Low seed pool pickers - How common are they?
Tom Adams
seeds for the Final Four?
Based on the little bit of info that I have, I figure that about in
in every 40 pool sheets pick the one seed for the Final Four.
The reason I want to know. Picking the one seed tends to
maximize one's expected score for that round, but if too many
people do it, then it might not be the best policy.
BTW, If you are not saving the info that your pool manager gives you
from previous year's pools, you should. It may be the only way you have
to size up your competition.
--
http://members.aol.com/tadamsmar/poolcalc.htm
Java applet that calculates your best college
hoops tourney pool entry
Tom Adams
>Based on the little bit of info that I have, I >figure that about in
>in every 40 pool sheets pick the one
>seed for the Final Four
I meant 1 in every 40.
Otto Bahn
>
> How common are office pool entry sheets that pick all the number one
> seeds for the Final Four?
>
>
> The reason I want to know. Picking the one seed tends to
> maximize one's expected score for that round, but if too many
> people do it, then it might not be the best policy.
My impression is that it rare for folks to pick *all four*
one seeds...it is also rare for them to pick less than
two one seeds. In general picking all four one seeds does
not win the pool (I'm usually in 1-2-4-8-16-32 pools) as
the FF and next two games pretty much determines the winner
(with the earlier rounds amounting to a big tie breaker).
In my region, you can maximize your odds by dissing the
ACC...but that probably won't work this year.
Another trend I use is never pick less than a six seed for
the final eight. A four seed is about as low as I go for
the FF, but I could go as low as six if there was a sleeper
I *really* liked (I rarely pick that low, but historically
six is about the limit IIRC).
OB
Mr. Poologic
JGA...@khepburn.mc.duke.edu wrote:
> Tom Adams wrote:
> >
> > How common are office pool entry sheets that pick all the number one
> > seeds for the Final Four?
> >
> > Based on the little bit of info that I have, I figure that about in
> > in every 40 pool sheets pick the one seed for the Final Four.
> >
> > The reason I want to know. Picking the one seed tends to
> > maximize one's expected score for that round, but if too many
> > people do it, then it might not be the best policy.
>
> My impression is that it rare for folks to pick *all four*
> one seeds...it is also rare for them to pick less than
> two one seeds.
I agree. I see no evidence that picking all low seeds a
common practice.
> In general picking all four one seeds does
> not win the pool (I'm usually in 1-2-4-8-16-32 pools)
I think that is the most common pool scoring rule.
How many entries does your pool have typically?
> as the FF and next two games pretty much determines the winner
> (with the earlier rounds amounting to a big tie breaker).
You are right that you need to win two contests to win
a pool: (1) you usually need to pick the teams in the late
rounds correctly and (2) you must win the tie breaker among
those that also got the late rounds correct.
But we differ on the best policy to do that.
The way to get the late rounds correct is to enter
multiple sheets that cover all the likely champions and
champion/runner-up matchups. Typically, you have a positive
return on investment for any champion or champion/runner-up
matchup with probability of higher than 1/T where T is the
total number of sheets entered in the pool. People would be
surprised at how many entries they can make where each entry
provides a positive return on investment. 8 or more entries
could be made in a pool with 50 total entries. (This might
raise the eyebrows of some pool managers!)
The way to win the tie-breaker is to go with the objective
favorites in the early rounds. These are usually the
low seeds, but in some cases the Sagarin ratings or betting
line spreads (which are available for the first round) differ
from the seeding order.
Usually, the first few sheets you bet should have the
objective favorites in the FF. These will usually be the
one seeds, but last year Maryland was a little better than
Auburn on Ken Massey's projections.
After your first 4 entries, you might have cause to vary
the FF from the objective favorites.
This will work well if no one else is using the same strategy.
Since picking low seeds is uncommon, few people are currently
using this strategy.
> In my region, you can maximize your odds by dissing the
> ACC...but that probably won't work this year.
