Pre-Season Spring 2004 Contest Discussion - Part 2 |
: | | | | This Topic has been marked closed. No additional messages may be posted. | | | Page 7 of 10 | | | From: Cthulhu | Posted: 3/15/2004 12:04:46 PM | Message Detail |
I've
just realized that there is a Spring 2004 contest and as I was looking
at the brackets, I cannot believe that there is not ONE Mega Man game... | From: Slowflake | Posted: 3/15/2004 12:29:57 PM | Message Detail |
Ummm... Zelda was a 10-seed, actually. Zelda was the second.
Well,
I didn't get to comment much on yesterday's PotD, thanks to my brother
and his whiny ways (and I still maintain that an N-Gage > siblings)
but I don't have much to think that hasn't been said.
We all
remember how Link beat Cloud 44-26 in that poll, right? Of course it
left the impression that Link's title might be disputed after all. Yet,
come the winning bracket list... Link sextuples EVERYONE, save for a
2-seed. Sephiroth. The other 1-seeds, Mario, Cloud and Snake, in that
order (Mario beat Cloud?!?) were 3rd, 4th and 5th.
And we all
know how it turns out... Link loses to Cloud. Ironically, that mid-June
poll turned out to be more accurate than CJayC's list.
However,
yesterday, it looked absolutely horrible. 48-17... this is monstruous.
It looks as though FF7 is even more the game to stop than Link last
year. Of course, it probably isn't, because I foresee a three-way dance
between CT, FF7 and OoT. But... OoT can probably be more of a key
factor in that huge gap than Sephiroth was last year
And what
about Halo beating SMB3? Well, that's probably nothing... after all,
Snake beat Mario handily in that poll last year if I remember
correctly. It just shows how passionate this little X-Box fanbase
actually is. With three of the strongest possible opponents, and a poll
with an OBJECTIVE nature, Halo still manages to get 15%. Don't get me
wrong, I don't think this is a sign that it'll beat KH, but as far as
Starcraft goes, it looks good.
Now, I had something on my mind
for the last few days. I'd like to start a board-wide project as soon
as the Rumble is completed (it's been relaunched just recently). My
idea is this: to gather as many brackets as I can all across the board,
then find out what games are the favorites to win given matches, and
not just the division championships like Popo did, but first, second
round matches, etc. all the way to the championship.
Do you think that could work? --- SpC2K4 Status --- Winner: FF7 --- Finalist: CT --- Semifinalists: SMB3, LoZ:WW | From: Ngamer64 | Posted: 3/15/2004 3:37:16 PM | Message Detail |
Not
a bad idea, Slowflake... But I like the one that was suggested earlier
even more, where we would base the odds off of just our own brackets,
as in, the 25-30 guys from just these topics. The Contest Discussion
Gurus, if you will (that's my favorite name so far). Speaking of which,
sign up, Slow! You too creative, solar, etc, when you guys get a chance.
And
on that same subject, I think when you put your topic online, Ulti, you
should make it clear that only the Gurus (i.e. those that signed here,
beforehand) are going to be eligible for our little Contest, and the
eventual sigging. That should, in turn, give us some better looking
odds.
--- the-elite.net Ngamer's Contest Archives: http://geocities.com/cyber1166/gamefaqs | From: Slowflake | Posted: 3/15/2004 3:55:14 PM | Message Detail |
Hmmm...
I don't know if that's a good thing or not. First, this behavior is a
little elitist, and second, having no basis for predicting levels the
field more than you think... besides, the casual poster can sometimes
see upsets coming that we don't (Squall over Luigi, Cloud over Link).
Plus, too many people here agree on Metroid beating Pac-Man, for
example: it doesn't do justice to Pac-Man's actual odds. The same case
could be made for Halo/Starcraft, although to a much larger extent. --- SpC2K4 Status --- Winner: FF7 --- Finalist: CT --- Semifinalists: SMB3, LoZ:WW | From: Haste2 | Posted: 3/15/2004 3:56:34 PM | Message Detail |
Heh,
all of this really DOES have little meaning, as characters will almost
always change in popularity anyway, so any arguments about variation is
pretty much pointless. I guess all this stuff will assume that
conditions are pretty much perfect.
I guess I'm just gonna argue to try and extend my mind a bit, you know? It's kinda fun, I guess. XP
Why
a t-test? We don't know the population mean, and this is a binomial
distribution. A simple z-test will suffice, where z=(p-.5)/sqrt(p*q/n)
What?
I did use a Z-Test, just one dealing with proportions, and I said it
right there. And I do not recall the formula that you have there, but I
beg to differ with your results if we're talking about binomial
distrubitions. Maybe if you could remind me what chapter it's in...
Okay, I admit, I haven't thought about the stuff in my Stats class
prior to inference for a while, so thinking of this in terms of
binomial distributions is probably (certainly?) better.
First of
all, binomial distrubutions are very difficult to work with if n is
very high. I don't know if you're genius enough to actually do a
calculation of that magnitude (and TI-83's can't do it), but with a
large number of n, the binomial distrubutions get very close to a
normal distribution. In fact, I will also calculate the standard
deviation of the distribution, as this calculation actually can work
with a large number of n...the standard deviation of Magus vs.
Ganondorf is 166, using (standard deviation) = Sqrrt of all of
(np(1-p)), rounded to the nearest vote. So, in 68% of the matches,
Magus should be no more than 166 votes off his "average." In fact, if
Magus gets 54,179 votes, 1000 votes below his "average," then it would
be 6.04 standard deviations away, which is far, far, below p
(probability) = .001. Now, the varying of total votes won't matter
much, as the proportions will always be about the same, so it wouldn't
affect a normalcdf test much (as in so tiny you couldn't even notice).
Anyway,
most of my discussion was centered around the utility and consistency
of the Linearly Extrapolated model....we have to work backwards from
the end--we can measure consistency across years, and use that as a
proxy for consistency.
Yes...and we only have 2 years to
work with. Unfortunately, with characters always changing in
popularity, it's still hard to do this even with several years. Hm...you know what I think the best character references are for just 2 years, though? Mario, Crono, and Solid Snake. They are almost exactly the same in comparison to each other in both 2002 an 2003.
Eliminate Cloud and Sephiroth, and they did decrease--by about 5.7%. Not very convincing.
It's
fairly convincing in my mind... Barring LInk, Cloud, and Sephiroth,
three Elite 9 characters from the North/East division had barely any
change at all, while the three from the West division did decrease much
more than 5.7%, but more like 10+% (Sonic went from 39% to 34%, Samus
went from 39% to 35%, Mega Man went from 43% to 38%). It's very
systematic. Think what you want with it, though...
