Fantasy Football - Footballguys Forums
Welcome to Our Forums. Once you've registered and logged in, you're primed to talk football, among other topics, with the sharpest and most experienced fantasy players on the internet.
2020 Presidential Election Polling Thread (3 Viewers)
- Thread starter TripItUp
- Start date
BobbyLayne
Footballguy
Alright, so the same principle applies to Trump's probability, right? So at +160 his odds are 100/260 = 38.46%. At +130 it's 100/230 = 43.48%.Minus means you are the favorite. -160 means that a person has to bet $160 to win a profit of $100 ($260 total).Could explain this like I’m 5 years old. Like, I never gamble so I don’t know what it means.Vegas odds have moved over the weekend. Biden was -190 to Trump's +160 on Friday. Today it's Biden -160/Trump +130.
To convert to a percentage, take 160 and divide by (160 + 100). 160/260 = 61.5% chance of winning.
A few days ago, Biden was at -190, which translates to a 65.5% chance (190/290 = .655).
For reference, on Election Day in 2016, some betting sites had Hillary at -900, which translates to a 90% chance of winning.
Hmmm
So if you're betting favorites on the money line number, you can wipe out your profits easily ($190 or $160 in the Biden example.) Whereas if you bet an underdog, you're only risking $100. And if I'm understanding the post before yours, the vig means you have to win quite a bit more than you lose just to break even, regardless of which side you take.
More or less correct?
Leroy Hoard
Footballguy
Just an example of a logical mind refusing to accept an illogical outcome.Don't remind me.
Goodell's Alias
Footballguy
Close. If you consistently bet favorites on the money line, yeah, you have to win a whole lot more often than you lose (although you should be winning more often, seeing as you are betting on the favorites). If you are betting underdogs, you can lose more than half and still make money, since you lose $100 each time but win $130 or $160 or whatever each time you win.
Note that bets don't have to be these exact increments. For instance, you could bet $50 at either -160 or +130. You're risking $50 either way (i.e. if you lose, you pay the book $50). In the former case (-160), if you win, you would collect $31.25. In the latter case (+130), you would collect $65 if you win.
Leroy Hoard
Footballguy
Not meant to be the perfect strategy, but if you only bet on plus money underdogs you only have to be right 50% of the time to show an overall profit. You have to be realistic though in thinking that they have a better than even chance of winning.
BobbyLayne
Footballguy
Y'all make money doing this? Outside of Trump in AC, you don't read too many stories about casinos going bankrupt. House always wins, right?
Anyway, thank you for taking the time to explain it. At least I have some idea what the money line means now; for years I was baffled and never cared to understand what it means.
Goodell's Alias
Footballguy
Some people can consistently make money. Most people don't. Remember that "house always wins" means that the house wins in the aggregate, not versus each individual gambler.
Personally, I don't gamble much these days. Every once in a while, there's a line that attracts me. Mayweather over McGregor, for instance, seemed like easy money even at -350. Turns out, it was.
Leroy Hoard
Footballguy
AAABatteries
Footballguy
But he did say he was going to win 2-3 weeks ago. I have no idea if he'll win or not (I don't think he will) but one thing I do know is I don't believe anything the man says. He's a proven pathological liar who can't be trusted.
To stay on topic - I'm curious how these internal polls differ from "regular" polls. What is their poll potentially showing that others aren't?
caustic
Footballguy
There's been lots of discussion about the divide between state polls and national polls, the former showing a slightly tighter race than the latter. Another interesting wrinkle is House district polls. House polls flashed warning signs of Hillary's collapse with non-college whites in 2016, and they were the most accurate polls of 2018. Right now they happen to be Trump's worst ones. Take NY-22 for instance -- it's based in Binghamton/Utica and has lots of white working class voters:
2012: Obama and Romney essentially tie at 49%
2016: Siena poll shows Trump +14, Tenney (R) +4. Actual results are Trump +15, Tenney +5.
2018: Siena poll shows Brindisi (D) +1. Actual results are Brindisi +2.
2020: Siena poll shows Biden +1, Brindisi +9.
That's a 14-15 point shift from 2016, and NY-22 isn't the only district where such gargantuan shifts are taking place. Biden is simultaneously clawing back Obama/Trump voters in districts like NY-22 while surging with suburban folks. The state/national polls are bad for Republicans, but the district polls are apocalyptic.
2012: Obama and Romney essentially tie at 49%
2016: Siena poll shows Trump +14, Tenney (R) +4. Actual results are Trump +15, Tenney +5.
2018: Siena poll shows Brindisi (D) +1. Actual results are Brindisi +2.
2020: Siena poll shows Biden +1, Brindisi +9.
That's a 14-15 point shift from 2016, and NY-22 isn't the only district where such gargantuan shifts are taking place. Biden is simultaneously clawing back Obama/Trump voters in districts like NY-22 while surging with suburban folks. The state/national polls are bad for Republicans, but the district polls are apocalyptic.