The contrarian approach is a good idea if you are
going to enter only one pool sheet. But in the multiple
entry strategy, you will usually get an return on investment
even when betting for a one seed to be champ even when that
one seed is a local fan favorite that is overbet in your
local office pool.
> Another trend I use is never pick less than a six seed for
> the final eight. A four seed is about as low as I go for
> the FF, but I could go as low as six if there was a sleeper
> I *really* liked (I rarely pick that low, but historically
> six is about the limit IIRC).
IMO it is not a good idea to bet a four seed in the FF. Why
pick a team that only has a 20% chance of making the FF when
you can pick four teams that all have a 45% or better chance of
making the FF?
A lot of people seem to think that one should bet an historical
pattern of upsets similar to those that occur in a typical year.
But this is not the best policy.
Consider the following contest: A biased coin is flipped 4
times. You get a point for each flip that you call correctly.
The coin is biased so that it give 55% heads on average.
You look at the historical record for the game and you see
that at least one tail occurs in about 90% of all contests.
Would call one of the flips a tail? Or would you aways bet
all heads? What would be the best policy?
Betting all heads gives the highest expected point total, so
it is the best policy regardless if the historical pattern.
The same goes for the pool. Betting the objective favorite
is the best policy, except for the need to enter multiple
sheets to cover all the likely champions and champion/runner-up
match-ups.
--
http://members.aol.com/tadamsmar/poologic.htm
Learn how to win your college hoops tourney
office pool
Sent via Deja.com http://www.deja.com/
Before you buy.
Otto Bahn
>
> In article <389EE9...@khepburn.mc.duke.edu>,
> JGA...@khepburn.mc.duke.edu wrote:
> > Tom Adams wrote:
> > >
> > > How common are office pool entry sheets that pick all the number one
> > > seeds for the Final Four?
> > >
> > > Based on the little bit of info that I have, I figure that about in
> > > in every 40 pool sheets pick the one seed for the Final Four.
> > >
> > > The reason I want to know. Picking the one seed tends to
> > > maximize one's expected score for that round, but if too many
> > > people do it, then it might not be the best policy.
> >
> > My impression is that it rare for folks to pick *all four*
> > one seeds...it is also rare for them to pick less than
> > two one seeds.
>
> I agree. I see no evidence that picking all low seeds a
> common practice.
>
> > In general picking all four one seeds does
> > not win the pool (I'm usually in 1-2-4-8-16-32 pools)
>
> I think that is the most common pool scoring rule.
> How many entries does your pool have typically?
25-30. The smaller size may be affecting my strategy,
but in the larger pools you just end up tearing up
your picks once you've lost a champion or 2 FF teams,
so I don't like those as much.
FWIW, the 1-2-3-5-10-15 pool is getting more popular
around here. The strategy doesn't change too much, but
the early rounds are more likely to come into play as
more than just a tiebreaker (especially in smaller pools).
> > as the FF and next two games pretty much determines the winner
> > (with the earlier rounds amounting to a big tie breaker).
>
> You are right that you need to win two contests to win
> a pool: (1) you usually need to pick the teams in the late
> rounds correctly and (2) you must win the tie breaker among
> those that also got the late rounds correct.
Exactly, except my experience is that the early round
tie breaker rarely comes into play (although if you do
really well or really poorly, it can negate a FF pick,
or in _one_ case I practically clinched once the FF was
determined due to dark horses).
> But we differ on the best policy to do that.
>
> The way to get the late rounds correct is to enter
> multiple sheets that cover all the likely champions and
> champion/runner-up matchups. Typically, you have a positive
> return on investment for any champion or champion/runner-up
> matchup with probability of higher than 1/T where T is the
> total number of sheets entered in the pool. People would be
> surprised at how many entries they can make where each entry
> provides a positive return on investment. 8 or more entries
> could be made in a pool with 50 total entries. (This might
> raise the eyebrows of some pool managers!)