--- "Ah, a party! We haven't had one of those. It could be fun! So...what is a party?" "Well, you drink punch and eat CAKE! ...I think." | From: Yesmar | Posted: 3/15/2004 4:33:26 PM | Message Detail |
Count me in also 1. UltimaterializerX 2. Shake 3. Starion 4. DomaDragoon 5. Ngamer 6. ChichiriMuyo 7. cyko 8. Z1mzum 9. Haste2 10. Neoatomtaco 11. Heroic Mario 12. i am vishnu 2 13. Yesmar --- Heh Heh... The wind... It is blowing...--Ganondorf
| From: ChichiriMuyo | Posted: 3/15/2004 5:20:16 PM | Message Detail |
Mario
and Crono are the same from 2002 to 2003 if you round to the
hundredths... if I had the time and patience I would have long ago
setup a list of character strength in comparison to those two. They may
not be the two most reliable guys to some people, but their performance
is consistent. --- This spot is reserved for the stat-head with the best Sp2k4 prediction. | From: smitelf | Posted: 3/15/2004 5:27:40 PM | Message Detail |
And
what about Halo beating SMB3? Well, that's probably nothing... after
all, Snake beat Mario handily in that poll last year if I remember
correctly. It just shows how passionate this little X-Box fanbase
actually is. With three of the strongest possible opponents, and a poll
with an OBJECTIVE nature, Halo still manages to get 15%. Don't get me
wrong, I don't think this is a sign that it'll beat KH, but as far as
Starcraft goes, it looks good.
Yes indeedy. I didn’t laugh
at SMB3 being beaten by Halo because I thought it was meaningful…just
because it’s hilarious in its own right, independent of actual meaning
for my bracket. Although I will be voting for Starcraft in the polls, I
have Halo to win in my bracket right now.
Now, I had
something on my mind for the last few days. I'd like to start a
board-wide project as soon as the Rumble is completed (it's been
relaunched just recently). My idea is this: to gather as many brackets
as I can all across the board, then find out what games are the
favorites to win given matches, and not just the division championships
like Popo did, but first, second round matches, etc. all the way to the
championship.
Do you think that could work?
I like this idea. Please, do try it. I’m curious to see how it will turn out.
Hmmm...
I don't know if that's a good thing or not. First, this behavior is a
little elitist, and second, having no basis for predicting levels the
field more than you think... besides, the casual poster can sometimes
see upsets coming that we don't (Squall over Luigi, Cloud over Link).
Plus, too many people here agree on Metroid beating Pac-Man, for
example: it doesn't do justice to Pac-Man's actual odds. The same case
could be made for Halo/Starcraft, although to a much larger extent.
Actually,
I was just thinking of that Metroid v. Pac-Man match today. I still
have Metroid winning but Pac-Man’s name-brand appeal does give it more
of an edge than it is credited with here. More importantly, however, as
you were saying, we all think along very similar lines aside from a few
issue areas and it would be good to get some opinions from people who
think Halo will beat Kingdom Hearts. I think they’re full of crap but
they bring a different perspective and, educated or not, it’s the
perspective of a voter who might be more in-tune with the masses than
we are. Might not, too, but there’s still that possibility.
As for the binomial distributions being spoken of…nah, never mind, I’ll comment on it later if I get the opportunity. --- "Your fate is sealed, and none but yours." -- Auron Proud Supporter of Starcraft in the Spring 2004 Contest | From: Tequilla Gundam | Posted: 3/15/2004 7:23:23 PM | Message Detail |
GUYS..This
is the problem with all your brackets..THEY ARE TOO COOCKIE CUTTER. If
you want the games...you gotta think UPSET!...If you have OoT or FF7
winning it all..try Chrono Trigger*smirks at Slowflake*..Look at
Cloud/Link....IF you would have posted Cloud winning on this board you
would have been laughed at(I Was*crys*). Thats how I see CT vs
OoT/FF7...Im not saying go ALLOUT....but have some surprise picks
like...SH2>SMW,Tetris>Zelda I, SOTN>GE,SM>LLTP...PICKS LIKE
THAT get you the games...IF 65% of gamefaqs has FF7 winning..GO WITH a
DIFFERENT GAME....you have a better chance --- Fools. Do you really think KH has a chance against Starcraft?Korea knows about this contest. You have been warned.~Captain Roy Falcon
| From: smitelf | Posted: 3/15/2004 7:27:17 PM | Message Detail |
IF 65% of gamefaqs has FF7 winning..GO WITH a DIFFERENT GAME....you have a better chance
...and going against the game I expect to win gives me a better chance...how? --- "Your fate is sealed, and none but yours." -- Auron Proud Supporter of Starcraft in the Spring 2004 Contest | From: Tequilla Gundam | Posted: 3/15/2004 7:31:18 PM | Message Detail |
[This message was deleted by a GameFAQs Moderator] | From: smitelf | Posted: 3/15/2004 7:39:24 PM | Message Detail |
Because you touch yourself at night.
Damn, how did you know? The spies, they're everywhere... --- "Your fate is sealed, and none but yours." -- Auron Proud Supporter of Starcraft in the Spring 2004 Contest | From: Haste2 | Posted: 3/15/2004 9:27:49 PM | Message Detail |
Argh, I typed have the message and lost it...
Phantasy
Star vs. Contra - remember when I said I was going to by Phantasy Star
Collection for GBA? Well, I did. :) PS is a good game (for it's time),
but I've realized that old RPGs don't stand up today as well as old
action games do... or at least it seems that most agree with that. I
played Contra a couple years ago(I never played it back when it was
popular), and I had a blast playing it! That's another reason I'd have
Contra taking it still. Of course PS may (only may) have more
recognition, but we should realize this as well.
Pac-Man vs.
Metroid - indeed, Pac-Man is not getting the credit it deserves.
Pac-Man as a game could be far stronger than as a character. I mean, I
would never vote for the character Pac-Man in a contest unless it was
to anti-vote someone else. Metroid is no Super Metroid, either...not
even close to it. It's starting to worry me a bit. We could see another
Wario vs. Shadow the Hedgehog....but much earlier in R1. *_*
Oh,
and I also must comment on Halo's "performance" in yesterday's poll...I
don't think it means much at all that it's ahead of SMB3. In fact, I
think the majority of the people don't think an NES or SNES game has a
chance of winning the contest. I think that's what helped Halo get
ahead of SMB3. I'm still no more confident that Halo will beat
StarCraft.
--- "Ah, a party! We haven't had one of those. It could be fun! So...what is a party?" "Well, you drink punch and eat CAKE! ...I think." | From: UltimaterializerX | Posted: 3/15/2004 9:32:45 PM | Message Detail |
This behavior isn`t elitist. It`s just our little contest, that`s all. --- MIASU!! My SC2K4 Petition: http://cgi.gamefaqs.com/boards/genmessage.asp?board=7&topic=12558738 | From: Team Rocket Elite | Posted: 3/15/2004 10:13:00 PM | Message Detail |
Anybody
remember how many brackets had Master Chief winning the Summer 2003
Contest? I don't remeber the exact number, but it was quite a bit. This
might explain why Halo beat SMB3 in the poll. SMB3 might have a lot
more fans overall, but Halo fans consist of a much greater percentage
of hardcore fans who beleive it can overcome anything. --- Do you have any advice for filling out our brackets? ''Never underestimate the true power of fanboyism.'' ~CjayC | From: DaruniaTheKing | Posted: 3/15/2004 11:05:42 PM | Message Detail |
Over 2000 brackets had MC taking it all. There were the same ammount for Vercetti, IIRC. Go figure.
