What does the Brewers offense
need to do to improve? I considered this very question some time ago,
arranging the issue in a series of questions
about the Brewers 6 major core players between the ages of 24 and 30.
That post contains a good amount of year-to-year data, which is nice,
but it's lacking something:
Context.
It seems to me that with the loss of Sabathia and Sheets, various rotation projections
could put the Crew anywhere from 30-50 runs behind the 2008 pitching
staff; maybe less, maybe more. It is also my opinion that even with
Sabathia and Sheets, the rotation likely does not improve upon their
2008 marks, and the end result is the same: we're looking for extra
runs from the offense.
While some might mourn losing the Brewers' healthy 750 RS, 689 RA
runs differential, we can learn from utilizing a Pythagorean Record
projection that replacing 50 RS in either direction results in a rather
minimal difference in expected wins; whereas the Brewers' 2008 expected
winning percentage was .5385 -- based solely on runs scored and runs
allowed -- a 2009 exoected winning percentage with a runs differential
closer to 800 RS, 739 RA yields an expected winning percentage of
.53602. In terms of 162 games, those winning percentages display the
difference between an 87.24 win season and an 86.84 win season.
So, my question in the last few weeks was simple: how do the Brewers score 50 more runs?
My initial thinking about this issue centered around the fact our
entire young core -- minus Hardy -- regressed in 2008 from their 2007
performances, sometimes in large quantities (this is especially true
for Fielder and Braun in one sense -- they couldn't possibly repeat the
luck of their 2007 campaigns, and it is also true for Hart, Hall, and
Weeks in a different sense -- Weeks' late-season explosion was probably
unlikely to be sustained for a longer period of time, and Hart and Hall
maintained habits that were not conducive for long-term hitting
success).
The result of this line of thinking had me searching for "grand"
outcomes. Who could increase their OBP from a .340 range into a .365
range? Or get that slugging up another 3% or 4%...
The only trouble is, in terms of baseball statistics, a 3% or 4%
difference in one category is rather large. This is true for individual
and team stats; for instance, the fantastic Cubs (855 RS) and awful
Padres (637 RS) were separated by a 3.7% difference in the OBP category
(.354 vs. .317). We probably feel the same way about Chase Utley vs.
Rickie Weeks, Matt Holliday vs. Carlos Beltran, Carlos Beltran vs. Mike
Cameron -- you get the picture. Each of these players have an OBP
differential somewhere between 3% and 4%, and that's just my eyeballing
stats and picking players off the top of my head.
So, I started to think about a way to build a more realistic, level,
and less-optimistic projection of the offense that still maintained the
premise of improving runs production. Luckily, I happened to be reading
through The New Bill James Historical Baseball Abstract (New York: Free Press, 2001) for a second time; a particular comment of his, about runs created, spurred me forward:
"A hitter's job is not to compile a high batting average. A hitter's
job is not to maintain a high on-base percentage, not to create a high
slugging percentage, not to get 2000 hits. A hitter's job is not to hit
home runs. Some hitters might hit home runs as the primary part
of their job, but only some hitters, and even those only some of the
time..." (James, 2001; 329).
James is correct:
"The objective of each team is to win by scoring more runs than the opponent" (MLB Official Rules, "Objectives of the Game," 1.02).
I could easily sit here and cite to you numerous shifts in BABIP
(Batting Average on Balls in Play) for the Brewers' youngsters, coupled
with their shifts in Balls in Play %, and the impact those luck factors
will have on OBP and SLG and AVG for players such as Braun, Fielder,
Hardy, Hart, and Weeks. I can explain to you why I think Braun and
Hardy are most likely to regress in 2009, Hart and Hall most likely to
improve, and Prince Fielder most likely to explode in 2009.
But you could still ask, "how do the Brewers score 50 more runs?"
So, I came to runs created. Runs created is a series of various formulas created and revised over time, initially conceived by James, in order to project the runs
production of a particular player. Which makes sense; we can easily and
readily cite a player's AVG, how many HR they hit, and their RBI; but
we'd be lost for words if we were asked about their runs production.