This will certainly increase your odds...the key of course
is determining "all the likely winners", and if you can
consistently find the winner/runner-up in eight picks, then
I recommend moving to Vegas where you can make a handsome
living. :-)
I see your strategy, but I'm in it for the fun, though I'll
take the money. My strategy is for two picks maximum.
> The way to win the tie-breaker is to go with the objective
> favorites in the early rounds.
Unless you are really "on to something," picking an upset
is generally a bad strategy. Exactly what constitutes an
upset is problematic, and I can see picking as low as 12
(though not very often).
> These are usually the
> low seeds, but in some cases the Sagarin ratings or betting
> line spreads (which are available for the first round) differ
> from the seeding order.
I certainly take such ratings into account.
> Usually, the first few sheets you bet should have the
> objective favorites in the FF. These will usually be the
> one seeds, but last year Maryland was a little better than
> Auburn on Ken Massey's projections.
And last year I just _knew_ it wouldn't be Auburn (over-rated)
or Maryland (big game chokers), thus I was left with St. Johns
or Ohio State. I thought MD would take St. John's and then
choke to Ohio State, so I got lucky. I lost the MD pick, but
Ohio State was more important.
> After your first 4 entries, you might have cause to vary
> the FF from the objective favorites.
Since all four ones almost never make it, the key for me
is figuring out who really are the four strongest teams
(in each region). This is what wins the pool. If you
correctly identify the 3-6 team that makes it, you have a
huge edge over the others who likely chose 1 or 2.
> This will work well if no one else is using the same strategy.
> Since picking low seeds is uncommon, few people are currently
> using this strategy.
>
> > In my region, you can maximize your odds by dissing the
> > ACC...but that probably won't work this year.
>
> The contrarian approach is a good idea if you are
> going to enter only one pool sheet. But in the multiple
> entry strategy, you will usually get an return on investment
> even when betting for a one seed to be champ even when that
> one seed is a local fan favorite that is overbet in your
> local office pool.
>
> > Another trend I use is never pick less than a six seed for
> > the final eight. A four seed is about as low as I go for
> > the FF, but I could go as low as six if there was a sleeper
> > I *really* liked (I rarely pick that low, but historically
> > six is about the limit IIRC).
>
> IMO it is not a good idea to bet a four seed in the FF. Why
> pick a team that only has a 20% chance of making the FF when
> you can pick four teams that all have a 45% or better chance of
> making the FF?
Because you're over-generalizing. What you say would be true
if I had _no other information_. IMHO picking Auburn was a
less than a 20% bet last year. I would have gone with any of
MD, St. Johns, or OSU before Auburn.
> A lot of people seem to think that one should bet an historical
> pattern of upsets similar to those that occur in a typical year.
> But this is not the best policy.
Correct. Even if the pattern holds up, most games are _not_
upsets. The key is picking which ones, and that is very
difficult, hence the minimal gain in picking an upset is
usually lost in the other missed upsets you pick. I find the
easiest strategy is simply concentrating on what "upsets" will
occur among the top four to six seeds, depending on the bracket.
> Consider the following contest: A biased coin is flipped 4
> times. You get a point for each flip that you call correctly.
> The coin is biased so that it give 55% heads on average.
>
> You look at the historical record for the game and you see
> that at least one tail occurs in about 90% of all contests.
>
> Would call one of the flips a tail? Or would you aways bet
> all heads? What would be the best policy?
>
> Betting all heads gives the highest expected point total, so
> it is the best policy regardless if the historical pattern.
> The same goes for the pool. Betting the objective favorite
> is the best policy, except for the need to enter multiple
> sheets to cover all the likely champions and champion/runner-up
> match-ups.
In a large enough pool this strategy almost always loses badly
as the odds also say you will very rarely get all heads, so
the winner will be among those went out on a limb and got lucky.