Also,
today's poll is interesting. The 7.63% (And growing*, 15 minutes ago
they were .16 less) of GameFAQs is European... that means they haven't
played (Most likely) CT, FF6 (And the ammount of people with the PSX
port is so low is not even funny), and probably others, too...
That could
be a factor, but we won't see until CT vs. SoM and FF6 vs. MK (Both SoM
and MK were released in Europe), so hey... it could be pointless...
But
it's probably more of a factor it was on the Summer Contests... think:
characters, you can vote for pictures, but... pictures will do **** in
this contest, so games will have to advance soley by name (Which can
give Non-European games a small throwback, they're losing several
potential voters)
It's just rambling, but meh...
*Think It's 8:05 AM here, on GMT+1 time, so we have to wait for the Europeans to wake up before they all vote.
--- Supporting Alis Landale and Phantasy Star 1. Contra is dum. AIM: DaruniaTheK1ng | From: UltimaterializerX | Posted: 3/15/2004 11:06:49 PM | Message Detail |
Last year, I believe over 3400 people had MC as their champion, and roughly 3000 also picked Vercetti. --- MIASU!! My SC2K4 Petition: http://cgi.gamefaqs.com/boards/genmessage.asp?board=7&topic=12558738 | From: Mega Psycho Crusher | Posted: 3/15/2004 11:08:50 PM | Message Detail |
Think It's 8:05 AM here, on GMT+1 time, so we have to wait for the Europeans to wake up before they all vote.
Yeah,
but it's 2:05 AM on the east coast, and so most Americans are asleep
(as it is a weekday). Only reason I'm up is because I have to pay the
price for massive procrastination =/. --- SpC2K4 Predictions: Winner: CT | Finalist: OOT | Semi-finalists: SMB3, FFX | From: Sir Shake | Posted: 3/15/2004 11:11:51 PM | Message Detail |
08.10 here now.
Goedemorgen.
--- Shake : Can ya cook? Kali101 : Not really. I nearly killed myself once, trying to make spaghetti. | From: UltimaterializerX | Posted: 3/16/2004 12:04:26 AM | Message Detail |
3:05 AM here, baby. --- MIASU!! My SC2K4 Petition: http://cgi.gamefaqs.com/boards/genmessage.asp?board=7&topic=12558738 | From: creativename | Posted: 3/16/2004 12:55:32 AM | Message Detail |
but
with a large number of n, the binomial distrubutions get very close to
a normal distribution. In fact, I will also calculate the standard
deviation of the distribution, as this calculation actually can work
with a large number of n...the standard deviation of Magus vs.
Ganondorf is 166, using (standard deviation) = Sqrrt of all of (np(1-p))
I
did use this in my post, Haste2. I made a point of showing that the
result was the same using a normal distribution to approximate the
binomial.
In fact, if Magus gets 54,179 votes, 1000 votes
below his "average," then it would be 6.04 standard deviations away,
which is far, far, below p (probability) = .001
No. This is a poll.
Assuming constant vote totals--which we are--then one vote lost for
Magus is one gained for Ganon. The Z-Score is about 1.96, which has a
p-value of roughly .025. Ganon had a 1 in 40 chance of winning that
match.
Also, why are you using a 1,000? The difference in votes between Magus and Ganon was 650.
but I beg to differ with your results if we're talking about binomial distrubitions.
I found this example on the net; perhaps it will help. Look at Sample Problem:
http://www.drake.edu/artsci/faculty/wrights/s158f03/binomial.html
Just plug in the numbers: p=.502962 "pi"=.50 (1-sided test) N=109,708 => denominator = 0.001509562 and numerator = 0.002962409 => z = 1.962430151
Heh, all of this really DOES have little meaning, as characters will almost always change in popularity anyway
The
major characters really don't, though. That's why the Linearly
Extrapolated model was so dominant last year. Using it, I utterly PWNed
the Oracle challenge towards the end. I just wish I had used it
earlier...it's just that I used it at the beginning, but the result of
Jill vs. Squall made me completely lose faith in it. Little did I
suspect the real reason for that. But anyway, it was stunningly
accurate for most matches where it could be applied, and caused me to
predict the Cloud vs. Link results very accurately. It also made me
predict Cloud to beat Sephiroth with 52% of the vote, but I was rooting
for Sephiroth, so I picked him in the contest despite being almost
certain Cloud would win.
You see, even though the model implied
that Cloud vs. Sephiroth would be close, when it comes to heavyweights
we can have a lot of confidence in extrapolated results. If
heavyweights' matches imply a victory by one guy with only 52%, you can
be fairly confident that guy will win. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: creativename | Posted: 3/16/2004 12:59:11 AM | Message Detail |
My
idea is this: to gather as many brackets as I can all across the board,
then find out what games are the favorites to win given matches, and
not just the division championships like Popo did, but first, second
round matches, etc. all the way to the championship.,
I was
actually planning on doing this too. I was going to make a website
where on one page you could replicate your bracket and submit it. It
would've been in PHP to save myself data entry.
I was also
thinking of maybe writing something to parse through the bracket files.
That way, people would just have to save their bracket as an .html
file, then submit them and have their predictions parsed automatically.
However, if I ever do getting around to doing that, it probably won't
be for a month or so, thus eliminating its usefulness in helping
predictions. Another big problem is that people's contact information
is on the same page as the bracket. So they'd want to empty those
fields, update their bracket, then save the file and submit it to me,
then type in their information and update again. Seems like a hassle,
though much less than typing or replicating the whole bracket.
...and going against the game I expect to win gives me a better chance...how?
As
stated, it's a matter of Game Theory. If you think FF7 and Ocarina of
time is very tough to call, and yet FF7 is a heavy favorite in the
brackets, then obviously you have a better shot of winning the contest
if you pick Ocarina of Time to win that match. It would be illogical to
pick FF7.
Of course, if you are somewhat confident that FF7 will win it all, then obviously you should pick it.
Also, today's poll is interesting. The 7.63% (And growing*, 15 minutes ago they were .16 less) of GameFAQs is European
Ah yes...the infamous European factor.
If
someone had stored the evolution of this poll across the day, then that
might give some interesting results. Presumably, Europeans would be a
greater presence in the early morning hours (for the US), because of
the time difference. I expect that the % of Europeans (currently
13.92%) will decrease during the day time, when the US is awake. (if it
doesn't, it might indicate that people are *lying*, heh)
I think
this would be of limited use though. If the European factor was really
significant, then Crono--who is presumably not popular in
Europe--would've performed differently in his matches with Mario.
Instead, Crono did quite well in the "prime Europe hours" last year. We
haven't really seen anything to indicate that Europeans vote any
differently from North Americans. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: creativename | Posted: 3/16/2004 1:22:20 AM | Message Detail |
No. This is a poll.
