Simulating the US Election

I attach an Excel file which simulates the US election.
The methodology is as follows.
From the FT Poll Tracker I accept their figure for Biden's "in the bag" Electoral College votes (EC). I also take from that source the states which are leaning toward Biden and those which are a toss up. These are my selected swing states. The rest seem to be in the bag for Trump.
I then use Betfair to get the odds of a Biden win in these swing states, this gives me the central scenario (Scenario 2).
I then guesstimate two alternative scenarios Scenario 1 is a bias towards Trump and Scenario 3 is a bias towards Biden.
I simulate 100 runnings of the election. I first simulate which scenario comes up. There is an input parameter to determine the likelihood of the scenarios. The version attached is for a 30% chance of each of Scenarios (1) and (3) with Scenario (2) occurring 40% of times.
The output should be self explanatory. (The simulations can be repeated by pressing Fn key 9.)

It points to me that Betfair's 66% chance of a Biden win is significantly understated, this is echoed by the link produced by @EmmDee and I find it very hard to explain the Betfair odds on the EC tranches.

Duke of Marmalade said:

I attach an Excel file which simulates the US election.
The methodology is as follows.
From the FT Poll Tracker I accept their figure for Biden's "in the bag" Electoral College votes (EC). I also take from that source the states which are leaning toward Biden and those which are a toss up. These are my selected swing states. The rest seem to be in the bag for Trump.
I then use Betfair to get the odds of a Biden win in these swing states, this gives me the central scenario (Scenario 2).
I then guesstimate two alternative scenarios Scenario 1 is a bias towards Trump and Scenario 3 is a bias towards Biden.
I simulate 100 runnings of the election. I first simulate which scenario comes up. There is an input parameter to determine the likelihood of the scenarios. The version attached is for a 30% chance of each of Scenarios (1) and (3) with Scenario (2) occurring 40% of times.
The output should be self explanatory. (The simulations can be repeated by pressing Fn key 9.)

It points to me that Betfair's 66% chance of a Biden win is significantly understated, this is echoed by the link produced by @EmmDee and I find it very hard to explain the Betfair odds on the EC tranches.

Click to expand...

Betfair odds aren't calculated probabilities. There is probably a fair amount of money on trump which is betting the models are wrong... Which explains the difference.

But if you take the view the models have been improved and the polls more accurate then Biden is a good price on Betfair.

Just FYI... 538 had trump as a 1 in 3 chance last time around. So the result wasn't "off the charts" in their models. It's a 1 in 9 or less this time for a number of reasons. But a lot to do with reduced undecideds in the polls

I have a somewhat less sophisticated way of trying to gauge the outcome. Basically gut instinct.
As much as I would like to see Trump trounced, I am wholly underwhelmed at this Biden / Harris ticket.
Trump, in no small part, got elected because many Democrats are disillusioned with the political system and could not bring themselves to vote for an establishment figure like Clinton, or worse, wanted to give the establishment a slap in the face and voted Trump.
Now they have had four years of the most debased performance by any President, election to the WH should be cake-walk for the Democrats. All they needed to do was to find a younger candidate that could articulate a half-reasonable policy platform with a little bit of charisma and job done.
Instead, they spent 4 yrs chasing after Russian bogey-men on FB instead of building base support. They have propelled an ageing, establishment figure who, as far as I can see, has offered absolutely nothing to the democratic base that didn't vote last time, or voted Trump.
So with one week to go, while Biden is polling ahead, there is still a significant doubt if he will get over the line. This is a pathetic situation to be in against the most unpresidential candidate of all time.
My gut instinct is that Trump will win again.

Bear in mind that Trump was ahead in the polls all the time until the Covid crisis. His handling of Covid and the slump in the economy as a result have flipped the polls.
Only in the last few weeks is rural America being hit by Covid.

Covid has had one huge positive this year so far; reduced air travel.
It may now oust Trump.
With Trump gone, Boris will have lost an anti-EU ally and may choose a Brexit deal with the EU.

I know several fairly intelligent and influential Americans who work in Ireland. All of them couldn't vote for Trump if they were paid, but they inform me their parents voted for Trump last time out and probably will do so this time. For some reason four years ago Trump was deemed to be giving Americans "Hope" but you can fool Americans once! They are not dumb.

The television coverage of the election in Ireland is pretty braindead with gun-waving redknecks shouting "Vote Trump" merely because they say so. I think the Black Vote will change in Biden's favour this time round along with several other voting factions.

So, I'm stocking up on the crisps and Seven-Up and chocolate for the all-night television coverage next week. I think Biden will win by a greater margin than most think.

Drakon said:

With Trump gone, Boris will have lost an anti-EU ally and may choose a Brexit deal with the EU.

Click to expand...

Yep, he will need to cut a FTA with an aged Irish papist.
I also think (hope) Biden by a wider margin than expected. But still as high as 1.53 on Betfair, clearly some Wolfie money at play their.

Drakon said:

Bear in mind that Trump was ahead in the polls all the time until the Covid crisis. His handling of Covid and the slump in the economy as a result have flipped the polls.

Click to expand...

It's slightly complicated by the fact that the Democratic nominee wasn't selected before Covis hit. But at the start of the year Biden (or a generic Democratic nominee) had a 5% lead roughly. Though that is a national average and whether that would have meant a presedential win is unknown.

Covid resposne has certainly impacted Trump and potentially with an important demographic.