In a small pool it depends -- are there a bunch of heads fans
in your area? How small is the pool? etc.
OB
Mr. Poologic
JGA...@khepburn.mc.duke.edu wrote:
> Mr. Poologic wrote:
> >
> > In article <389EE9...@khepburn.mc.duke.edu>,
> > JGA...@khepburn.mc.duke.edu wrote:
> > > Tom Adams wrote:
> > > >
> > > > How common are office pool entry sheets that pick all the number
> > > > one seeds for the Final Four?
> > > >
> > > > Based on the little bit of info that I have, I figure that about
> > > > Four.
> > > >
> > > > The reason I want to know. Picking the one seed tends to
> > > > maximize one's expected score for that round, but if too many
> > > > people do it, then it might not be the best policy.
> > >
> > > My impression is that it rare for folks to pick *all four*
> > > one seeds...it is also rare for them to pick less than
> > > two one seeds.
> >
> > I agree. I see no evidence that picking all low seeds a
> > common practice.
> >
> > > In general picking all four one seeds does
> > > not win the pool (I'm usually in 1-2-4-8-16-32 pools)
> >
> > I think that is the most common pool scoring rule.
> > How many entries does your pool have typically?
>
> 25-30. The smaller size may be affecting my strategy,
> but in the larger pools you just end up tearing up
> your picks once you've lost a champion or 2 FF teams,
> so I don't like those as much.
>
> FWIW, the 1-2-3-5-10-15 pool is getting more popular
> around here. The strategy doesn't change too much, but
> the early rounds are more likely to come into play as
> more than just a tiebreaker (especially in smaller pools).
>
I agree that the strategy stays the same. All pools seem
to award a relatively high number of points for the late
rounds. I think this is done to avoid having the pool
winner decided before the championship game. Therefore in
all pools, it is usually necessary to pick the champ and also
win the tiebreaker.
>
> > > as the FF and next two games pretty much determines the winner
> > > (with the earlier rounds amounting to a big tie breaker).
> >
> > You are right that you need to win two contests to win
> > a pool: (1) you usually need to pick the teams in the late
> > rounds correctly and (2) you must win the tie breaker among
> > those that also got the late rounds correct.
>
> Exactly, except my experience is that the early round
> tie breaker rarely comes into play (although if you do
> really well or really poorly, it can negate a FF pick,
> or in _one_ case I practically clinched once the FF was
> determined due to dark horses).
>
> > But we differ on the best policy to do that.
> >
> > The way to get the late rounds correct is to enter
> > multiple sheets that cover all the likely champions and
> > champion/runner-up matchups. Typically, you have a positive
> > return on investment for any champion or champion/runner-up
> > matchup with probability of higher than 1/T where T is the
> > total number of sheets entered in the pool. People would be
> > surprised at how many entries they can make where each entry
> > provides a positive return on investment. 8 or more entries
> > could be made in a pool with 50 total entries. (This might
> > raise the eyebrows of some pool managers!)
>
> This will certainly increase your odds...the key of course
> is determining "all the likely winners",
See:
http://www.mratings.com/cb/ncaa99.htm#win
Massey calculates the probability that any team will advance
to any round. I use this or similar methods to determine
all the likely winners.
> and if you can
> consistently find the winner/runner-up in eight picks, then
> I recommend moving to Vegas where you can make a handsome
> living. :-)
>
I think you are right that you can't consistently pick the
winner/runner-up in eight picks (but I have not done the
math on this).
However that is not the point. The point is just to enter
all the pool sheets that provide a positive return on
investment (ROI). If a winner/runner-up pick has a probability
greater than 1/T where T is the total number of entries in the
pool, then you may get a positive ROI. You need advancement
probabilites and win probabilities to calculate the probability
of a winner/runner-up.
