consistent, throughout the entire expansion era (except approx. 1977-1982, when relief appearances surged for relief aces, and there were more relievers pitching ridiculous IP totals), and perhaps most importantly, (c) there is no pitcher worth signing and depending upon for 30-36 starts in the off-season market. In this regard, using a swingman rotation would allow the team to maximize an injury reclamation project, or perhaps break any young, promising pitchers into the rotation (eventually; maybe this would only be a relevant issue in 2011, 2012, or 2013 when more notable young pitchers might be available to the Brewers).

One of my growing concerns with this plan is the workload required of Yovani Gallardo. In such a scheme, he would need to start approximately 36 games (or more), which would mean well over 210 IP given his career averages. Given Gallardo's almost-shocking inefficiency, I hesitate to offer full-fledged support of a 4-man/swingman rotation without breaking down his workload.

What is more crucial than pitch count or innings pitched in a true 4-man/swingman oriented rotation is working on that 5th day regularly -- and pitching consecutive 5th day starts for just about the entire season.

Let's cut to the chase; here is how Gallardo fared in one-or-more consecutive 5th day starts in 2009:

 

Dates			IP	R	H+BB+HBP	Pitches
1. April 13-18, 2009			11	7	16	203
2. April 29-May 9, 2009			20	5	22	326
3. June 16-July 1, 2009			26.7	7	34	455
4. July 30-August 29, 2009			41.3	23	62	756
16 GS			99	42	134	1740

 

 Gallardo's pitch efficiency in these stretches of 5th day starts is downright shocking, standing at nearly 17.6 NP/IP over those starts. Even worse, from the stretch of three April-May 5th day starts, to four June-July 5th day starts, to seven July-August 5th day starts, his inefficiency worsened as the season progressed.

Oddly enough, Gallardo's 3.82 runs average, 6.18 IP/GS, and 1.35 WHIP during those 16 games started was not that far off from his numbers in 14-other starts (non-consecutive 5th day starts, and 6th+ day starts); during this grab bag of 14 other starts, Gallardo's efficiency dropped to a cool 16.96, with a runs average of 3.74, 6.19 IP/GS, and 1.33 WHIP.

So, here's the difficulty: how do we gauge Gallardo's alarming inefficiency and decreasing performance as the season wore into the dog days of summer alongside the fact that in his other 14 grab-bag starts (following longer rest periods, or less consistent rest patterns), well, Gallardo was pretty damn inefficient anyway, and has the same alarming workload issues...

I am inclined to suggest two solutions to fitting the problem of Gallardo's inefficiencies into a 4-man rotation:

(1) Simply limit inefficient starts by pulling Gallardo earlier when he doesn't have it. This follows logically from the fact that Gallardo will receive approximately six extra starts in this system, and frankly, with more starts from a better pitcher otherwise, the Brewers can afford to pack it in and call it a day earlier when Gallardo doesn't have it. More starts by an above average pitcher (the best #2 starter in the 2009 NL) will lessen the significance of a few more short outings; should Gallardo be pulled in the 4th-5th when his inefficiency places him around 75 pitches after 4 or near-100 after 5 aq couple of times a year, the extra six starts basically serve as a buffer.

(Nevermind that even in the course of inefficient 5 inning starts, Gallardo can still deliver a performance better than most Brewers starters on a somewhat more regular basis).

(2) Simply build in an extra 2-3 starts during the year for the main swingman, or a second swingman. Or, pitch Gallardo out of the bullpen every now and then, say, to cut his workload during the colder parts of the year, when his performance may decline in outdoor stadia, therefore concentrating his starts into the warmer parts of April and September, and the middle of the season, too. 

Either of these solutions will either (a) provide Gallardo three more starts on five days' rest, rather than four, or (b) an opportunity to maximize potentially long stretches of consecutive starts during times of the season when Gallardo is more likely to be fresh, and then avoiding the possible impact of a late season "meltdown" or regression by building in bullpen time after the middle of September.

This is no doubt radical to think about when you imagine a stretch run without an ace, or something like that, but on the flipside, imagine a stretch of baseball from April 15 to September 15 where Yovani Gallardo makes 36 starts; 

 Here's Casey McGehee as a minor leaguer:

Minors (2836 PA): .279/.332/.409; 75% BIP, 1.9 HR%, .316 BABIP

AAA (1153 PA): .282/.334/.410; 75% BIP, 2.1 HR%, .313 BABIP

2009 (394 PA): .301/.360/.499; 70% BIP, 4.1% HR, .330 BABIP

Two notes: (1) if someone could double check my BABIP math on McGehee's minor league stats, that'd be awesome. I've calculated it about 4-5 different times, and .316, .313 don't look right. (2) If someone could point me to PCL park factors, that'd really be helpful. I can't find 'em.

Now, here's the problem: A 26-year old high-contact hitter (75% minor league BIP rate) comes to the majors and in a neutral offensive environment doubles his HR% and suddenly puts the ball in play 5% less frequently (huge shift!) as a result (he also increased his BB%), while collecting hits on balls in play nearly 2% more frequently (huge shift!) and 4% above average league defensive efficiency.

