Re: Single elimination at PB9

boson0 · Mon, 24 Jan 2000 13:24:28 -0800

By: mickmars_2000 Date: 1/24/00 12:14 pm 
" In
the RR, Illinois, Michigan and Maryland played on 13
common rounds. On those rounds, they compiled a record
of 39-0.

Given an average margin of victory
and standard deviation, we could predict the
likelihood that all three would lose on the same
round.

That probability would have to be considerably less
than 1% or even 0.01%. Then consider that Chicago A
(w. Andrew Yaphe would have also gone undefeated in
the RR and lost on that round. The tournament
occurrence was so statistically impossible that it cannot
hardly have been caused by a Random
Variation."

There are some problems with your "statistical"
analysis. First of all, the 4 undefeated teams were not
playing "average" teams in the first round of the
playoffs - they were playing other teams that qualified,
and hence were significantly above average. I would
argue that the odds of an upset of a team that went
14-0 by a team that went (say) 10-4 are much closer to
50% than your prediction of the cube root of .01% (=
4.6%) which you obviously took out of thin air.

"Penn created and set up a situation where the
consensus #1-4 teams in the nation would go undefeated in
the RR and ALL lose an opening playoff round to teams
well outside the top ten, and with less than 10% of
the votes that the #1-4 teams recieved in QB
polls."

There is no such thing as a "consensus #1-4 team". The
qb poll is a measure of reputation, and hence is
always months behind reality. Depending on who shows up,
what type of questions, and dynamite new freshmen the
results can change rapidly. This latest poll, in
particular, was plagued by lack of votes, protest votes, and
was based on early season results. Use match results
to modify the poll, and not vice versa.

"It
is not coincidental that all 3 of the Nation's top 4
in attendance lost on that round - to teams the
others had decimated in the RR. It was an "upset
special" placed in that position by the PB9 staff to favor
the #4 seeds over the #1's and maximize the
likelihood of defeat. Having read the paacket in advance,
seeing the preliminary statistics, and seeing the
playoff pairings, only a fool would have bet on any of
the #1 seeds to survuve the first round."

I
have not read the packet, I was not at Penn Bowl, and
haven't seen the playoff pairings. However, these teams
that won are not awful - they beat about 10 teams each
to get to the playoffs. It is not a stretch of the
imagination to imagine them winning. 

I may be a fool,
but I find it difficult to believe you can write (or
recognize) a packet on which upsets are guaranteed. The
"better" team will know more stuff more rapidly than the
"worse" team. Unless you intimately know the 32 players
involved your "upset special", seems impossible to me. A
packet could be written to favour one team, or one
player. A packet could be made so difficult or easy or
bad that its results become random. A packet can not
be written to favour a "worse" team over a "better"
team without knowing both the teams in advance. To do
this simultaneously for 4 matches, as you accuse, is
ludicrous.

Rob