Last edited by a moderator:
Terminalxylem
Footballguy
I don’t gamble either, but it seems like you could just subtract the vig (assuming that number is known) and perform the calculation as Joe Summer stipulated. For example, in the even odds scenario you mention, with 10% vig, the likelihood of winning is (110-10)/[(110-10) + 100] = 100/200 = 50%?
But that doesn’t explain why the opposing team’s number isn’t always complementary to the favored team, ie., how -160 for the favorite isn’t always matched with +140 (vs. +130), or whatever adds up to 100% using Joe’s calculation. I suppose there’s always the chance of a tie, or third party candidate winning in the election scenario.
Last edited by a moderator:
BobbyLayne
Footballguy
10% vig? Who runs this, Hesh Rabkin from The Sopranos?
Dr_Zaius
Footballguy
I think Nate deserves some credit for 2016, because I don't remember anyone giving Trump as good of a chance as 538 did, and I remember they were taking heat prior to the election from people arguing they were grossly overestimating Trump's chances.
That being said, I think the argument they give about not being wrong because they acknowledge it's not 100% is lame. I mean, if that's the standard they could just use a random number generator to predict the outcome - how can they ever be wrong? It's not like you can go back and run the election 100 times to see how close they were. They've seemingly missed pretty badly a few times (one of the recent British elections comes to mind), and I think it's self-serving to act like the actual outcome of the election has no bearing on the correctness of their model.
BobbyLayne
Footballguy
Polls like this and Georgia tell quite a tale. Trump has had to hit a bunch of states he steamrolled 4 years ago. He’s playing wack-a-mole while Biden is cruising along.
Gr00vus
Footballguy
They assigned probabilities to outcomes. I don't know how "wrong" becomes a metric for measuring that kind of effort. We've had this discussion a bit on this board before. 28% is a non-trivial probability. I have a hard time saying someone was wrong because something they assigned 28% probability of occurrence to actually happened. Also, in the end, they aren't a primary source of measurement - they're performing analysis on other peoples' data. If that original data isn't good, there's only so much secret sauce they can spread on it to make it better. I don't know what kind of crazy model they could have come up with that would have put Trump at 50% or better to win on election day. I think they'd be more suspect if they had ended up in such a place than where they actually ended up.
Last edited by a moderator:
Leroy Hoard
Footballguy
AAABatteries
Footballguy
Thunderlips
Footballguy
With voting already taking place across much of this country......are these polls asking people if they voted already and then including them; or is admitting to voting an automatic qualifier?
The Z Machine
Footballguy
I would think that if someone said they already cast their ballot, they would be counted in those surveys as a very likely voter.
Dr_Zaius
Footballguy
I actually like 538 a lot and think they do a really difficult task quite well, but I think people in general put too much weight in what they say without really gleaning a lot of the uncertainties. My main beefs:
- Any percentage prediction is non falsifiable over only one result. If I give you two dice and predict that you have a 20% chance of rolling an 8 or better, my model is wrong. It doesn't matter whether you roll a 10 or a 3, and it's silly to after you roll a 10 declare that I was right because I said it might happen. To your point, the only way to really check 538's work would be, for example, to group all of the races over the years that they declared a 70% chance of winning and see how often the favorites actually won. It still wouldn't tell you for certain they were right (as errors could cancel out), but if it was 68% (vs. 50%), that would be a big point in their favor. I agree with your statement on Trump - I don't see any way a reasonable model would have made him the favorite, but that gets to my next point...
- They are dependent upon polling, and I think that the uncertainties associated with this are understated. In the last week I've seen polls that have Biden +18 and Biden +3. One of those polls (at least) is just wrong. They do the best they can with weighting polls and looking at who was accurate in the past, but you can only do so much, and if a bunch of pollsters have the electorate modelled wrong in the same way, it's going to make 538's result off. Again, nothing they can really do about it, I just think it is underappreciated by many.
- Nate to his credit pointed this out in 2016, but in a national race the state poll errors to some extent are correlated. If you looked in 2016 and saw Clinton had a >85% chance of winning each of Pennsylvania, Wisconsin, and Michigan, the tendency is to treat them as independent outcomes and declare that the election is over because there's no way Trump "draws to an inside straight" over and over, to use one of the phrases I heard a lot. We see it repeatedly in elections that the swing states typically swing together (e.g. 2012 when Obama overperformed a lot of the state polling).
Juxtatarot
Footballguy
One of my biggest concerns is the electoral map looking like this shortly after election night with PA taking several drama filled days to count votes.
Anarchy99
Footballguy
Part of the problem in making a predicative election model is that while there is one election, the components involve 50+ states and a handful of districts. Like we found out last time, a few thousand votes in key states swung the election. In pretty much 40 states, there is almost always a clear winner. The polls for the toss up states usually fall within the margin for error. Places like 538 then have to figure out in close polling state if the mix of people polled is right, what the bias in the polling organization bias leans, what the total turnout will be, what bump an incumbent gets, voting history, etc. In some states, it could easily come down to getting a single county right in predicting the winner of that state.