All you need to know about runs created is that it basically
conceives scoring runs in terms of producing runners (or getting
runners onto the basepaths, and keeping them there) and advancing
runners (once they are on the basepaths). Producing runners and
advancing runners occurs in concrete opportunities to produce and
advance runners -- basically, this concept allows us to analyze H, BB,
HBP, CS, SB, SH, SF, TB, and AB in a dfferent way, with a different
objective in mind: projecting runs production.
There are multiple runs created formulas, and it's not necessary to
know them all here. I used the "technical" version for this survey,
simply because it was readily available on Baseball-Reference
player pages, and I could easily correct the data for my survey by
calculating the runs created of the 2008 Brewers as a whole (using team
stats), calculating runs created of the 2008 Brewers as an aggregate
(adding different Brewers together), and comparing that data with the
actual runs scored of the 2008 Brewers.
This is a wonderous fact about the 2008 'Crew -- we know how many
runs they scored, exactly (750), and we can thus correct the runs
created projection. In his 2001 Abstract, James notes that most
teams in history fall within 5% of runs created projections, with
certain teams deviating for various reasons. The Brewers' deviated more
than 5% of their RC projections in 2008, probably because of the dearth
of situational hitting coupled with the overall prevalence of extra
base hits. If there is a discrepency between extra base hits and
situational hitting, that will show up in a formula such as RC, given
that Total Bases (TB) is utilized as the basis of advancing runners
(along with SB, SH, SF, and BB) -- if a team posts a high TB total due
to a high number of extra-base hits, but continually fails to execute a
single or a sacrifice fly in situations with a high-probability for
scoring actual runs, that difference will eventually be recorded in
Runs Created if it occurs frequently.
So, here's one calculation -- the 2008 Brewers' starters, bench, and
pitchers (while batting) created 813 runs, which is .92251 of the 2008
Brewers' actual runs scored 750. One of the methods that James
suggested that spurred me forward is the following correction of runs
created, which suddenly made the formula incredibly useful for the
specific projection of projecting an improvement in 2009 for the
Brewers:
"But isn't it possible, someone wonders, that on this team that has
700 runs created, and only 680 runs scored, some players actually
created MORE runs than our estimates show? Of course it's possible; in
fact, it's inevitable. We're not making the estimates perfect; we're only making them better.
If a team scores 680 runs but has 700 runs created, we know that the
average player on this team is 2.9% less productive than our best
estimate shows. If we reduce all the estimates by 2.9%, we will be
making some of them more accurate and some of them less accurate. What
we can be absolutely certain of is that those we are making better
out-number those that we are making worse" (James, 2001, 331).
Now that we have a method for projecting runs created, and a method
for correcting those estimates and projections, we can ask a question
about the little things: what if the 2009 Brewers improve because
no member of the offense explodes, produces a career season, or
produces at an outstanding level; what if the 2009 Brewers improve
because of minute changes here and there in their approaches, their
games, and their luck?
I want to analye the following issue: with a few hits here, a few
less caught stealing there, a couple more sacrifice flies thrown in,
perhaps a couple of walks, maybe a few more extra base hits...how much
can the Brewers' offense improve?
Which is to say, how many runs will they create in such a scenario?
Using the "technical" version of runs created, which includes SB and
CS stats in estimating a batter's success in staying on base and
advancing runners, as well as SF and SH in the advancing runners
section of the equation. The version I used, as well as
Baseball-Reference (as far as they stated and as far as I know), goes
something like:
( "Producing Baserunners / Keeping Baserunners on base" (H+BB+HBP-CS-GIDP) x
"Advancing Baserunners" (TB + (.26 x (BB - IBB + HBP)) + (.52 x (SH + SF + SB))) ) /
"Opportunities" (AB + BB + HBP + SH + SF)
All together:
( (H+BB+HBP-CS-GIDP) x (TB + (.26 x (BB - IBB + HBP)) + (.52 x (SH + SF + SB))) ) /
(AB + BB + HBP + SH + SF)
According to this metric, the 2008 Brewers as a unit created
approximately 795 runs, which is approximately 5.7% runs above the
Brewers' actual runs scored. I was mistaken in Part One, when I stated
that the Brewers' 2008 runs created was almost 8% above the actual runs
scored; the estimate of runs created I used to inform that statement
featured an aggregate of the individual 2008 Brewers' runs
created projections with the Brewers. That RC total was approximately
813 runs, which is approximately 2.7% above the 2008 Brewers' RC using team stats; how is this possible you ask?