WolfeTone said:

I have a somewhat less sophisticated way of trying to gauge the outcome. Basically gut instinct.
As much as I would like to see Trump trounced, I am wholly underwhelmed at this Biden / Harris ticket.
Trump, in no small part, got elected because many Democrats are disillusioned with the political system and could not bring themselves to vote for an establishment figure like Clinton, or worse, wanted to give the establishment a slap in the face and voted Trump.
Now they have had four years of the most debased performance by any President, election to the WH should be cake-walk for the Democrats. All they needed to do was to find a younger candidate that could articulate a half-reasonable policy platform with a little bit of charisma and job done.
Instead, they spent 4 yrs chasing after Russian bogey-men on FB instead of building base support. They have propelled an ageing, establishment figure who, as far as I can see, has offered absolutely nothing to the democratic base that didn't vote last time, or voted Trump.
So with one week to go, while Biden is polling ahead, there is still a significant doubt if he will get over the line. This is a pathetic situation to be in against the most unpresidential candidate of all time.
My gut instinct is that Trump will win again.

Click to expand...

Your gut instinct may be a great predictor but it certainly doesn't rely on data

The problem with most of the younger / alternative candidates were that they were either unelectable in large numbers of key states or if they were electable, didn't have the weight of organisation to get out there. Bernie and the other "radical" candidates were fundamentally undone by black electors in the Carolinas - and if they couldn't bring them along, they had no hope in a general. Biden proved himself capable of bringing back the traditional vote which was a large reason Clinton lost.

I don't think there is significant doubt the Biden will win. There is currently an 85% chance he will win the electoral college. However, even with an adjustment for state-level poll errors similar to 2016, Biden's probablility remains over 75% - unless something structural happens (i.e. Trump actually persuading somebody with his "stop counting votes after 3rd" message - which ironically would disqualify his own vote).

The Betfair odds don't reflect the data - I think there is a large amount of "gut-instinct" money at play

EmmDee said:

Your gut instinct may be a great predictor but it certainly doesn't rely on data

Click to expand...

True, and I'm not going to dispute the detail you extrapolate from data. Sitting 7,000km away all I can is observe, and from this park bench it's a shambles.

EmmDee said:

The Betfair odds don't reflect the data - I think there is a large amount of "gut-instinct" money at play

Click to expand...

There is easy pickings so?

WolfeTone said:

There is easy pickings so?

Click to expand...

I don't think there is ever a "sure thing" in betting (or investing for that matter). But I think it's fair to say there is a mis-match between the data and modelling and the odds on Betfair. Depending on your view on what some of the modellers are doing (and it would be an idea to understand something about their approach), there is certainly a potential for a "value bet" even at these odds.

If you wanted somethign a bit more speculative - I'd keep an eye on the Betfair markets as the initial tallies come in. The "on the day" votes are counted first on the night. The early and postal votes are counted in the days after. If the markets moved because of some early vote reporting, it might also be an opportunity

Duke of Marmalade said:

Yep, he will need to cut a FTA with an aged Irish papist.

Click to expand...

And there is the Democrats refusal to see the GFA/BA tampered with.

Leper said:

For some reason four years ago Trump was deemed to be giving Americans "Hope" but you can fool Americans once! They are not dumb.

Click to expand...

“Hope” was Obama. “Great” is Trump’s offering.

Leper said:

The television coverage of the election in Ireland is pretty braindead with gun-waving redknecks shouting "Vote Trump" merely because they say so.

Click to expand...

It always is. It’s head wrecking.
Do you remember 2016? Matt Cooper took his entire Today FM Last Word team over to New York on a junket and did a live broadcast from New York to celebrate Hilary Clinton’s historic win...
The Irish media love to focus on Biden’s Irishness. And it is true, he is proud of his Irish roots.
Mike Pence has closer family ties and worked in a bar in Clare when he was a student.

Leper said:

So, I'm stocking up on the crisps and Seven-Up and chocolate for the all-night television coverage next week. I think Biden will win by a greater margin than most think.

Click to expand...

Previously the result was known the night of the count. The results came in thick and fast. One late night and that was it.
I’d say it’ll be the weekend before we’ll know this time. The results will come in slowly and in tranches. Usually the early votes and mailed votes don’t make a difference because the volumes are minuscule. But this time there’s been about 75 million already.

Duke of Marmalade said:

I attach an Excel file which simulates the US election.
The methodology is as follows.
From the FT Poll Tracker I accept their figure for Biden's "in the bag" Electoral College votes (EC). I also take from that source the states which are leaning toward Biden and those which are a toss up. These are my selected swing states. The rest seem to be in the bag for Trump.
I then use Betfair to get the odds of a Biden win in these swing states, this gives me the central scenario (Scenario 2).
I then guesstimate two alternative scenarios Scenario 1 is a bias towards Trump and Scenario 3 is a bias towards Biden.
I simulate 100 runnings of the election. I first simulate which scenario comes up. There is an input parameter to determine the likelihood of the scenarios. The version attached is for a 30% chance of each of Scenarios (1) and (3) with Scenario (2) occurring 40% of times.
The output should be self explanatory. (The simulations can be repeated by pressing Fn key 9.)

It points to me that Betfair's 66% chance of a Biden win is significantly understated, this is echoed by the link produced by @EmmDee and I find it very hard to explain the Betfair odds on the EC tranches.

Click to expand...

In this simulation I put the chances of 300-329 Biden EC votes at 31% (favourite) compared to Betfair’s 11%; result 306
I also earlier touted the arbitrage Trump win Florida at evens, to lose overall also at evens; fairly easy money; was never losing Florida and winning overall.