As for running the PA / WI / MI gambit, last time Trump barely won all of them. Even a minor defection to Biden would be enough for those to go blue this cycle. In that regard, I am not sure I buy into that if someone wins one, he will win all three.
Anarchy99
Footballguy
I mentioned yesterday that RCP has most battleground states closer than 538, and others explained the different methodologies of how each site used or ignored polls. 538 has Biden +6.4 in PA (using the same polls as RCP). They also had Biden +7.3 a few days ago. Apparently they are reading the tea leaves differently.
ETA: It looks like RCP takes the simple average of the most recent 6 polls they have listed. In this case, the last 6 polling results in PA average out to +3.8.
Last edited by a moderator:
gianmarco
Footballguy
Well, unless I'm off, Biden should get both electoral votes from Maine, which puts him at 259. Then he only needs Arizona to get to 270. Right now, Arizona is 50.9% for Biden and 47.8% for Trump according to 538.
Yes, I know what you posted could happen, but I like the odds that it doesn't, especially with PA even more favored for Biden than Arizona.
PennStater77
Footballguy
Trump is going to win
Juxtatarot
Footballguy
Winning Arizona allows Biden to eek out 270 without winning PA in a map like this (he would have to win one of the contested NE or ME districts).
Anarchy99
Footballguy
Maine has a mostly statewide election for 2 electoral votes. But they also have elections in District 1 and District 2, each worth 1 electoral vote.
538 currently has Biden 90% to win the 2 electoral votes for the state, 98% to win District 1, and 51% to win District 2.
Is there any good swing state simulators out there where you can plug in an outcome and it spit out path to victory for each candidate? For example, if Biden wins Florida, how many other swing states does he need to win?
Nevermind, it’s already linked here.
Nevermind, it’s already linked here.
Last edited by a moderator:
Anarchy99
Footballguy
While it is not a simulator, you can start with this roadmap from 538. With PA in the Biden column, Biden is at 278. If you want to give Trump the 20 electoral votes for PA, that would push Biden back to 258. The states they have leaning Biden are:
NE2 - 1
AZ - 11
FL - 29
NC - 15
ME2 - 1
There are also few other toss ups in GA (16), OH (18), and IA (6).
As for your question, if Biden wins FL (and holds on to the states where he is heavily favored and slotted in his column), that would give him 307 electoral votes. So Biden could lose OH and FL and still have multiple ways to still win.
AAABatteries
Footballguy
I think at the end of the day the single biggest factor that will determine the election (outside maybe Covid) is that Biden isn't Hillary. We can argue/discuss all you want about whether it was fair or not and whether it was something nefarious (like misogyny) or not but the fact was that there was a lot of hatred of Hillary. She was not a good candidate to run against Trump. Biden, with his warts and all, is.
Last edited by a moderator:
AAABatteries
Footballguy
Biden has several paths to victory. Trump basically has to sweep FL/NC/OH/GA and then win 2 of 4 of PA/WI/MI/AZ*. The former wouldn't surprise me if he does it. I just don't see him doing the later.
*I think I have that right that he needs 2 of those 4 but I may be off on that.
caustic
Footballguy
In the RCP average, Clinton briefly hit 50% nationally in April 2016, but that was it. In the Fivethirtyeight average, Biden is currently leading 52.2% — 42.0%. Two weeks away from election day 2016, Clinton was leading 45.8% -- 39.7%. Joe is definitely in a stronger position.
Redwes25
Footballguy
Agree with this which is why I point out the over 50% point. It feels different if more than half the country likes him. Honestly, I haven't followed all the state polling closely but if Biden gets like 51% of vote and Trump wins electoral college it is going to get ugly. I really hope it doesn't happen.
Redwes25
Footballguy
When numbers are like that there is a lot of undecides out there. It just doesn't feel the same but who really knows until all votes are counted.
Drunken Cowboy
Footballguy
This was also near her best polling the entire election cycle. This is after Access Hollywood and before the Comey letter.
![[scooter]](/data/avatars/m/22/22287.jpg?1659477690){kind=avatar}
[scooter]
Footballguy
Hillary hit 50% in several individual polls, but her average would remain far below 50%. Looking at 538, I see 19 individual polls where Hillary hit 50% or more. By contrast, I just counted 250 national polls where Biden topped 50%. That bodes very well for Biden, but it doesn't mean much if Trump pulls off upsets in 3 states.
[scooter]
Footballguy
Four days ago:
Nine days ago:
That's progress, folks!
Chicago Hooligan
Footballguy
Unless it's some weird site no one's ever heard of there's basically no good poll result for Trump at this point. Fortunately our friends have their new thread to hang out in while they fantasize about a silent army of Trump voters carrying the day.
AAABatteries
Footballguy
Early votes don't necessarily mean mail-in, right? I'd love to see that breakdown (if it exists) because it's going to be a disaster if it takes weeks like some are predicting.
AAABatteries
Footballguy
Similar threads
