topbanner_forum
  *

avatar image

Welcome, Guest. Please login or register.
Did you miss your activation email?

Login with username, password and session length
  • Thursday March 28, 2024, 10:17 am
  • Proudly celebrating 15+ years online.
  • Donate now to become a lifetime supporting member of the site and get a non-expiring license key for all of our programs.
  • donate

Author Topic: Mouser's Election Predictor - A Memory from 2004  (Read 18473 times)

mouser

  • First Author
  • Administrator
  • Joined in 2005
  • *****
  • Posts: 40,896
    • View Profile
    • Mouser's Software Zone on DonationCoder.com
    • Read more about this member.
    • Donate to Member
Mouser's Election Predictor - A Memory from 2004
« on: November 05, 2012, 11:41 PM »
On the eve of the 2012 election I thought I would bring up a memory of mine.

This application precedes the creation of DonationCoder by a few months: Released in 2004 was my "Mouser's Election Predictor".

"MEP is a program for estimating the likelihood that a given candidate will win the United States Presidential Election based on estimates of how each state is leaning, and taking into account the electoral college all-or-nothing voting system.  It lets you fill in a table of how likely each state is to vote for a given candidate, and based on these numbers, calculates the odds that the different candidates will win. It can also use poll results to estimate probabilities.  It can be used on election night as states results come in to constantly give you an updated estimate of your candidates chances."

Read the online help file from 2004: here.

Mep1.png
evbiases2.gif






Nowadays there are great online sites that run essentially the same Monte Carlo electoral college simulations and show results online, e.g. Nate Silver's 538 Blog.

mouser

  • First Author
  • Administrator
  • Joined in 2005
  • *****
  • Posts: 40,896
    • View Profile
    • Mouser's Software Zone on DonationCoder.com
    • Read more about this member.
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #1 on: November 05, 2012, 11:47 PM »
Just wanted to share a bit of history :)

TaoPhoenix

  • Supporting Member
  • Joined in 2011
  • **
  • Posts: 4,642
    • View Profile
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #2 on: November 05, 2012, 11:48 PM »
Heh Mouser you don't want to release an Obama Romney version just for the giggles? :)

Nate Silver is almost as smart as you. (/Brownnose)  ;)

Plus you could make a new category - NANE - New Apps for the New Election!
« Last Edit: November 05, 2012, 11:56 PM by TaoPhoenix »

Renegade

  • Charter Member
  • Joined in 2005
  • ***
  • Posts: 13,288
  • Tell me something you don't know...
    • View Profile
    • Renegade Minds
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #3 on: November 06, 2012, 01:14 AM »
Heh Mouser you don't want to release an Obama Romney version just for the giggles? :)

I was going to ask the same question, but I thought the race was between Goldman Sachs Red Team and Goldman Sachs Blue Team. ;D hehehehe  :P
Slow Down Music - Where I commit thought crimes...

Freedom is the right to be wrong, not the right to do wrong. - John Diefenbaker

Renegade

  • Charter Member
  • Joined in 2005
  • ***
  • Posts: 13,288
  • Tell me something you don't know...
    • View Profile
    • Renegade Minds
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #4 on: November 06, 2012, 03:09 AM »
Nate Silver is almost as smart as you. (/Brownnose)  ;)

Found it! I saw an article that went on a bit about Nate Silver's methodology and another methodology:

http://www.globalres...recast-model/5310583

Based on the LV polls, Obama has a 98% probability of winning the Electoral Vote , not 80% as Nate Silver would have you believe. His model is obviously contaminated by extraneous variables. Nate tries to impress with 100,000 simulations. In reality only 500 are necessary to determine the EV win probability.

It goes on a bit about methodology and stuff.

What would be interesting is to see models that account for voting fraud based on historical tendencies. My guess is that it would simply be impossible though, with it basically being stochastic, etc. etc.

@mouser - Did your software predict the right outcome? (Accounting for fraud or for the popular vote?)


Slow Down Music - Where I commit thought crimes...

Freedom is the right to be wrong, not the right to do wrong. - John Diefenbaker

TaoPhoenix

  • Supporting Member
  • Joined in 2011
  • **
  • Posts: 4,642
    • View Profile
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #5 on: November 06, 2012, 10:08 AM »
Heh Mouser you don't want to release an Obama Romney version just for the giggles? :)

I was going to ask the same question, but I thought the race was between Goldman Sachs Red Team and Goldman Sachs Blue Team. ;D hehehehe  :P

(Satire) Come on, there are other banks out there, it could be at least between two bank consortiums! Goldman Sachs and Citibank on one side, Bank of America and Chase on the other side! Or we could cast it as RIAA + MPAA on one side and your choice of the other side! (/Satire)


TaoPhoenix

  • Supporting Member
  • Joined in 2011
  • **
  • Posts: 4,642
    • View Profile
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #6 on: November 06, 2012, 10:14 AM »
Nate Silver is almost as smart as you. (/Brownnose)  ;)

Found it! I saw an article that went on a bit about Nate Silver's methodology and another methodology:

http://www.globalres...recast-model/5310583

Based on the LV polls, Obama has a 98% probability of winning the Electoral Vote , not 80% as Nate Silver would have you believe. His model is obviously contaminated by extraneous variables. Nate tries to impress with 100,000 simulations. In reality only 500 are necessary to determine the EV win probability.