Quite simply, the Brewers as an aggregate of individual production are different from the Brewers as an actual unit, a
unit that plays through situations, and produces in sequence with one
another. There are ways in which we can intertwine the individual
aggregate projections in order to bring them back to the 795 runs
created projection, and closer to the actual 750 RS. Those methods
were, quite frankly, more difficult and in-depth than I wanted to
employ here; I wanted this to be a somewhat easy survey to follow!
So, for the sake of convenience, I used the Brewers' individual runs created projections from Baseball-Reference,
simply because of their availability. You should follow along via B-R
player pages; it's rather easy to do -- runs created, as well as other
stats, are listed under " Special Batting" sections on each player
page. In order to bring the individual RC projections closer to the
actual RS by the Brewers, I divided the Brewers' 2008 RS (750) by the
2008 aggregate RC (813), and then multiplied each player's individual
RC by that percentage (which is .92251; meaning, each RC projection is
approximately 92% of that player's actual runs production in the
Brewers' team environment.
I am aware that there are more accurate methods to rectify this
issue, but as I stated above, I wanted to keep this rather simple, and
I wanted to use methods I am more familiar and comfortable with.
Here's the fun stuff:
2008 Brewers' starters (Braun, Cameron, Fielder, Hall, Hardy, Hart,
Kendall, and Weeks): 649 aggregate RC in 4746 PA; corrected to
approximately 598.71 RS.
| Player |
|
RC |
|
Correction |
|
PA |
|
|
|
|
|
|
|
|
|
| Kendall |
|
59 |
|
54.43 |
|
587 |
|
| Fielder |
|
114 |
|
105.17 |
|
694 |
|
| Weeks |
|
72 |
|
66.42 |
|
560 |
|
| Hall |
|
47 |
|
43.36 |
|
448 |
|
| Hardy |
|
90 |
|
83.02 |
|
629 |
|
| Braun |
|
111 |
|
102.4 |
|
663 |
|
| Cameron |
|
75 |
|
69.19 |
|
508 |
|
| Hart |
|
81 |
|
74.72 |
|
657 |
|
|
|
|
|
|
|
|
|
| Starters |
|
649 |
|
598.71 |
|
4746 |
|
2008 Brewers' main bench (Branyan, Counsell, Dillon, Durham, Gross,
Gwynn, Kapler, and Rivera): 144 aggregate RC in 1083 PA; corrected to
approximately 132.48 RS.
2008 Brewers' auxiliary bench and pitchers: 20 aggregate RC; corrected to approximately 18.45 RC.
The starters and main bench account for the vast majority of the
Brewers' production in 2008, of course; the starters' 4746 PA accounted
for just over 75% of the team's total PA, and their approximate 598
runs scored / adjusted runs created account for almost 80% of the
Brewers' total run production in '08. The bench's 1083 PA account for
over 17% of the team's total PA, and their 132 runs scored / adjusted
runs created account for just under 18% of the team's runs scored.
***
Now, an exercise involving the starters, and an exercise showcasing
some of the value of Runs Created, as it concerns "small" or
"background" stats that nonetheless affect advancement of runners and
ultimately, creating runs...here are eight "little" things for our
starters, which alone account for almost one win above the Brewers' '08
Pythagorean W-L (which was approximately 87 wins):
“True” Little Things (-8.80 runs created) 758 RS / 689 (174179.86686) / (320585.90346) = .54332 = 88.02 = (almost 1 win)
Corey Hart: Increase in GDP (94.33 PA/GDP in 2007; 38.65 PA/GDP in 2008) due to increase in GO/AO. Difference: Approximately -4.00 runs created.