Wow...I just read that and it sounded very rude. I apologize if it was, I didn't intend it to come out that way.
I can also probably make myself clearer with regards to the Normal Approximation:
The
total number of votes was 109,708. Thus, the break-even point is
54,854. Magus had 55,179 votes. The standard deviation is 165.6. Magus
was 325 votes from the break-even point.
*55,179 - 54,854 = 325 *325 / 165.6 = 1.962 *p-value(1.962) = .0249 *1 / .0249 = 40.23
Thus, there was a 2.49% chance that Magus would not win that match. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: creativename | Posted: 3/16/2004 1:27:57 AM | Message Detail |
Oh,
and as I said earlier: the actual problem is more complicated and that,
and all the complications are on the side of greater volatility; thus,
the 2.49% should be treated more like a minimum bound for Ganon's odds
of winning the match. Though I'm thinking his actual odds of winning
probably wouldn't be too much different.
But with the
"outside interference" and "cheating" factors (the former being a lot
more important and powerful than the latter, from what we've seen), any
match that's close will have some air of unpredictability around it.
The actual standard deviation should be greater than 165.6. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: solarshadow | Posted: 3/16/2004 4:25:56 AM | Message Detail |
And here I thought this would be the one place I could go to forget about my Probability Theory homework...
No
kidding. I study all week for my stats exam, to the point where I'm
seeing statistics in my sleep, write my exam, then come to the board to
relax and what do I see? A freaking z-test in my discussion topic! You
guys are killing me... ;p
Actually, I suppose it was just a
matter of time before the stats topic produced a z-test. Maybe I should
be surprised that it took until the third contest. :) In fact, the
probability of a z-test appearing in here at any given time is normally
distributed with a standard deviation of... (ahem)
My idea is
this: to gather as many brackets as I can all across the board, then
find out what games are the favorites to win given matches, and not
just the division championships like Popo did, but first, second round
matches, etc. all the way to the championship.
I think that
is an excellent idea. It sounds like an enormous task though. I'd be a
little worried about poster honesty, but it would give us a very
interesting look at what the board really thinks.
Anybody
remember how many brackets had Master Chief winning the Summer 2003
Contest? I don't remeber the exact number, but it was quite a bit.
Here, I'll give you the top 10:
1. Link - 17571 2. Sephiroth - 3265 3. Mario - 2953 4. Cloud - 2482 5. Snake - 2378 6. Vercetti - 2260 7. Samus - 1688 8. Master Chief - 1488 9. Crono - 650 10. Mega Man - 599
So about 3.62%. Which is significant.
That's
why the Linearly Extrapolated model was so dominant last year. Using
it, I utterly PWNed the Oracle challenge towards the end.
Okay, you're officially disqualified from this year's. ;)
As
stated, it's a matter of Game Theory. If you think FF7 and Ocarina of
time is very tough to call, and yet FF7 is a heavy favorite in the
brackets, then obviously you have a better shot of winning the contest
if you pick Ocarina of Time to win that match. It would be illogical to
pick FF7.
This is not a strategy I necessarily agree with
(mainly because it has served me poorly in the past). Being different
from the majority in critical matches only works if the majority ends
up being wrong. If the majority is right (as it often is) then you're
screwed. But if you side with the majority in a key match like FFVII
vs. OoT, you can always pick up extra points in a less-critical match
(maybe something in the mid-rounds that you have a particularly good
read on). I can see the rationale in matches that you think are 50/50,
but the inescapable fact is that you need to get the winner right
in order to have any chance of winning. So picking with your gut is
going to serve you better than relying on what other people didn't pick. (Says me, anyway.)
Only reason I'm up is because I have to pay the price for massive procrastination =/.
And that price is... hanging out on the contest board? =)
Oh, and just in case you guys thought you could have a competition in here without me, take this:
1. UltimaterializerX 2. Shake 3. Starion 4. DomaDragoon 5. Ngamer 6. ChichiriMuyo 7. cyko 8. Z1mzum 9. Haste2 10. Neoatomtaco 11. Heroic Mario 12. i am vishnu 2 13. Yesmar 14. solarshadow --- Contest Stats: http://solarshadow-stats.tripod.com | From: Cromage | Posted: 3/16/2004 4:34:05 AM | Message Detail |
Mmm.
But to win the contest, you've got to beat all else. If the majority is
right, and you go with them, then you've got to distinguish yourself
some other way. Which may or may not be possible. Go against the
majority, and guess correctly, and you've just increased your chances
by a factor of 10.
...
Don't quote me on that. --- http://crolapras.tripod.com/archivist.html ~Archivist of the SC2k3 board..... a year late. | From: cyko | Posted: 3/16/2004 4:40:22 AM | Message Detail |
alright, Solar!
Hmmm...
I don't know if that's a good thing or not. First, this behavior is a
little elitist, and second, having no basis for predicting levels the
field more than you think... besides, the casual poster can sometimes
see upsets coming that we don't (Squall over Luigi, Cloud over Link).
Plus, too many people here agree on Metroid beating Pac-Man, for
example: it doesn't do justice to Pac-Man's actual odds. The same case
could be made for Halo/Starcraft, although to a much larger extent.
come
on, Slowflake. it's not elitist. a couple of people that don't post
very frequently already joined our little competition. when NGamer said
to only allow people who posted in this topic to post their brackets,
he meant the people that already signed up in this topic and not to
allow any late joiners. and besides that, the majority of people in
this topic are a little different (at least while posting in this
topic) by remaining impartial when it comes to games and characters;
and not resorting to flaming or fanboyism. they try to base their
predictions on stats, logic, and cold hard facts. not that cold, hard
facts always work, though..........................
--- Please go nominate Secret of Mana for SNES. Right Now. A Cheese Legend of Trivia 12 | From: InsuIt Comic Dog | Posted: 3/16/2004 4:46:52 AM | Message Detail |
Elitist?
Yes, you're elitist alright. Nobody is allowed into your inner circle
ruled by complex math and glasses three times the size of your own
eyes. Nice list, yés. If I ever need a geek to fix my DVD player I'll
give yoú a call, okáy?
And if anyone of you has a hot date, make
sure to call an expert for that one. You guys have...other qualities.
Now, don't let me interrupt your Z-test.
--- I keed, I keed. I make a little joke! You will DIE alone. | From: Heroic Tails | Posted: 3/16/2004 5:06:18 AM | Message Detail |
solarshadow, the fact is that you can't be sure
of who is going to win. All you can extremely loosely estimate are the
chances of winning of each characters in a match. So whatever your
choices are, with or against the majority, it may very well be that
you're going to be wrong.
But the thing is, if the majority
decides a match is a given when you feel it is going to be a 50/50
chance, then by choosing the least popular choice, not only do you not
decrease the expected number of points this match is going to earn you,
but in the case you're right, you'll have a huge advantage towards the
majority of people, which is the important thing here.
If we consider again the Mario/Crono match for example, I think we can fairly assume two things: 1. it's close to a 50/50 chance. 2. more people chose Mario to win than Crono.
