A
little mailbag this morning … This subscriber didn't want
his letter published so I will paraphrase his questions to protect the
innocent …
Dear
Lou, I
enjoyed your article about the value of analyzing splits to make
pitching line-up determinations. I have a couple of questions about it.
Thanks
for the kind words! Fire away... (Note: That was word for
word! ) 1)
How much of a sample size is necessary for the split to be meaningful?
(He cites a 21 AB sample)
I
think your premise for the question is flawed and as such, it can drive
you crazy.
I
wouldn't look at any sample being as meaningful and not meaningful. A
baseball date sample is not a zero-sum argument. Instead, I would look
at it as "more" or "less" meaningful.
In
your example 21 ABs is not a sample I'd bet the house on but it's not
throw away either. I wouldn't lean on 21 ABs too heavily in almost all
cases, but if that sample combines with other data to support a trend
or theory or even an established model, I might look at it differently.
For
example knowing that a left-handed hitter likes to go with power to
LCF, if I see he's had 21 great ABs at Fenway, I can put apples and
dumps trucks together and give that little more weight than if I just
have 21 ABs with nothing to back it up. I can theorize that Fenway fits
his offensive style well and those 21 ABs support that ...
Now,
for example, if Jimmy Rollins hits 11 of his 15 HRs in a year in 21 ABs
at RFK, I might look at that sample differently, knowing what I know.
In that case I am probably thinking "aberration" until more ABs prove
me wrong.
And
keep in mind that I have to take 21 ABs and make a guess. That is my
job. I make a living getting to the truth before the average guy ... or
more accurately that's what I try to do. You don't have to, although
that skill is useful in competition, so you don't have to lean as on
heavily on small samples.
I
guess the short answer here would be there is no "truly meaningful"
answer ... 7 ABs has little meaning ... 21 has more meaning
... 210 has a lot of meaning ...
All
of these things are puzzle pieces ... They all relate to the
big picture ... If what you are looking at is an all white piece, that
doesn't help you see what the puzzle is a picture of, then it is of
little help. But If that piece is the black nose that tells you it's a
polar bear in a blizzard, it doesn't matter if the piece is small. It
still reveals a lot.
Use
any sample, but generally the smaller it is the less you lean on
it.
2)
Fantistics has always endorsed the weekly line-up rater as a method to
decide between two players of similar caliber. How does the weekly
line-up rater reconcile with the idea of using splits in making line-up
determinations? So say I have a lefty hitter whose splits suggest that
he won't do well facing lefty pitchers. And he will be facing 4 lefties
in the coming week.
But
according to the weekly line-up rater, the pitchers he will be facing
in the coming week have not done well recently, so his rating is very
high.
On
the other hand, I have a hitter of similar caliber whose splits say he
is a good early season hitter. However, his number on the weekly
line-up rater is low because the pitchers he will be facing have done
well in the last few weeks.
Which
player would you choose to use in this situation? Or put another way,
which tool is more reliable in making line-up judgments when they are
telling me different things?
Again,
think pieces of a puzzle. And again, the premise of the question is
flawed. A doctor has all kinds of tests at his disposal. An MRI isn't
"more reliable" than a blood test, but it sure works better if the
question is whether your elbow ligament is strained or torn.
Nevertheless, it won't do you any good if there are issues with your
cholesterol level.
The
Lineup Rater takes one thing into account ... what that hitter's
scheduled opposing starters for the week have given up for OBP lately
... or in the case of pitchers, what their scheduled offensive
opponents have posted for OBP lately.
When
considering a choice between two players of equal caliber this is a
good thumbnail starting point, but it only looks at one factor. One
guy's opposing starters may be cold but they still may own that hitter
historically. You have to understand that the Lineup Rater is one piece
of the puzzle, a handy way of helping you make your initial decisions
about which guy to start. There are a number of ratings like this, or
splits, that would be helpful, but it is similar to sample size. The
more you take into account the higher the level of "reliability" ...
In
your example, working just on what you gave me, I lean more heavily on
the fact that first hitter can't hit lefties because that's what he's
facing. If his opposing pitchers are cold there could be any number of
other factors involved there, for example his last three starts has
been against the Red Sox, Yankees, and Blue Jays on the road. That
relates less to your hitter than does his problem with left-handed
pitching.
If
you are convinced the second guy's early season success is a
characteristic of the player and not just the result of a good two
weeks two years ago, then I would go with him because his split relates
to his abilities more than the OBP of his opposing starters against
other teams. His split is more relevant ...
But
both are useful. Both have meaning. But in any given situation, some
indicators, and some sample are more useful and more meaningful than
others. The trick is to figure out which is which.
Hey
if it were easy anyone could do it! <g>...
In
the end it is a matter of experience. You do it for a while and find
your own level of faith in all the numbers that are available to you.
Ask around our staff and I bet we all have varying levels of confidence
in any given stat and that confidence is fluid depending on the
question, and the player in front of me. If you look at these things
enough you'll make some good decisions and bad ones but you'll start
developing faith in the indicators that work well for you, the tools
that fit the job in front of you.
I'll
bet they'll end up a lot like ours.
Hope
that helps ...
Thanks
for the discussion and most importantly thanks for being one of us here
at Fantistics!
There are over 100
player news blurbs posted in the member area each
morning, members can read the rest of today's player news by clicking
here: www.fantistics.com/baseball00/. Not a member? Join today:
www.fantistics.com/join/join.php3
Comments (0)