Bill Hall: Continued low SB% (.444 SB% in 2007; .454 SB% in 2008) accounts for wasted outs. Difference: approximately -2.00 runs created (w/o any SB attempts whatsoever)
Rickie Weeks: Decrease in SB attempts (18.74 PA/SB attempt in 2007;
23.33 PA/SB attempt in 2008), leading to decrease in SB and SB%. Difference: Approximately -2.00 runs created (vs. 2007 SB)
J.J. Hardy: Increase in GIDP (49.08 PA/GDP in 2007; 34.94 PA/GDP in 2008). Difference: Approximately -2.00 runs created.
Prince Fielder: Increase in SF (170.25 PA/SF in 2007; 69.4 PA/SF in 2008). Difference: Approximately +1.00 run created.
Ryan Braun: Decrease in SB attempts (24.6 PA/SB attempt in 2007;
36.83 PA/SB attempt in 2008), failing to capitalize on increased SB%. Difference: Approximately -0.50 runs created.
Mike Cameron: Increase in SB attempt (28.3 PA/SB attempt in 2007; 23.09 PA/SB attempt in 2008). Difference: approximately +0.50 runs created
Jason
Kendall: Increase in SB attempt (73.47 PA/SB attempt in 2007; 53.36
PA/SB attempt in 2008), leading to more runs created with increased
SB%. Difference: Less than +0.20 runs created.
Notice how the little things add up -- without
considering issues with extra base hits, luck with BABIP and
balls-in-play %, K% or BB%, without concerning any major hitting
issues, we've accounted for almost one win's worth of production in
stolen base attempts, sacrifice flies, and double play ground balls.
And that's just looking at one aspect of the player's game that
changed; if we scrutinized the record, we might gain a run or so back,
but we might also find more "little thing" blindspots, too...
This is just a start to answering my initial question, which was,
How do we account for 50 runs scored for 2009? We've accounted for
almost 9 runs thus far....now it's time to move forward....
Now that we understand the power of better baserunning, sacrifice
flies, and double play groundouts, I think that we can move on to other
runs created exercises, with other areas of production, with hopes of
learning the impact of production increases for each player in the
Brewers' starting eight.
Keeping with the theme of "little things," I purposefully kept these
increases small, with the hopes of drawing further conclusions about
runs created (and using runs created) with small increases, realistic
increases -- usually within 1% in OBP and SLG categories, and never
more than 2%.
To keep things interesting, I eye-balled the statistical lines of
2008, adding and subtracting just a few elements here or there,
according to conclusions or general guidelines I put together in the
original "Six Questions" appraisal of the offense. Thus, I focused on
Hart's HR drop in 2008, Braun's hitting drop, Fielder's 2B drop, Weeks'
hitting drop, and Hall's discipline decline (i.e., lower K/BB
discipline without the hitting results of previous seasons).
Hardy was an interesting case -- I could not, with an honest mind,
project him for any type of improvement, even small, given his luck
increase in 2008 (he put fewer balls in play and a higher percentage of
those balls in play dropped for hits). So, I turned Hardy into my
"counter" case -- I purposefully set him up for a rather large
regression, in hopes of seeing how the smaller increases from a few
players stacked up against a larger regression from one player.
Outside of the young core, I did a few simple things with Kendall.
Cameron was a tough case, given the anticipated playing tiem
discrepency between '08 and '09. So, I put together three models: (1) I
used his career averages to build a startling regression season; (2) I
used his career averages to build a realistic full seaon; (3) I used
his 2004 campaign, which was the last full season in which Cameron
played in an offensive environment that was near the offensive
environment of Miller Park (for this I used Baseball-Reference's AIR
index, which is rather intuitive and easy to read and digest).