Then the best is to choose Crono. You can't say afterwards
"oh yeah, Mario won, so that was obviously the best choice". Crono
could very well have won this one, I think we can all agree on this.
Now
of course, this doesn't work if we assume the existence of the infamous
Mario clutch, which skyrockets his chances of winning (thus nullifying
the first hypothesis).
So yeah, I chose OoT over FF7, Sonic 2
over SMW, Tetris over Zelda, because I feel all of these matches are
going to be very close. --- "I tried a SMB speed run once but I
ran into the first Goomba and died because I forgot to jump. After
that, I never tried again." - PsychoSwordsman | From: ChichiriMuyo | Posted: 3/16/2004 7:14:22 AM | Message Detail |
But
if Mario had been a finalist and the favoring was heavily towards him
your choice of Crono lost you the entire contest before it even ended. --- This spot is reserved for the stat-head with the best Sp2k4 prediction. | From: smitelf | Posted: 3/16/2004 8:13:48 AM | Message Detail |
If we consider again the Mario/Crono match for example, I think we can fairly assume two things: 1. it's close to a 50/50 chance. 2. more people chose Mario to win than Crono.
Then the best is to choose Crono.
Considering that you therefore lost points I'd say this strategy did not serve you well. --- "Your fate is sealed, and none but yours." -- Auron Proud Supporter of Starcraft in the Spring 2004 Contest | From: torey luvullo | Posted: 3/16/2004 8:24:01 AM | Message Detail |
i understand the desire to try to tame the beast of choicemaking by mathematical models and game theory.
but
the fact of the matter is, the contest is so competitive and the
contestants are so numerous and knowledgable that our only chance to
survive is to pick nearly every match correctly. since among those
matches are going to be several which are destined to be close matches,
good luck - we'll need it. | From: creativename | Posted: 3/16/2004 8:36:08 AM | Message Detail |
Maybe I should be surprised that it took until the third contest
I
too was surprised no one's discussed this before. I suppose it must be
because its not very useful for predictive purposes; its only use is
for interpreting matches after they've already happened, and saying
"So-and-so had __% chances of winning this match..." For instance, we
can say that AiAi's chances of winning against Link were...um...0%
<cough>. OK then...stupid Excel. It doesn't go beyond 30 decimal
places. Which means AiAi's chances show up as 0%. What an outrage!!
Okay, you're officially disqualified from this year's. ;)
NOOO! I was just being facecious :(
But
anyway, the bad thing about the LE model is that anyone can easily use.
So if all of a sudden some people use it--and I know kawaiifan used it
last year, though I don't know to what extent--it'll lose its "Oracle
Challenge Value". <sigh> That stupid Game Theory stuff again :-|
Being
different from the majority in critical matches only works if the
majority ends up being wrong. If the majority is right (as it often is)
then you're screwed.
But to paraphrase what Cromage
said--since I can't quote him on it ;-) --...if you always go with the
consensus, your chances of winning are basically non-existent.
But
if you side with the majority in a key match like FFVII vs. OoT, you
can always pick up extra points in a less-critical match (maybe
something in the mid-rounds that you have a particularly good read on).
On
the contrary; going against the majority in lesser matches is unlikely
to matter much. Due to the extreme weights given to the later matches,
that's what its all about. If you rely on earlier matches to somehow
distinguish you at the end, you're wasting your time.
If FF7 is
indeed as heavily favored in this contest as Link was in SC2K4 (and I
don't know if this is really the case, since the poll might not
completely represent the brackets), then unless you're very convinced
FF7 really is a significantly better bet to win it all than anyone
else, you're better off going with an "underdog" choice (if you really
want to finish #1). But if you're just looking to maximize your
expected points--rather than odds of winning--and believe that
the consensus is best, you should pick it. Maximizing expected points
is about risk aversion; maximizing odds of winning relies on volatility
dynamics. This is because your bracket is like an Option, in Finance:
you can relate your finish as the Stock Price (lower is better), with a
Strike Price of 10. This option only finishes with value, or "in the
money", if you finish in the Top 10.
So picking with your gut is going to serve you better than relying on what other people didn't pick
Well, the assumption is that you only use this strategy when you think its appropriate anyway. Its based on your
belief that a match is 50/50. If you think that--to use the same
example we've been using--Final Fantasy 7 vs. Ocarina of Time is not
50/50, or close to it, then obviously you'd pick whoever you thought
was going to win.
From: InsuIt Comic Dog | Posted: 3/16/2004 7:46:52 AM | Message Detail Elitist? blah blah blah some random nonsense blah blah blah Z-test.
All right, fess up. Which one of you created this alt account?? My money's on Ultimaterializer.
I love Triumph, BTW.
But
if Mario had been a finalist and the favoring was heavily towards him
your choice of Crono lost you the entire contest before it even ended.
Given
that Mario's chances of making the final were essentially the same as
Crono's--and both weren't likely given that Sephiroth and Mega Man were
in the same half of the bracket--I don't see why this matters. Like
Heroic Tails said, this is relying entirely on hindsight (unless you
believe someone will always cheat for Mario). Not choosing Cloud to win
cost every one of us the whole bracket before it even ended. So what? --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: creativename | Posted: 3/16/2004 8:39:04 AM | Message Detail |
Considering that you therefore lost points I'd say this strategy did not serve you well.
Huh?
Picking Link to win the whole thing would've served you far worse than
picking Mario vs. Crono incorrectly, which was only worth a pathetic 4 points
and not at all essential towards winning the whole thing. Mario vs.
Crono has got to be the most "overrated" match there is, in terms of
the attention it gets. Which makes me wonder...how many people in the
Top 10 even picked it correctly? I doubt more than 6 or 7. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: ChichiriMuyo | Posted: 3/16/2004 8:41:13 AM | Message Detail |
[This message was deleted at the request of the original poster] | From: Heroic Tails | Posted: 3/16/2004 8:50:44 AM | Message Detail |
Considering that you therefore lost points I'd say this strategy did not serve you well.
You missed my point. In case you didn't notice, the goal of the contest is to make the best predictions before knowing the results. In this case, the best choice was Crono.
Let's
say that there is a game where you roll a dice, and you must bet on
either "6" or "anything but 6". Let's suppose we play once, and get a
6. Are you really going to say that the "anything but 6" choice wasn't
the best one? This is just an example btw, I know it isn't that
clear-cut in the Mario/Crono case.
By the way, I picked Mario in this match, because I wasn't thinking of the contest in a game-theoretical way last year. --- "I
tried a SMB speed run once but I ran into the first Goomba and died
because I forgot to jump. After that, I never tried again." - PsychoSwordsman | From: creativename | Posted: 3/16/2004 8:52:35 AM | Message Detail |
BTW,
I didn't mean to imply that I think Mario vs. Crono shouldn't be
discussed. Mario vs. Crono II was definitely the match I was looking
forward to more than any other, and it was also the one that left the
worst taste in my mouth (for obvious reasons). It's just that it really
wasn't that big of a deal to pick the match right. You could easily
have won the whole contest without it. Plus, given that Shameless
Square Fanboyism™ was basically what won you the contest last year, a
good number of the top finishers no doubt were biased in favor of Crono.
