Predict the Election

[mr_genius]

If it finishes 269-269. Romney gets it. It'll be a close one, that's for sure.

 

[mr_genius]

But ask any actuary if they can with 100% accuracy predict when I die, and they would say no.

 

if i have an actuary predicting 70%-80% outcomes on a macro scale and is wrong consistently that's less than optimal. For example if I have an actuary predicting with 80% certainty that a combined loss ratio in a property/casualty cycle will be at 95% with their tables, and they miss at 105% every time... not good. But we still have random major catastrophic loses; is the tea party in 2010 Silver's cat loss? Very possible. I just need more wins from Mr.Silver before I give him his (rightful - earned) place as a statistical modeler.

 

anyways votes are predicted in the millions not 1 to 1. You are not predicting one persons vote, or one person to die (in your example). you are predicting the overall trend with a macro outcome which you can abstractly apply to that 1.

Edited October 27, 2012 by mr_genius

[mr_genius]

To be fair, Silver isn't the only election prediction site that has Obama as favorite to win, right now, if Romney were to win in Nov. it would be seen as an upset.

 

I'm getting kinda tired of the hyper politics going on today, I wish the elections were held tomorrow just so we could move on...

 

agreed

[knightni]

I'm not saying that with bias; It's true. The Republicans control the House and they decide an electoral tie.

[mr_genius]

I'm not saying that with bias; It's true. The Republicans control the House and they decide an electoral tie.

 

dude... we have the same prediction.

 

on page 2

 

 

http://www.270towin.com/2012_election_pred...s.php?mapid=SUY

 

and i have Biden as VP on page 1, per the Senate vote.

 

http://www.soxtalk.com/forums/index.php?s=...t&p=2719140

 

Edited October 26, 2012 by mr_genius

[2nd_city_saint787]

Ron Paul...by a landslide...shocks the world...he ends up being the Anti Christ...The Mayans were right.

[Soxbadger]

if i have an actuary predicting 70%-80% outcomes on a macro scale and is wrong consistently that's less than optimal. For example if I have an actuary predicting with 80% certainty that a combined loss ratio in a property/casualty cycle will be at 95% with their tables, and they miss at 105% every time... not good. But we still have random major catastrophic loses; is the tea party in 2010 Silver's cat loss? Very possible. I just need more wins from Mr.Silver before I give him his (rightful - earned) place as a statistical modeler.

 

anyways votes are predicted in the millions not 1 to 1. You are not predicting one persons vote, or one person to die (in your example). you are predicting the overall trend with a macro outcome which you can abstractly apply to that 1.

 

The part you seem to not understand is Im not talking about voters being 1 to 1, Im talking about the results. Silver has only predicted 1 presidential election. When you are talking about millions of people and their life expectancy, you have millions of deaths to base it on. Thus you can find trends and see if the model is working. When you have a model that can be tested only 1 time every 4 years it is almost impossible to tell if the model is working. Which is why I compared it to them predicting my death. Because they only have 1 shot at being right, not millions of deaths to be right within a certain tolerance.

 

More wins or losses require more games being played. When you work with actuaries you know that they need lots of results. There havent even been 100 elections, its just an impossibly small sample size with a ridiculous amount of variables. Once again, not talking about voters, talking about how many elections there are to try and gauge a model thats purpose is to predict the outcome of presidential elections.

[SouthsideDon48]

I played around with an electoral map yesterday and now I'm predicting Obama will win with 319 electora votes versus Romney's 219.

[mr_genius]

The part you seem to not understand is Im not talking about voters being 1 to 1, Im talking about the results. Silver has only predicted 1 presidential election. When you are talking about millions of people and their life expectancy, you have millions of deaths to base it on. Thus you can find trends and see if the model is working. When you have a model that can be tested only 1 time every 4 years it is almost impossible to tell if the model is working. Which is why I compared it to them predicting my death. Because they only have 1 shot at being right, not millions of deaths to be right within a certain tolerance.

 

More wins or losses require more games being played. When you work with actuaries you know that they need lots of results. There havent even been 100 elections, its just an impossibly small sample size with a ridiculous amount of variables. Once again, not talking about voters, talking about how many elections there are to try and gauge a model thats purpose is to predict the outcome of presidential elections.

 

bro stick with Matlock. you ain't no actuary or stats guy.

 

 

[Jake]

bro stick with Matlock. you ain't no actuary or stats guy.

 

[Soxbadger]

bro stick with Matlock. you ain't no actuary or stats guy.

 

Dont worry, I have felled far greater trees than you. Keep your head up, there just is no way you are beating me in an internet debate. Your best line is resorting to comparing me to a famous television show, which is pretty much a compliment because (I actually have never seen the show) I am pretty sure that Matlock was successful.

 

Maybe next time compare me to Lionel Hutz, that would have made more sense.

[Rex Kickass]

This is a close election, but it could end up a lot less close than we think. I'm suspicious of Likely Voter screens which is about every poll that we see. Not because they lean Republican per se, but because part of Obama's key to victory involves registering a lot of new first time voters. It's why I think Virginia will stay blue - and although I think North Carolina and Florida will flip Republican again this year, there's a lot of evidence to point out that either of those states could be very very close.

 

Ultimately I think Obama with 294 EV and a razor thin popular vote margin because of the huge numbers coming in from the South for Romney.

[SOXOBAMA]

If it finishes 269-269. Romney gets it. It'll be a close one, that's for sure.

 

If that happens, Joe Biden is Romney VP.

[mr_genius]

next time compare me to Lionel Hutz

 

What about the lawyer from Night Court?

[EvilMonkey]

I think whichever side wins it is going to be by a bigger margin than anyone predicted because people are lying in polls. There is so much attention to polls I can see people making crap up just for fun.

[Jake]

I think whichever side wins it is going to be by a bigger margin than anyone predicted because people are lying in polls. There is so much attention to polls I can see people making crap up just for fun.

 

It's all accounted for in the math

[greg775]

I predict Obama is a lock to win.

[clyons]

I'll predict a popular/electoral split, with Obama winning the latter by a whisper, but I honestly can see any number of scenarios.

[Brian]

I predict Obama is a lock to win.

 

Wish I felt as confident as you.

[kev211]

The election is going to come down to Ohio. If Obama wins it he is sitting pretty. If Romney wins it things will start to look bad for Obama, but he can still find 270 without it.

 

This is all based on Romney's ability to win Florida. If he can't do that then Obama most likely ends up with 300+.

 

I personally feel Romney wins Florida, Obama wins Ohio. Obama wins with 290ish EC votes.

[StrangeSox]

Obama declares a State of Emergency, postpones election indefinitely (channeling paranoid libs circa 2004)

[Jake]

Obama declares a State of Emergency, postpones election indefinitely (channeling paranoid libs circa 2004)

 

link?

[southsider2k5]

link?

 

I'm pretty sure that was a joke.

[StrangeSox]

I do not have a link for the paranoia a few on the left exhibited in the run-up to the 2004 election. Maybe check democraticunderground?

[Jake]

Too much homework for me. Would be a brilliant political move, potentially. He may get shafted in turnout if things are still s***ty on election day. The only trouble is that changing the day might hurt turnout even worse.