A Whole New Ballgame, part 4
June 13, 2005
Well, it's been almost a week since the draft and about the only player
who is still on my team is Alex Rodriguez. Actually there are a
few more players who've managed to survive the roster churn, but it's
still
a very different looking team than the one I envisioned before the
draft. Why?
1) Unlike fantasy baseball, injuries are devastating and in part
predictable in Strat-o-matic. It's reasonable to take a chance
on a player who's had some tough breaks with injuries in fantasy
baseball because there's a decent chance he could stay healthy for a
full season. So for the price of three or four months of expected
production, one might get six. In addition, an injury-prone
player will often exceed his expected rate of production in a year in
which he stays healthy. That's not the case in Strat.
If a player missed 30-40 games in real life, that's about how many he
will miss in the Strat simulation and his production rates will remain
similar as well. Also, unlike real baseball, there's no way to
take precautions to keep a player healthy. For example, DHing a
player has no
appreciable effect on keeping him off the DL. So even though
Joe Mauer was one of the 5 most productive hitters by rate against
right-handers, the fact that he only played 35 games last year due to
injuries meant that he wasn't going to play very many games even if he
didn't play catcher, which is how he got injured in the first
place. Having him for 100-120 games would be sufficient enough
reason to carry his salary and injury risk, but the likelihood of him
playing that many is quite remote. A more reasonable expectation
would be something like 60-80 games. So one would need to carry
at
least a decent back-up plan for the other 100 or so games, the cost of
which
would be almost as much as Mauer himself. Plus, the back-up
player would have to be hardy (an injury rating of 1 or 2 on a scale of
1-6) or else the same issue could arise with
him, too. With injuries in Strat-o-matic, the possibility of a
never-ending cycle dragging a team to the cellar is all too
real. The best plan appears to be taking injury-free
starting players
and grabbing some very cheap back-ups in case they get hurt for a
couple of
games.
2) Starting pitching - and the bullpen to a lesser extent - is another
conundrum. It's next to impossible to win with cheap starting
pitching. In fantasy baseball, it's possible to hedge based on
indicators from last year's performance. Pitchers like John
Patterson and Erik Bedard are fine examples. Neither guy had a
good ERA or WHIP last season. But there were several indicators
like strikeout rate that pointed toward a breakout season this
year.
The problem in Strat is that cheap pitchers are cheap for a reason -
they weren't very good. That doesn't mean the real player doesn't
have
talent. But for that season and for whatever reason - poor
defense, injuries, bad luck - he wasn't good. The difficulty
is finding which ones were simply bad and which ones were victimized by
factors that were beyond their control. That is no easy task when
looking at a card that defines 72 different possible outcomes from each
plate appearance and the probabilities that each will occur. For
example, Mark Mulder is a groundball pitcher (2.05 g/f ratio last year)
and pitched in front of a good defense last year. A friend of
mine ran a simulated season with him pitching in front of an even
better
defense in a 4-team division in which two of the ballparks were
extremely favorable to pitchers. The results? He gave up 55
homers on his way to an 11-28 record and a 6.68 ERA. I only have
suspicions as to why - injuries, maybe the bullpen wasn't properly
configured or perhaps the computer manager replaced quality defenders
with ones who had been
designated as strictly DHs or pinch hitters... there are any
number of possible explanations. Or he simply could have been a
really bad pitcher last year (mostly in the second half) and was simply
bailed out by a good A's defense and bullpen. The point is that a
computer
will be running
the simulation and that if Mulder is redeemable, he will require more
careful handling than the
computer manager appears capable of.
3) Speaking of the computer manager, HAL as he is affectionately known
in Strat circles, is not known for being particularly insightful or
intuitive when it comes to managing the roster during the game.
In essence, HAL is Dusty Baker. He doesn't look at the individual
splits: if the opposition brings in a lefty, he brings in a
righty. He'll leave in his lefty specialist after he's gotten the
lefties out because "he's got a hot hand" only to be crushed by the
impending right-handers. Even with fairly specific instructions
designated in the manager strategy settings, HAL tends to blunder
against his programmer's wishes. The best strategy is to have
generally good players that don't require a lot of manipulation to be
successful. Those players are generally more expensive, making it
harder to build a resilient and competent staff.
So there are two factors that work against filling out a complete
pitching staff - the unreliability of the
computer manager and the difficulty of getting a handle on what the
player card actually represents.
Fortunately there are hobbyists who spend a good amount of time
distilling the latter info for public consumption. One such
individual
produced a database of the probabilities for each event for each
pitcher card. With this info in hand, I decided to try an
experiment. Since my original plan to go with cheap starters like
Mulder, Barry Zito and Matt Morris was destined for the dumper, I
thought I might try use the database to see if there really was way to
discover cheap quality starting pitching. I won't know
whether I'm successful or not until the season is over, but I've
already
chalked this season up as a learning experience, so what the hey.