I'm
certainly going to mark the inevitable Mario vs. Crono III on my
calendar though (I hope CJayC relies on nominations for the Summer
Contest as well, but I'm pretty sure he'll fix this one no matter
what). And the expected matchup of Chrono Trigger vs. Super Mario Bros.
3 should be quite insane as well (though I have Legend of Zelda beating
SMB3 in my bracket). As we know, a lot of people are going to treat
that as a pseudo-Mario vs. Crono III.
Note that I don't think
Chrono Trigger vs. SMRPG or Chrono Trigger vs. Super Mario World are
really worth of even being a pseudo Mario vs. Chrono III, because the
outcome will never be in much doubt (though I must point out Super
Mario World is one of my all-time favorite games). And while the latter
should happen, I'm not at all convinced SMRPG can beat Street Fighter 2.
...hmmm.
Actually, come to think of it, I wonder if SMW has better chances of
winning than people think (assuming it can beat Sonic 2, which I think
it will). SMW really was a great ****ing game, and sold only a bit less
than SMB3. So maybe that will be a worthy pseudo-Mario vs. Crono III. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: creativename | Posted: 3/16/2004 8:59:57 AM | Message Detail |
Wait
a minute...did Super Mario World come packaged with the SNES? I think
it did, when the system first came out. In which case the sales factor
isn't as important.
Does anyone remember? --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: ChichiriMuyo | Posted: 3/16/2004 9:02:41 AM | Message Detail |
Okay,
that works for Crono/Mario where it's not worth a lot of points, but
were talking about FF7/OoT here, and that's the whole contest as most
people consider it. Sure, you can gamble and hope that you come out on
top, but you probably won't. As stated the majority believes FF7 will
win, but now you're not distinguished from the people who picked OoT
and your chances on the big prize hasn't gotten that much bigger. Oh,
and FF7 is pretty well set to win, it could go the other way, but this
match isn't a 50/50.
When all is said and done, if you've picked
every game with the majority support your bracket will still be quite
different from 99% of the predictors. The majority on one match will
not all be the majority on another. Most people will have FF7 going all
the way... some of those people have Halo losing round 1, many have
Halo round 2, and I'm sure there are some taking it even further. Even
if FF7 Win/Halo round 2 is the majority consensus of the possibilities
provided how does that "majority" get broken up over Contra vs.
Phantasy star? If that's a true 50/50 match we've already broken our
majority. What about VC vs Melee? Majority says SSBM, but will the
majority say Contra over PS , FF7 champion, Halo round 2, AND SSBM over
VC? Not a chance. If Contra has more support than PS then we'll see
that is the largest group, but at that point it's not likely to even be
20%. Throw in DK vs DH, do we even have a 10% group that agress on all
those matches? Nope. The point is, if you go with the majority on the
big matches where one is clear you still have a lot of matches to set
yourself apart.
2002 top 10 results 1 Burney 178 Link - 1500 2 Fat Morton Jellyroll 178 Link - 1254 3 LeLoup 177 Link - 22500 4 synergy 177 Link - 20211 5 I am spam 177 Link - 6908 6 magnolia brandon 177 Link - 165 7 TRA 177 Link - 323500 8 Moogle Man 174 Link - 17000 9 Young Janus 174 Link - 7000 10 SeladonEQ 173 Link - 18321
2003 top 10 results 1 Leo0820 183 2 Lawn Flamingo Lamp 180 3 TheM0ose238 179 4 Johnathon117 178 5 cervesaswatter 177 - Best Tiebreaker (289 off) 6 silvermoon9000 177 7 shahid 177 8 cloudstrifesheart 177 9 ZeonAce0079 177 10 spicymcspice 177
6-7
points is all it takes to make the difference between grandprize and no
prize. If the majority proves to be correct in this contest, following
the majority on every clear cut match will get you up there, and get
the real 50/50s will make the difference between winning and losing.
DK/DH, Contra/PS, Mario RPG/SF2, and the bang/fang triangle are matches
we can't come with an actual majority for, and getting those one right
while say slowflake gets half of them wrong puts me in first and him in
second, or worse. Sure, you can set yourself apart from the majority
pick on one match, and it may even pay off. But then when it comes to
winning the big one you still have to worry about all the other people
who took OoT, which isn't a small ammount, and hope (like if you had
picked FF7 and been correct) that enough of your early matches will be
correct to set you apart. I wouldn't be surprised if everyone from Leo
to Spicy had Cloud beating Seph in the finals, so if shahid may have
gotten 1 four point match right (say, maybe he missed Crono/Mario) he'd
get 2nd place. And you'd dare say Crono/Mario doesn't have much pay-off
in the end? --- This spot is reserved for the stat-head with the best Sp2k4 prediction. | From: torey luvullo | Posted: 3/16/2004 9:34:07 AM | Message Detail |
btw,
here is a noted gambling columnist's take on how to win a 64 entry
bracket, depending on how many people enter. all games theory, no
knowledge needed, is his take...
http://www.freep.com/news/casinos/casino16_20040316.htm | From: creativename | Posted: 3/16/2004 9:49:42 AM | Message Detail |
but were talking about FF7/OoT here, and that's the whole contest as most people consider it
I
don't know about it being the whole contest; Chrono Trigger, and maybe
SMB3 or even another SNES game, has a good shot. It's worth 24 points
though, since the winner should easily beat the 128 representative. So
it is very important.
Oh, and FF7 is pretty well set to win, it could go the other way, but this match isn't a 50/50.
That's
why it makes sense for you to pick Final Fantasy 7: you don't think
it's an even match. We're trying to say, that for those who do think it's even, they should pick Ocarina of Time to increase their odds of winning.
And
yes, everybody's going to have a different permutation. But as has been
stated: guessing right on the early matches is not what's going to put
you in the Top 10. Maybe it'll separate you from number 11...but unless
you get the finals correct, you're odds are pretty small of ending up
in first few hundred, so you can just forget about worrying over
separating yourself from somebody a couple spots below you.
but
now you're not distinguished from the people who picked OoT [...] you
still have to worry about all the other people who took OoT, which
isn't a small ammount,
From the poll though, it might be a
small amount. While the poll didn't include OoT, and while on the
boards it seems that about 40% or so are picking OoT to beat FF7, we
don't know about the actual brackets. In the end, who you should pick
comes down to what you think other people picked, and what you think
the odds are of each game winning. If there's a big asymetry there in
terms of you thinking a match is much closer than others do, picking
the game that the majority didn't is the logical thing to do.
As
for the rest, you basically have it exactly backwards: going with the
majority on the earlier matches is the thing to do. Then, you try to
separate yourself by taking risks at the end.
Going with the
favorites late, and trying to earn separation by picking Duck Hunt over
Donkey Kong or Phantasy Star over Contra, is a losing proposition.
