Robots, Pollsters & Gamblers: Why Biden is Almost Certain to Win


As I am sure you are all painfully aware at this point, on Tuesday, November 3rd the US polity will be voting in their 59th quadrennial presidential contest to either: Retain their 45th Commander in Chief, or, and almost certainly, elect their 46th. As someone who enjoys gambling on political events I’ve settled on a crude formula that has helped me win the last handful of elections. This involves heavily weighting 3 primary sources of voting information and then adding in some key elements that marginally sway it. So without getting bogged down with debates, or wild speculation, about voter suppression, rigged ballot boxes, gerrymandering, mail in ballots, Supreme Court rulings, and post-victory calamity, let's simply stick to the numbers we have available and put down some smart money (“Don’t ever say smart to me”).

This election is being characterized by bookmakers as one of, if not THE, biggest gambling event in the history of mankind. Thus we should review the statistics, probability, and information gathered as processed and presented by the professional nerds. This should allow us to aggregate the data into a rudimentary picture showing that, by my admittedly amateur estimation, Joe Biden has at least an 87% chance of victory. This should be a no-brainer bet for anyone with a few bucks in their hand.

The experts I think we should be listening to come in the form of:

  1. The mathematicians, programmers, and technologists behind cutting edge AI internet data gathering/analysis tools such as “Polly” (of Advanced Symbolics)

  2. Traditional pollsters in aggregate form (i.e. 538, RCP, & 338Canada)

  3. Las Vegas book makers and their odds

In addition, I will tack on a few contributing factors that point towards a likely Biden win (job approval numbers, Covid deaths, and early voting stats). Taken on the whole, I think we can safely say, and more importantly bet, that Joe Biden will win a solid victory on election night.

Classical Polling

To start we should look at the legacy polling data and what they are projecting. Heading into the final weekend of the campaign former VP Biden is widely seen to be enjoying a 9-point national lead and is comfortably leading in 4 out of the 10 battleground states (versus Trump's lead in just Ohio and marginally in Iowa). The remaining 4 states are statistical ties, but also have shown momentum in Biden’s favour over the last few weeks (Florida, Arizona, North Carolina, Georgia).

Yet, even while typing that last sentence, I can vividly hear the shrill, moronic, and dismissive wailings of ignorami that are unfamiliar with polling methodologies, statistical analysis, and data probability. They will loudly proclaim their ideologically driven disbelief with defiant statements like: “The polls were bullshit in 2016!!”, “Polls are intentionally biased to corral sheeple into voting differently”, and “Elites doing the polling aren't asking real people what they think!” All nonsense and all deeply unaware of the actual science/art of making political projections. And if you are sympathetic to any of those sentiments ask yourself this: “Why would political parties, corporations, entertainment insiders, and social scientists spend hundreds of millions of dollars a year tracking public perception if it was largely useless?” If your answer has anything to do with fake news, Koch money, Soros schemes, or Mossad then please stop reading and get bent.

To quickly address the 2016 Presidential polling discrepancies (which of the last 5 years was the only outlier and only if narrowly understood), I will use a poker analogy. Heading into the final hand of the 2016 campaign, Clinton had three Aces while Trump held an inside straight draw (but not an actual hand). After the votes were tallied Trump essentially hit his belly buster straight, on the river, while knowing full well the next card in the deck was another Ace (hence the preemptive charges of electoral fraud and his surprise at winning the election). An objective look at the national polling averages of 2016 reveal that they were very accurate and this was duly reflected in the popular vote numbers.

In the subsequent post-mortem on the projections, many polling agencies rightly adjusted their methodologies when collecting, analyzing, and soliciting data. Their updated measures included an intensified formula that weighted more heavily for education, cut a sharper distinction between urban and rural profiles, and also selected citizen samples from registered voting lists as opposed to random landline dialling. So it seems as though they have recognized the flaws from their past performance and have tweaked them accordingly to estimate closer results (thus the accuracy they had in predicting the 2018 Midterms).