It goes on a bit about methodology and stuff.

What would be interesting is to see models that account for voting fraud based on historical tendencies. My guess is that it would simply be impossible though, with it basically being stochastic, etc. etc.

@mouser - Did your software predict the right outcome? (Accounting for fraud or for the popular vote?)

Careful Renegade. As much fun as you are, something feels wrong about that site. It tries to paint Obama winning by 98%...
"Based on the LV (Likely Voter) polls, Obama has a 98% probability of winning the Electoral Vote..."

That just feels wrong. That kind of percentage is the Give Up and Go Home level, and that just feels way off. I'm a little out of my depth on spiraling statistics, but that article is too simple to believe completely. "Everyone" agrees there are some 5-7 swing states, so if everything else is a wash, the percentages in those swing states are absolutely not 98% for Obama. So then the rest of that article goes down the tubes.


Renegade

  • Charter Member
  • Joined in 2005
  • ***
  • Posts: 13,288
  • Tell me something you don't know...
    • View Profile
    • Renegade Minds
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #7 on: November 06, 2012, 10:57 AM »
(Satire) Come on, there are other banks out there, it could be at least between two bank consortiums! Goldman Sachs and Citibank on one side, Bank of America and Chase on the other side! Or we could cast it as RIAA + MPAA on one side and your choice of the other side! (/Satire)

Meh, they're all the same. :P

Careful Renegade. As much fun as you are, something feels wrong about that site. It tries to paint Obama winning by 98%...
"Based on the LV (Likely Voter) polls, Obama has a 98% probability of winning the Electoral Vote..."

That just feels wrong. That kind of percentage is the Give Up and Go Home level, and that just feels way off. I'm a little out of my depth on spiraling statistics, but that article is too simple to believe completely. "Everyone" agrees there are some 5-7 swing states, so if everything else is a wash, the percentages in those swing states are absolutely not 98% for Obama. So then the rest of that article goes down the tubes.

Not really. It's simply a prediction from polls. The prediction isn't a function of public opinion. It's just a mathematical model.

But really, I just don't care all that much. The Americans are hosed period. If they get Obomney, they're hosed. If they get Robama, they're hosed. I could care less, but not much. What I *do* care about is the poor Americans that will be hosed. I feel sorry for them. But not as sorry as I feel for Canadians that have a... <Renegade is no longer capable of speaking civily>... ;)
Slow Down Music - Where I commit thought crimes...

Freedom is the right to be wrong, not the right to do wrong. - John Diefenbaker

tomos

  • Charter Member
  • Joined in 2006
  • ***
  • Posts: 11,959
    • View Profile
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #8 on: November 08, 2012, 03:44 AM »
Not something I know much about, but Nate Silver was giving a 90.9% chance of winning in the end and apparently he predicted 49 of 50 states correctly.
Nate Silver correctly predicts 49 out of 50 states (timesofindia)
& his blog is here
http://fivethirtyeig...t.blogs.nytimes.com/


ironically mouser's PopupWisdom had this to say just now:
Ninety percent of everything is crap.
-- Theodore Sturgeon
Tom
« Last Edit: November 08, 2012, 05:07 AM by tomos, Reason: minor »

mouser

  • First Author
  • Administrator
  • Joined in 2005
  • *****
  • Posts: 40,896
    • View Profile
    • Mouser's Software Zone on DonationCoder.com
    • Read more about this member.
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #9 on: November 08, 2012, 04:56 AM »
A quick comment about Nate Silver's work, and similar work.

The trivial thing that he does, which my Election Predictor did, is Monte Carlo simulation to predict the presidential election winner by taking the odds of winning each state and then running thousands of simulations to see how those state odds affect the overall winning odds (This is what my World Cup Predictor does also).

That's not a very impressive thing to do -- it's fairly straightforward and simple -- so you won't be surprised to find out that the pundits on tv are oblivious to this and prefer to just "go with their gut" when making predictions.

The *SIGNIFICANT* and non-trivial part of what Nate Silver does is combine lots of ever-changing state-wide polling data in smart ways and properly weight this polling data.  That's the hard part and that seems to be one area where Nate Silver knows what he is doing -- and that's the part that requires real domain expertise and good judgement.

TaoPhoenix

  • Supporting Member
  • Joined in 2011
  • **
  • Posts: 4,642
    • View Profile
    • Donate to Member
Re: Mouser's Election Predictor - A Memory from 2004
« Reply #10 on: November 08, 2012, 05:33 AM »
Running with a hint from Mr. S's own site, he wants to predict ... what will actually happen. However in the last couple of elections the two sides with a slant on the Repubs, are making statements that they want to *alter* the predictions *with*!

(Satire)
Mouser has got zero of the vote in Alaska right now. So Mouser needs to work extra hard in those other states. Remember folks, that's *Mouser* that needs to win in *Conneticut* if he has any chance to win! Don't forget, *Mouser* has been showing strong in the midwest, because he lives out there and knows the area well. People like *Mouser* out in the midwest!

(This message was contributed by Tao for Mouser for the role of Zaphod Beeblebrox in any future production of Hitchhiker's Guide to the Galaxy and was not endorsed by Mouser. )