Now, even without minor league records, that season looks flukey; somehow, a clunker like McGehee was able to beat defenses by 4% on balls in play, while also nearly doubling the league average HR% -- without collecting an overly high K% or BB%. With minor league records we can see that McGehee's HR rates were completely out of the norm in 2009, completely out of nowhere, alongside his BABIP shift (in a neutral offensive environment!) and BIP% rate.

Just think about this for a second: McGehee managed to put the ball in play significantly less frequently while collecting hits on those BIP notably more frequently, all the while doubling his HR% and increasing his BB%. Of course, he did this as a 26-year old rookie, with absolutely no MLB playing time to ramp up to his prime years or anything like that.

Does that scream "I will regress for 2010" to anyone else? (That's why I say trade McGehee now. He's served his purpose, proven Melvin can still pull 'em off the scrap heaps, and now it's time to turn him into a trade piece).

***

Here's what McGehee's corrections look like (I've kept the walks, correcting only the BABIP and HR, and therefore the BIP%):

League average efficiency: 400 PA, 288 BIP; .299 BABIP, 86 BIPH; 8 HR, 32 BB, 70 K, 2 HBP; 366 AB, 94 H, 22 2B, 0 3B, 8 HR; .257/.320/.383

 Note: I hope that this is part of an on-going series on this issue. I specifically plan to survey 1969-mid-1980s next, and then 1987-approx. 1999....

 I think that it is rather clear that run scoring is framed by how the ball is put in play. However, it's not always clear that putting the ball in play works as we expect -- our intuition is that "if you just put the ball in play, you have a chance for something to happen." But, it is not the case that contact -- or putting the ball in play more frequently than average -- is more conducive to scoring runs.

 I worked on this problem in the past (most notably here), and have found that in recent years, it is clearly the case that above average offenses are rarely contact offenses; instead, offenses in which batters regularly strike out, hit home runs, and walk are more frequently above average in terms of runs scored. 

However, what I want to learn is how this changes over time; in the last two months, I began looking into contact rates from 1962-1968 -- baseball's second deadball era -- and I was surprised at what I found. At first glance, even in a severely depressed offensive environment, I found that offenses put the ball in play less frequently were more likely to be above average. 

In the decade, there were 53 "high" contact offenses -- offenses that put the ball in plat at an above average rate (28 NL, 25 AL). Of those offenses, 14 were above average in the NL, 10 in the AL, meaning that 24 of the 53 high contact offenses were above average. By contrast, there were 87 "low" contact offenses (42 NL, 45 AL), and of those offenses that put the ball in play less frequently than average, 50 were above average (26 NL, 24 AL).

Overall, even in a depressed era, there were not even as many contact offenses -- teams still were playing towards the 1940s and 1950s norms, to some extent, and the "contact" norms we might expect in such an era didn't actually take over until the 1970s. 

Another lesson in the exploration of balls in play: it is not always what we intuit about the game that corresponds to run scoring, and it is not the case that contact necessarily -- or even regularly -- leads to more runs scored.

What this means (I worked on the NL today, the AL comes tomorrow or Friday):

CONTACT OFFENSES

Of the above average offenses from 1962-1968 in the NL, only five teams (a) put the ball in play more frequently than average and (b) were below average in BB%, HR%, and K%. These teams are marked by unrealistic BABIP in some cases, or exceptional situational hitting:

1. 1962 Cardinals (.296 BABIP, .270 RISP, .275 Men on)

2. 1964 Cardinals (.308 BABIP, .266 RISP, .275 Men on)

3. 1967 Pirates (.314 BABIP, .265 RISP, .280 Men on)

4. 1968 Cardinals (.283 BABIP, .250 RISP, .255 Men on) '

5. 1968 Pirates (.289 BABIP, .244 RISP, .248 Men on)

It is rare for truly contact offenses -- offenses that do not walk, do not strike out, and do not hit home runs -- to score runs at an above average rate.

Notable offenses:

These offenses matched at least two of three HR%, BB%, K%:

1. 1963 Braves (2.2 HR%, 8.5 BB%)

2. 1963 Phillies (2.1 HR%, 15.7 K%)

3. 1964 Braves (2.6 HR%, 7.8 BB%)

4. 1964 Pirates (2 HR%, 15.9 K%)

5. 1964 Giants (8.2 BB%, 2.7 HR%)

6. 1965 Giants (2.6 HR%, 7.8 BB%)

7. 1965 Braves (3.2 HR%, 16.1 K%)

8. 1966 Braves (3.3 HR%, 8.2 BB%)

9. 1966 Pirates (2.5 HR%, 16.2 K%)

10. 1966 Phillies (8.2 BB%, 15.5 K%)

11. 1967 Cubs (2.1 HR%, 8.3 BB%)

Only two of these eight offenses managed to put the ball in play more frequently than average (1964 Braves, 1965 Giants).