And you'd dare say Crono/Mario doesn't have much pay-off in the end?
"Dare"?
But it's true. Nothing daring about it. You can play the "what if" game
'til the cows come home, or whatever. No doubt that guy who ended up in
78th place was this close to making a couple of relatively
low-point-value picks that would've won him the whole thing. He can see
in retrospect that this "cost me the contest", but...you know. That's
in retrospect.
When you're talking about importance, you have to define it relative to your Objective Function. How sensitive
is the objective function to a particular variable? Obviously, the
objective function of "finishing in the Top 10" is almost completely
insensitive to 1st round picks that everyone knows won't advance past
the 2nd round, and only senstive to a small degree to 4-point things
like Mario vs. Crono. It is however entirely dependent on getting the
winner right, and heavily dependent on getting the Final Four results
right.
If the late matches were very predictable, this wouldn't
be the case. Even though they'd still be worth a lot of points, if
everyone saw what was coming, that's not where the separation between
contestants would take place. That's not true though: the late results
have shown to be very unpredictable. People know who the juggernauts
are, but picking between them is very difficult. And that is what the Objective Function is most sensitive to.
FYI,
I'm also picking FF7 to win that match (and the whole contest). That's
purely out of bias though, not Game Theory based logic, because I think
that Ocarina of Time and Chrono Trigger are almost as good bets as
Final Fantasy 7 is. I just don't think any game deserves to win more
than Final Fantasy 7 does (despite the immense hate for it), so I have
a hard time picking something else. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: Slowflake | Posted: 3/16/2004 9:56:01 AM | Message Detail |
He
can be an authority on the subject all he wants, I'm sticking with what
I have. I rode a cookie-cutter bracket to the top 10 with only three
matches to go, and it took an upset only dumb fanboy luck could foresee
to kill my chances at a prize. Shame that I'm not eligible anymore...
I'll just aim at getting the highest score possible instead of taking
huge risks and winning it all.
(Ironically, it was Mario beating Crono that first put me in the top 50.)
Though I guess some will think WW beating MP and FF10 is a huge upset no matter what. --- SpC2K4 Status --- Winner: FF7 --- Finalist: CT --- Semifinalists: SMB3, LoZ:WW | From: torey luvullo | Posted: 3/16/2004 9:58:54 AM | Message Detail |
actually,
according to glazer's "knowledge-free" hypothesis, the definition of
"upset" is very cookie cutter: anytime a higher seed beats a lower
seed. | From: Team Rocket Elite | Posted: 3/16/2004 9:59:12 AM | Message Detail |
Wait
a minute...did Super Mario World come packaged with the SNES? I think
it did, when the system first came out. In which case the sales factor
isn't as important.
Does anyone remember?
SMW was packaged in with the SNES. Although, I think some of the later ones had a different pack in game. --- Do you have any advice for filling out our brackets? ''Never underestimate the true power of fanboyism.'' ~CjayC | From: Slowflake | Posted: 3/16/2004 10:02:54 AM | Message Detail |
Except
knowledge is actually required on this kind of bracket. 16-seeds aren't
even supposed to come close to 1-seeds, but if you don't think twice
about Halo/Starcraft, then you're the exception.
And who the heck is Sam Fisher, and why is Magus seeded 13? --- SpC2K4 Status --- Winner: FF7 --- Finalist: CT --- Semifinalists: SMB3, LoZ:WW | From: creativename | Posted: 3/16/2004 10:09:57 AM | Message Detail |
http://www.freep.com/news/casinos/casino16_20040316.htm
Nothing there I wouldn't agree with. But the system is very different. This is not a Sports bracket.
In
sports, there is high volatility in the individual match-ups. The best
team will rarely win the NCAA tourney, no matter what people say. A
single-elimination tourney is a horrible way of isolating the best team
that there is (but it's a great way to make money, because of the
excitement). Repeat the same exact NCAA tourney a bunch of times and
the results will vary wildly. One time, a #1 seed will win; another
time a #3 seed; another time yet another #1 seed; and so on. Repeat the
same NCAA tourney a few hundred times, and you're going to get more
than a dozen different champions.
To maximize your Objective Function of winning, the strategy is simple: Go with whoever Vegas goes with.
Why? Because Vegas already takes Game Theory into account for you.
The odds are calculated based on evening the betting on both sides. The
favorite is a *consensus based*. NO ONE knows better than Vegas. Vegas'
favorite will rarely win, because the odds are against it...that's why
the favorite will obviously have less than even odds of winning. But if
Vegas says Duke has the best chance of winning at 5:1, then they do.
It's that simple.
It's the same exact thing in Finance:
consensus rules. The market stock prices are efficient: unless you have
insider info, you can't beat the market (unless your name is Warren
Buffet or Peter Lynch). The consensus is just TOO. FRIGGIN'. POWERFUL.
However, this is not sports. Repeat the 2002 tourney a bunch of times, and Link wins almost every time.
Cloud also wins a vast majority of SC2K3 tourneys. This is because, as
Haste2 pointed out, of the small variance in poll results when dealing
with tens of thousands of votes. In basketball, the variable is on a
much smaller scale: 60-100 field goals taken by both teams (and free
throws, which complicate the issue). If basketball games were a few
hundred hours long, then they'd actually be as accurate and dependable
as these polls are. But in basketball, even the '96 Bulls lose 10
games. Have Link face 82 random opponents in 2002, and he doesn't lose
one match. Cloud might lose a couple in 2003, but not nearly 10.
It's also not equities. There's only one winner here. This is more derivatives,
than it is equities (while you can treat equity as a call option on a
firm's assets, if you increase volalitity, you increase the required
rate of return and thus increase the rate at which future earnings are
discounted, and thus decrease the stock price). And derivatives is what
I went to grad school for ;) Volatility is king here. It's simple: the
value of an option increases with volatility. And you can treat picking
a certain game to win as a call option on that game. Volatility is what makes an option more valuable.
For
instance, SSBM is a wildcard. While, on average, SSBM might be expected
to perform just like another game, you have to take into account its distribution
of expected performance, rather than just the average. SSBM has greater
volatility, and a fatter tail: there's a chance it can go pretty far,
because its so unpredictable. Thus, as an option, SSBM is worth more
because of its volatility. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: creativename | Posted: 3/16/2004 10:27:31 AM | Message Detail |
Oh,
and of course the seedings in the NCAA tourney tend to be a helluva lot
better than the seedings in these brackets. Like Slowflake just said,
you don't have Maguses with 13 seeds in the NCAA.
If Vegas
actually got involved in these tourneys, then what would be your best
bet to both maximize points, and your odds of finishing in the Top 10?
Simple. Go with the favorite every single time. That's because Vegas
has the best system of devising odds: even the money.
We don't
have that system. Just taking the % of brackets that have a certain
character as winning is a poor proxy for a true favorite. In Vegas,
Master Chief and Tommy Vercetti would've been bigger underdogs than
they were with the bracket-makers. Fanboys may put them on their
brackets, but in order to increase their odds with Vegas, they'd have
to put their money where their fanboyism is. Even if the random fanboy
did is, it would get drowned out by the serious money.