So regardless of the protestations about 2016, we can quickly look at my three favourite aggregate polling sites and find a common theme:

  1. Real Clear Politics: Rolling poll averages show Biden with a 7.8 point lead, the “No Toss Up States” electoral map shows a 345 to 193 win for Biden, and their rolling average for betting odds show Biden with a 64.1% chance of winning.

  2. 538: Based on their polling and simulations, Biden wins 89 times out of 100, the electoral college goes 347 to 191 for Biden, and Biden also wins the popular vote 97 times out of 100 (with 53.3% going to Biden and 45.5% to Trump). Additionally their site polling puts Biden up with an 8.9% lead nationally to date.

  3. 338 Canada: Rolling poll averages show an 8.6% lead for Biden, an 84.3% chance of winning the electoral college, and a 97.2% chance of Biden winning the popular vote (To boot they project that the minimum Biden electoral college count is 279 and could reach as high as 374. So at bare minimum a very narrow win for the Biden ticket).

And in addition to these findings, Biden's October polling data looks much better than Hillary’s in 2016. Joe has either slightly increased or held his voters in the final stretch of the campaign in a manner that Clinton couldn't. On the final day of polling in 2016 Hillary was ahead nationally by an average of 3.2% whereas Biden is ahead by more than double that at 7.8%.

So say what you will about individual polls, or particular polling firms, but the aggregate sites are all strongly pointing towards a Biden victory. Now onto the robots.

The Emerging Robotic Eye

Since becoming aware of a company called Advanced Symbolics (AS), and their AI program “Polly” in 2017 (courtesy of Steve Paikin and his show “The Agenda”), I have followed their predictions and have seen amazingly accurate projections that factor into my personal betting equation. They have correctly predicted an impressive string of elections that include: The Brexit Vote, the 2018 Congressional Midterms, the 2018 Ontario Provincial Election, and the 2019 Canadian Federal Election.

Polly is the name given to AS’s artificial intelligence program that scours social media sites and gathers information “to understand how people’s online behaviour and connections predict their real-world actions.” AS uses a patented “Conditional Independence Coupling” program mixed with a “unique qualitative methodology, ‘Digiographics™’, to identify a randomized, controlled sample of a representative target audience.”

Whereas legacy surveys rely heavily on small groups and are often tainted by response bias, Polly is released into the ether and gathers everything from comments, posts, and popular sentiment anonymously. After gathering data from a much larger pool of representative individuals than traditional polling agencies, AS is able to make significantly more accurate predictions on electoral outcomes. Traditional pollsters will lean on a few thousand respondents to create their study, Polly looks at hundreds of thousands of people that more accurately reflect the makeup of the voting public in terms of age, sex, education, race, class, region, etc.

As currently stands, Polly is giving President Trump less than an 8% chance of winning this time around and that his only salvation may come in the form of a Coronavirus vaccine that is widely believed to be at least months (if not years) away from realization. There is simply not much more to say about Polly’s predictions since large parts of it’s processes and algorithms are heavily protected by patents. So the best I can offer is to put forth that they accurately (and within reasonable margins of error) predicted the biggest elections of the last 5 years including Brexit which was largely missed by traditional polling companies.

All hail our eventual computational overlords! But before they eventually coerce us into the robotic collective you should plunk down some fiat currency on it's recommendations.

The Vegas Factor

The final major contributor to my ‘aggregate of the aggregates’ method of political betting is observing and respecting the Vegas gambling odds on elections. As a general betting strategy, and as well as observed wisdom, the favorite’s odds are more closely pegged to reality in almost any gambling situation.

According to aggregated betting odds, Joe Biden has been the favourite to win the election since June 2nd, with his probability rising ever higher since the first Presidential debate. As of writing this article the combined knowledge of Vegas bookies has set the likelihood of a Biden victory at 64.1% against Trumps 34.9%. And even when considering bookies that are offering the best odds for a Trump win (currently with Bovada and Vbet) you are still looking at a mere 36% chance of Trump pulling it off.