Anyway, I multiplied the
expected
run production for each event - .33 for a walk, .78 for a double,
1.40 for a home run, etc. - with the corresponding probability for each
on each pitcher's card. Excel rocks, by the way. Then I
incorporated the pitcher's probability of
getting a strikeout. In theory, the result should rank the
pitchers by the strength of their player card. I was fairly
pleased with the results as it listed Randy Johnson, Jason Schmidt,
Carlos Zambrano, Roger Clemens, Jaret Wright, Jake Peavy, Johan
Santana, AJ Burnett, Oliver Perez and Rich Harden as the best cards
available. Other highly ranked cards but perhaps somewhat of a
surprise were Doug Davis, Bobby Madritsch and Kip Wells.
Unfortunately, my team doesn't have any of those pitchers. But
Roy
Halladay, Matt Clement and Jeremy Bonderman were ranked reasonably
highly
so maybe I'm not in as bad a shape as I thought. I went down the
list and found a few pitchers who were relatively highly ranked for
their price tag..
I think Erik Bedard's strikeout rate might have been overvalued on the
table. Bedard had solid defense at short, second and center
backing him up in Baltimore so I'm not sure why his card was so strong
on the
table. Maybe the less than average defense in the Oriole corners
was the reason his numbers were so high last year. Whatever the
reason, I will find out if the defensive upgrade and larger ballpark
will be enough to turn him around.
Wil Ledezma's card was surprisingly strong. It had no home run
possibility and only 1 triple which was split 25%/75% with a
single.
That seems like a pretty incredible card. His strikeout rate
wasn't very good, but if the lack of extra base hits on his card is for
real, team defense should be the most important factor in his
success. However, against right-handers - with whom HAL will
automatically oppose him because he's a lefty - get lots of singles and
walks against him, meaning he would likely be nickle-and-dimed to death
each time out.
So instead I opted for Joel Piniero. Seattle had excellent
defense at second base, center and right field. Occasionally,
left field got an upgrade when Hiram Bocachica played. But the
rest of the defense was pretty poor. I considered Mike Mussina
for the same reasons - poor defense everywhere except for short and
third - and whose numbers were very similar, but Piniero had the slight
advantage of a higher endurance.
It's interesting to see the difference between what I started with and
what I'm beginning the season with. The first number is the
player's health rating. The second is his defensive range and the
third is his error rating. The lower the number the better:
Before
After
Catcher
Miguel Olivo (2/4e2)/Joe Mauer (6/2e1) Miguel Olivo
(2/4e2)/Brian Schneider (2/2e1)
First
Base
Paul Konerko
(1/3e7)
Ben Broussard (1/2e9)
Second
Base
Adam Kennedy
(1/2e14)
Brian Roberts (1/2e8)
Shortstop
Alex Rodriguez
(1/2e8)
Alex Rodriguez (1/2e8)
Third
Base
Eric Chavez
(3/1e17)
Mike Lowell (1/2e8)
Left
Field
Jason Bay
(3/3e3)
Jose Guillen (1/3e7)
Center
Field
Torii Hunter
(1/1e5)
Torii Hunter (1/1e5)
Right
Field
JD Drew
(1/2e4)
JD Drew (1/2e4)
DH
Chipper Jones (2)/Cliff Floyd
(3)
BJ Upton (1)/Jose Offerman (2)
The Chipper/Floyd combo was more productive and obviously had more
power than Upton/Offerman. But the possibility of losing one
or both to injury for 15 games at a time then having to DH Alfredo
Amezaga is not overly appealing. Plus, the injury-prone duo
cost more, thereby limiting options elsewhere. Pokey Reese
replaces Amezaga as the designated late inning defensive replacement
and should provide considerably more offense should it be needed.
While the new squad doesn't have as much offensive potential, it has a
very slight advantage defensively. The main advantage is that
they will play everyday and I shouldn't have to worry too much about
injury replacements - not that I really have any to begin with.
Reese is about the only back-up plan I can afford. The quality
inexpensive back-ups were rostered long before my eureka. Here's
hoping
that Strat hasn't incorporated surprise injuries this year.
The starting staff started out as Matt Clement, Barry Zito, Mark
Mulder, Matt Morris and Jeremy Bonderman. Clement and Bonderman
made the cut, but the rest of the rotation now features Roy Halladay,
Joel Piniero and Erik Bedard. I'm not sure how that will play
out, but I'm hoping the defensive upgrade in Strat over their real life
counterparts will make a significant difference in their performance,
especially with two pitcher's parks in the division. The bullpen
still has Eric Gagne, Eddie Guardado and Brendan Donnelly to finish
games. Neal Cotts' significant split against right-handers made
him the choice as a middle reliever.
I chose Scott Kazmir for the final spot because I wanted to see how
much defense can affect a player with a crazy card but significant
upside. Scott Kazmir's walk rate in 2004 was pretty frightening,
but his strikeout rate was just nuts (11.07 per 9). With the
exception of centerfield and left, the Devil Ray defense in 2004 could
be downright terrible. He's available both as a starter and
reliever, but I plan on using him primarily as a mop-up man. I
figure blow-outs are the perfect situation to run an experiment that
has the potential to go horribly awry.
Will it work? To be honest, my brain is so overwhelmed with all
the Strat-o-info I've crammed in the last week I haven't the slightest
idea. At best, I've developed some interesting theories that may
provide someone inspiration about their own team. So I'll go with
this team and see what happens. Anyway, that's the plan.
For now.