Here are the results...enjoy!
(1)
Braun (663 PA): .295/.341/.568 (.909) = 107 RC (+.006 OBP, +.015 SLG).
(Braun: +6 H, +3 2B, -2 BB, +2 SB, +2 SF)
(2)
Fielder (694 PA): .282/.378/.520 (.898) = 110 RC (+.006 OBP, +.013 SLG).
(Fielder: +4 H, +4 2B)
(3)
Hall (448 PA): .246/.308/.394 (.702) = 47 RC (+.015 OBP, -.006 SLG)
(Hall: +9 H, +5 2B, -1 3B, -4 HR, -2 BB, -3 SB, -5 CS (-8 SB attempt), -1 SH, +1 SF)
(4)
Hardy (629 PA): .273/.326/.447 (.773) = 76 RC (-.017 OBP, -.031 SLG)
(Hardy: -5 H, -2 HR, -4 BB, -2 SH, +4 SF, -3 GDP)
(5)
Weeks (560 PA): .239/.348/.412 (.760) = 70 RC (+.006 OBP, +.014 SLG)
(Weeks: +2 H, +1 2B, +1 HR, +4 SB, +1 CS, +2 BB)
(6)
Hart (657 PA): .268/.298/.479 (.777) = 79 RC (-.002 OBP, +.020 SLG)
(Hart: +4 HR, +2 SB, -2 CS, -4 GDP)
(7)
Kendall (587 PA): .254/.337/.333 (.670) = 58 RC (+.010 OBP, +.009 SLG)
(Kendall: +2 H, -6 SB, -3 CS (-9 SB attempts), +6 BB, +3 SH)
(8)
Cameron (a) (568 PA): .231/.320/.389 (.709) = 60 RC; (b) (568 PA):
.255/.342/.443 (.785) = 72 RC; (c) (562 PA): .231/.319/.479 = 73 RC
(a) -.011 OBP, -.088 SLG; (b) +.011 OBP, -.034 SLG; (c) -.012 OBP, +.002 SLG
(Cameron career averages w/ adjustmnets for hitting,
power, and speed regression; Cameron career averages; Cameron 2004; his
last full season in a favorable offensive park)
The results of this survey: from the lowest Cameron estimates for
playing time and runs created, to the highest estimates, the Brewers
starting eight can be projected to perform anywhere from 4800 to 4806
PA, creating 607 Runs to 620 Runs.
Adjusted for playing time, the lowest runs created projection -- 607
runs -- is pretty much the same offensive output as the 2008 Brewers'
starting eight. 620 runs created is basically a runs created increase
of 13 runs, which is significant in this model because (a) not all of
the gains from the "True Little Things" model (see Part Two) were
incorporated, leaving the chance at a few more "here and there" runs
created in the shadows behind the main stats; (b) these models include
at least one significant regression (Hardy), and standard, average and
below average projections for Cameron.
With slight improvements allotted to either Hardy or Cameron, from
their 2008 level of production, the aggregate runs created jumps
anywhere between 622 to 630 RC; if both Hardy or Cameron improve, that
means an aggregate of approximately 633 runs created from the core
(these improvements are anywhere between 15 runs and 26 runs above the
2008 core eight).
Remember, all of these improvements occur without assuming any
improvement above 2% in either OBP or SLG. In fact, many of the
projected improvements were easily between .005 and .015, and hardly
any of the projected improvements bumped against the .020 level (Hart's
HR correction is the only one that hits a 2% increase).
Marked improvement can occur without break-out seasons, and it can
also occur even with a case of one player regressing significantly.
Without assuming grand, optimistic positions, we have found scenarios
in which the Brewers reach higher levels of runs production from their
starting eight. This should favorably shape our opinion regarding the
chances of improved runs production, and furthermore, should make our
next consideration -- that of break-out seasons -- even more
interesting now that we know that the team is not doomed if any of the
starters fail to show radical improvement over 2008.