You can estimate the probability that one team will win over
another from the ratings spread. The Sagarin spreads are
unbiased when multiplied by 1.05. Game scores have a normal
distribution with a standard deviation of about 11. See "How to
Play Office Pools if You Must" by D. Brieter and B. Carlin,
_Chance_ Vol 10, No. 1, 1997, pp 5-11.
For teams A and B, the probability for an A/B outcome is:
(prob. that A advances to the final) * (prob. that B advances
to the final) * (prob. that A beats B)
But you may need to discount this probability a bit because you
will not win if someone else bets the same matchup and wins the
tie-breaker.
I will calculate these probabilities for the 10 to 20 most
likely winners and winner/runner-up matchups for this year's
tourney and post them on my web site at:
http://members.aol.com/tadamsmar/poologic.htm
You might want to take a look at that web site between
Selection Sunday and the first game of the tourney. Even
if you don't use my approach, the web site may provide
some useful information and links for you.
>
> I see your strategy, but I'm in it for the fun, though I'll
> take the money. My strategy is for two picks maximum.
>
Well, last year, Massey's page put Duke at 41%, UCONN at 19%,
and Mich State at 15% to win the championship. All these
look like good bets in a pool of size 25-30.
And last year was unusual, because Duke was so strong. In
1998 the top 4 had championship probabilities of 30%,30%,
10%, 7%. In 1997 the values were 20%,19%,12%,10%. So,
3 or 4 sheets are typically a very good bet in a pool of size
25-30.
You should consider betting any team with a championship
probability of greater than 1/25 or 4%. But you may need
to discount your local office fan favorite a bit, since
it will be overbet in your pool.
(But, of course, this approach may not maximize your fun,
depending on what is fun for you.)
> (Sniped out some stuff here)
> > IMO it is not a good idea to bet a four seed in the FF. Why
> > pick a team that only has a 20% chance of making the FF when
> > you can pick four teams that all have a 45% or better chance of
> > making the FF?
>
> Because you're over-generalizing. What you say would be true
> if I had _no other information_. IMHO picking Auburn was a
> less than a 20% bet last year. I would have gone with any of
> MD, St. Johns, or OSU before Auburn.
>
You are correct that 45% is an over-generalization. Massey put
Auburn at 25% to make the FF. He put OSU at 11%.
Massey put only MD over Auburn for that region.
But I still have not seen a case where a four seed should be
advanced to the FF unless you are entering a lot of pool sheets.
You played OSU on a hunch (or a reasoned analysis) and you were
right. But I don't think that an approach based on calculating
advancement probabilities from the spreads would ever put a four
seed in the FF.
Did you bet OSU because you thought that it was the team in that
region that was most likey to make the FF? Or do you deliberately
pick some underdogs for the FF because you view that as part of a
winning strategy?
That contest is not really meant as a pool example, just an example
to illustrate how expected point totals work. If you had to compete
against a bunch of people who were mostly betting all heads, then
it might be a good idea to bet for a tail. But we have already
established that low seed pickers are rare, so that is not really an
issue here.
If I was in a pool where lots of people were copying my strategy or
where lots of people betting all one seeds (or all Massey's highest
probability picks) in the FF, then I would revise my strategy. It
would be harder to win. There would be fewer entry sheets that
would provide a positive ROI.
> -- are there a bunch of heads fans
> in your area? How small is the pool? etc.
I agree that you have to keep an eye the stategies of other players.
I look at the proportion of people who pick only low seeds for the
FF as a indicator of people who are playing a strategy similar to mine.
And size is important, where pools are concerned. The larger the pool,
the more entry sheets there are that will provide a positive ROI. I
have not looked carefully at scaling up to large pools, but I have the
impression that a single player can enter 10% to 20% of the entries in
any pool and have a positive ROI for each of these entries. This
assumes that no other player in the pool is using the same strategy.
--
http://members.aol.com/tadamsmar/poologic.htm
Learn how to win your college hoops tourney
office poolIn article <38A062...@khepburn.mc.duke.edu>,