Cloud might lose a couple in 2003, but not nearly 10.
Actually,
I just worked out the bionomial odds and even against Link, Cloud had
basically 0 chance of losing in 2003. So yeah, even Cloud would
probably go 82-0 against Link from 2003, let alone random opponents.
Put the Bulls against the 64-18 Sonics, who had the 2nd best record,
for 82 games, and the Bulls are lucky to get 50 wins. (they actually
did face them in the finals that year and beat them in 6 games; I
remember it well...that Sonics team was one of the most underrated in
history...ah well)
Anyway, the point is that the strategies for
winning an NCAA office pool and the GameFAQs contests are very
different. Simply going with what Vegas says is the best way to
maximize your odds of winning an office pool--and if Vegas were here,
it'd be your best strategy here too. But Vegas ain't. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: creativename | Posted: 3/16/2004 10:59:02 AM | Message Detail |
On
the Cloud vs. Link thing: in order to have Link expect to get even one
win against Cloud in 2003, you have to increase the variance by a
factor of almost 6. While I've said I believe the variances are
underestimated by the binomial model, its doubtful they're
underestimated to anything near that degree.
Oh, and I actually mis-spoke when I said that picking the Vegas favorite would be the best strategy. It probably would be, but it's not that simple. To apply the Vegas model to Final Fantasy 7 vs. Ocarina of Time:
*If
Vegas said FF7 was the favorite against Ocarina of Time, maximizing
your odds of winning is still a question of Game Theory. For instance,
if every single bracket picked FF7 just because it was the Vegas
favorite, you'd have to be a moron to pick it as well. That's just an
extreme case; FF7 can still be over-favored. If Vegas said FF7 was a
1:2 favorite to win, you'd have to be an idiot not to expect it
to win, but that doesn't necessarily mean that you should choose it to
win--and you definitely shouldn't bet on it to win in Vegas ;) (see
below).
Now, FF7 being favored by Vegas probably would mean you
should pick it in your bracket, for this simple reason: most brackets
aren't going to go by Vegas. The average bracket will be stupider than
Vegas, just like the average office pool entry is stupider than both
Vegas and the selection committee. But if your objective was to
maximize your odds of winning, and if the average bracket *exaggerated*
Vegas' favorites systematically (i.e., every single favorite was
"over-favored" by the brackets), then you'd actually be better off
picking the underdog 100% of the time (I think). Though obviously, if
you wanted to maximize your expected points total, you'd still
systematically go with Vegas' favorite, no matter how slightly they
were favored. But if every bracket did that, and you actually cared
about winning, then you certainly wouldn't do it as well.
*You
couldn't make money on either one. One or the other will win and half
the betters will make money, but your net expectation is negative
profit. All you'll do is lose on the spread. That's why gambling sucks. --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: creativename | Posted: 3/16/2004 11:22:43 AM | Message Detail |
Yet
another note on this: [how many posts have I had in this topic the past
couple days? Geez. At this rate I'm going to single-handedly bring
about Part 3 of this topic. Sorry Solarshadow for making more work for
you :( ]
The definition of "overfavored by the brackets" is very
unclear. For instance, if a particular game was favored at odds of 2:3,
then you'd definitely expect more than 60% of brackets to pick it to
win; that wouldn't necessarily qualify as being overfavored, because
the relationship is non-linear. That's the problem with Game Theory, of
course--and the reason for its existence: the dynamics are just
ridiculous.
Which is exactly why relying on Game Theory to make
choices for your bracket makes little sense. I can only think of 2
places to apply it: 1) when you truly believe a match is a complete
toss-up, but one competitor or the other is a clear favorite in the
brackets; and 2) a game (or games) is favored to win it all (or still
be present late in the tourney) by so much, that simply choosing it to
win will virtually eliminate your chances of winning, because of the
conditional nature of the choice; i.e., if you do pick it to win and it
wins, you have no chance of winning the contest anyway. Such was the
case with Link last year: if you really wanted to win the tourney, Link
was not a wise choice for champion. (and this has nothing to do with
hindsight; I made this very point to my brother before the tourney
began, though I told him Sephiroth was a smart choice for the
"contestant who wants to win", rather than Cloud, and never would've
suspected Cloud would be the one to dethrone Link) --- Remember to nominate Frog from Chrono Trigger for Summer Contest 2K4! | From: smitelf | Posted: 3/16/2004 11:45:31 AM | Message Detail |
Huh?
Picking Link to win the whole thing would've served you far worse than
picking Mario vs. Crono incorrectly, which was only worth a pathetic 4
points and not at all essential towards winning the whole thing.
Still, the strategy lost you points. You should just go with your gut; don’t worry about everyone else.
You
missed my point. In case you didn't notice, the goal of the contest is
to make the best predictions before knowing the results. In this case,
the best choice was Crono.
No, it wasn’t. I don’t remember
who I chose last year but considering the fact that Mario beat Crono in
2002, why wouldn’t he win in 2003? Yes, it’s always a close match but
Mario has won in the end both years. There seems to be some unknown
factor in his favor…let us just call it the CJayC Factor and say no
more about it. You’re basing your choices off of everyone else’s, which
I think is stupid. If you just do your own damn bracket you’ll do much
better than trying to second-guess who everyone else is betting on.
Mind
you, for some absolute 50/50 matches, such as Mario/Crono (w/o CJF),
taking into account which is the favorite is a good thing, though you
shouldn’t base your whole prediction on it. However, there isn’t a
Mario/Crono in this contest that I can see thus far, not even FFVII vs.
OoT (in which I see FFVII having a significant advantage).
Let's
say that there is a game where you roll a dice, and you must bet on
either "6" or "anything but 6". Let's suppose we play once, and get a
6. Are you really going to say that the "anything but 6" choice wasn't
the best one? This is just an example btw, I know it isn't that
clear-cut in the Mario/Crono case.
No, it’s not as clear-cut
in the Mario/Crono case – I’d go so far as to say that this example has
absolutely nothing to do with what we’re talking about, actually.
By the way, I picked Mario in this match, because I wasn't thinking of the contest in a game-theoretical way last year.
And
I’m not going to think that way this year because it’s a stupid idea
that won’t work. I’m going to go with the evidence. The evidence says
that FFVII will likely win, just as it implied Link’s victory last
year. Looking back, I can see that some factors, such as Kingdom
Hearts, weren’t taken into account that year. I’m going to try to
account for more variables this year and thus have a better bracket.
The purpose of predicting the bracket is to guess who will win, and I’m
not going to second-guess myself with a bunch of game-theory junk that
won’t work in the end.
Wait a minute...did Super Mario World
come packaged with the SNES? I think it did, when the system first came
out. In which case the sales factor isn't as important.
Yup. --- "Your fate is sealed, and none but yours." -- Auron Proud Supporter of Starcraft in the Spring 2004 Contest | Jump to Page: | | | | | | 7 | | | |
|