TRUE OUTCOME OFFENSES:

These are the offenses that matched or surpassed the averages in HR%, BB%, and K%. I posted their top attribute:

1. 1962 Braves(9.4 BB%)

2. 1963 Reds (7.9 BB%)

3. 1964 Reds (15.8 K%)

4, 1965 Reds (2.9 HR%)

5. 1965 Phillies (17.7 HR%)

6. 1967 Giants (2.3 HR%)

Holy *** this year was crazy! So, fewer pitchers managed to reach 100 IP in the 2009 NL -- seven, to be exact -- and fewer runs were scored / allowed than in 2008. So, the value of a lot of pitchers entirely shifted around. You're going to be absolutely surprised at parts of this list. 

Also, the ace drop-out rate from 2008 was insane!

2008 First Starters

1. Lincecum, SF (72.5 VORP): 34 G, 33 GS,            227 IP, 72 R

2. Santana, NYM (73.4 VORP): 34 GS,                    234.3 IP, 74 R

3. Sabathia, Mil (52.2 VORP): 17 GS,                       130.7 IP, 31 R

4. Dempster, ChC (57.5 VORP): 33 GS,                   206.7 IP, 75 R

5. Peavy, SD (51.5 VORP): 27 GS,                           173.7 IP, 57 R

6. Hamels, Phi (56.3 VORP): 33 GS,                         227.3 IP, 89 R

7. Sheets, Mil (52.4 VORP): 31 GS,                          198.3 IP, 74 R

8. Billingsley, LAD (51.6 VORP): 35 G, 32 GS,       200.7 IP, 76 R

9. Haren, Ari (53.2 VORP): 33 GS,                           216 IP, 86 R

10. Lowe, LAD (49.7 VORP): 34 GS,                       211 IP, 84 R

11. Webb, Ari (50.8 VORP): 34 GS,                          226.7 IP, 95 R

12. Volquez, Cin (44.3 VORP): 33 G, 32 GS,           196 IP, 82 R

13. Hudson, Atl (37.0 VORP): 23 G, 22 GS,            142 IP, 53 R

14. Nolasco, Fla (44.7 VORP): 34 G, 32 GS,            212.3 IP, 88 R

15. Oswalt, Hou (44.1 VORP): 32 GS,                      208.7 IP, 89 R

16. Maholm, Pit (42.7 VORP): 31 GS,                       206.3 IP, 89 R

 

Sheets, Webb, Volquez, Peavy, and Hudson were injured, so that's five right there, Sabathia and eventually Peavy went to the AL, Dempster's season was a fluke (who knew!), and overall, Santana, Dempster, Hamels, Billingsley, Lowe, Nolasco, Oswalt, and Maholm regressed.

The only remaining #1 pitchers from 2008 that pitched as #1 in 2009 were Tim Lincecum and Dan Haren!

This season was wide open!

As always, I ranked pitchers by runs prevented, and then posted VORP, R, and IP, along with G and GS). The logic is simple: it is the main job of a pitcher to prevent runs. The value of the pitcher is then enhanced by IP, and how they pitch in their environment (this is why VORP is important). But, for a ranking tool, there is none better than runs prevented. It's quick, it's sharp, it's dirty -- as Aristotle said (sort of), cut at the joints!

Also also, I had no reason to move that many swingmen -- pitchers that pitched 100 IP, but did not enter half of their games as a starter. There was one swingman in 2009 NL, Jeff Karstens.  If you think his rating was unduly affected by my methods, please feel free to contact the Jeff Karstens Fan Club!

 Enjoy!

I visited this point in a series preview earlier this year, and now that the season is complete and Kershaw is making a playoff start, I think it's time to revisit it.

While Kershaw hasn't yet had the type of seasons that Koufax had late in his career, check out their early careers -- it took Sandy Koufax eight seasons in the majors, until he was age 26, to match Kershaw's 170+ IP and 140+ ERA+ campaign.

Koufax through age 21:

62 G, 28 GS, 204.7 IP, 182 K/108 BB, 1.42 WHIP, 103 ERA+

Kershaw through age 21:

53 G, 51 GS, 278.7 IP, 285 K/143 BB, 1.33 WHIP, 125 ERA+

Note that it took Koufax through his age 22 and into his age 23 campaign to reach 51 GS; beyond that, Koufax didn't have a value season close to Kershaw's 2009 campaign until age 25:

Koufax VORP:

age 19: 11.4 VORP

age 20: -1.4 VORP

age 21: 13.8 VORP

age 22: 7.6 VORP

age 23: 20.5 VORP

age 24: 19.2 VORP

age 25: 45.2 VORP

age 26: 49.9 VORP

age 27: 87.9 VORP

age 28: 69.4 VORP

age 29: 86.7 VORP

age 30: 99.7 VORP

Kershaw VORP:

age 20: 16.3 VORP

age 21: 49.1 VORP

Of course, the big difference is that we know that Koufax pitched some of the most dominant, elite campaigns ever, from age 27 through age 30. That's the test for Kershaw -- obviously we cannot compare him to Koufax in that regard, but thus far, Koufax is no comparison for Kershaw through age 21.

Where does Kershaw go from here? It should be fun to watch!