WAIBI SAID: HEY!! CHOOSE THE OPPORTUNE PLAYER, NOT THE OPPORTUNISTIC ONE!!
ABOUT DATABASE ON BROOKSBASEBALL
As we have advanced in our previous articles, WAIBI Sabermetric
System is focused on the qualitative evaluation of the players, not on their historical
performance. For these reasons the system discards (or tries to dismiss as
accurately as possible) the inconsistencies that arise in the application of the
conventional notation system, in which we identified “Default” and “Average” as
its fundamental syndromes.
We have been developing the first phase of the system, but
without possessing a database that meets
the necessary characteristics for our work, we have relied on other existing,
which provide a fairly detailed data flow. So we've selected BrooksBaseball.
The choice has been justified in the distribution of data
which can separate or identify the actions of each player progressively (whether
batter or pitcher), so, we can delete multiple "default" data.
Remember that even though “default” data doesn’t depend directly from Pitcher_vs_Hitter
relation, it affects significantly the numbers of each player, distorting the
decoding and final accuracy in the analysis. However, other "default"
data distort the interpretation, (and therefore, the real idea about the
quality of a specific player) could not be eliminated. This happens because the
error is created directly from the annotation rules in force in the Official
Baseball Rules.
For WAIBI, the distribution of each data (action) play by
play is indispensable. The essence of
its functionality isn’t in the details of the phenomenon (as in most of the
current sabermetric stats) but in the details of the pieces that make up the phenomenon,
and how intermesh one with the other. This
makes it undoubtely more accurate to evaluate the quality of a specific player,
even from this initial stage of implementation. The first Waibi rule is all
about avoiding as much as possible what we recognize as Average Syndrome. However,
the database implemented by BrooksBaseball, or at least the version we have
accessed, contains a series of erroneous or incomplete informations that are
hindering the successful processing of
all data. For example, the classification of "Des" and "PDes"
columns (the first is the description of the final result of each “at-bat”, and
the second is for each launch) often suggest absurdities in the conventionality
of the game. For example, “hits”
situations, “extra-base hits”, “groundouts” and others, sometimes end in “called
strikes” or “balls” instead of “In play no out” or similar. There’s some descriptions blank. “Hit by pitch”
with “In play no out”. Even “Homers” are described 4 times as “foul”, in the
latest throw of turn. There’re also “out” situations recorded as hits in the conventional play by play
score, just to mention some examples.
Despite this, BrooksBaseball is a complex data
base, with a large number of interesting graphics. It contains invaluable
information. We believe it’s an ideal site to assess specific players trends. This facilitates the characterization of them. In The
Book, Tom Tango refers the illustrative concept of "La Famiglia". The
charts and diagrams in BrooksBaseball allow us to interpret this concept quite rightly, being able to
discern by a deeper study, which are the fathers, brothers, children and grandchildren
of each "Famiglia". In this way, the incongruity of some data, or
even its absence, would not affect the interpretation, because of the enormous
size of the sample, to evaluated numerically
something as inespecific as the trend. It's like the difference between a
mortar and a rifle precision: both are great, the choice depends on the mission.