Yes of course, bookies largely lost money in their pegging of the 2016 race, but they seem to be more closely hedging their stacks in light of the low-probability victory Trump managed last time. In November of 2016, Trump was given an average of about a 20% chance of winning as opposed to the 35% chance this time around. Like my analogy earlier, he pulled out an improbable win with exceptional vote concentration in key districts within the swing states. This is something that seems to be severely lacking this time around with Biden holding the favourite position in 4/10 battleground states (a quick look at state-by-state odds from Bodog tell the story). Trump hit the inside straight draw, on the last hand, with the last card. But have you ever seen anyone do that twice in a row? Even if he did, I feel that Biden has the nut flush and is check-raising his way to a bigger pot.

All bookies considered, Biden is looking like a solid bet and the individual state-by-state lines back this up completely.

Other Significant Considerations

Aside from the aforementioned big 3 slices of my betting formula, here are some additional elements that I feel are working against Trump’s chances of reelection (Please keep in mind that these assertions are based on stats, but the conclusions I point towards are anecdotal):

  • 94% of US Covid deaths come from people aged 50 or older (199,822 deaths) and Trump's support from people in this demographic was about 8-9% points higher than Hillary Clinton in 2016. Which would seem to point towards the idea that at least 106,000 less Trump voters are alive today. This is especially noteworthy because the 2016 Trump win was decided by about the same amount of likely dead Republicans now (especially since Republicans are much less likely to take the virus seriously and thus more likely to die from the disease). This is further compounded by Trump’s seeming mismanagement of the pandemic and that an average of 61.5% of Americans think the country is on the wrong track.

  • Trump's job approval numbers are not great. He’s currently sitting at an average of 44.9% approval and 53.4% disapproval with the additional note that he hasn’t seen net positive approval numbers since January of 2017. Historically, presidents below a 45% approval rating do not get reelected (and in Trump’s case, he hasn’t consistently earned above 45% since Feb-May of this year i.e. before Coronavirus was more fully understood)

  • Nearly 69 Million people have already voted early or have mailed in ballots, and this seems to be a largely Democrat phenomenon. As the Washington Post has reported “Of the roughly 3.5 million voters who have cast ballots in the six states that provide partisan breakdowns, registered Democrats outnumber Republicans by roughly 2 to 1 (for Florida, Iowa, Maine, Kentucky, North Carolina and Pennsylvania). Again, anecdotally, Trump was dissuading voters from sending mail in ballots and, because of the Covid influence and behaviour patterns, I’m assuming more Democrats are opting to vote remotely and stay away from the large gatherings at polling locations. Advantage Biden since the hardest part of political activism is getting people to actually go out and vote.

  • There were about 9 million voters who cast ballots for Obama in 2012 who then voted Trump in 2016 (a solid 14.5% of Trump’s popular vote in 2016). So the question really is: “Are a majority of these swing/independents/centrists voters happy with the last 4 years of Trump?” I think not. This is because I find it hard to imagine that a voter who looked at Obama’s first term (2008-2012) with his emphasis on healthcare and using government to re-stabilize the markets (after the 08’ crash), on top of having to contend with the raucous Tea Party rise (2010 onward), would look at Trump’s first term and approve. I think a hatred for Hillary, a desire for change after 8 years of Democrat rule, and Trump’s willingness to speak openly about any controversial subjects (immigration, terrorism, military adventures, and China) largely spurred their 2016 Donald vote. A decision I do not think they are very proud of because they now would like a return to normalcy.

My Predictions

With that all said, I think I can sum up the eventual results on Tuesday night's impending election with two concepts: Coronavirus failure and Trump Fatigue Syndrome. In an alternate reality where Trump didn’t have to contend with a semi-lethal virus, and was about 20% more diplomatic in his approach to politics, he would easily be riding a wave of support into his second term. Despite the fact that Covid was not The Donald’s fault I believe he will ultimately be punished for the state affairs we find ourselves in. A thought I can’t seem to shake is that people who dislike him will be especially motivated to vote against him (which is a demographic I believe has grown significantly over the last 5 months of demonstrations, lockdowns, and existential risk to the Republic), because of his perceived failure on the pandemic front.

If you are prone to like him you’re probably wondering why he doesn’t force the economy back to normalcy, and, if you dislike him, put a disproportionate amount of blame on him for the quarter million dead Americans. But aside from Coronavirus I think that people are just tired. Tired of seeing his face, tired of hearing about the daily absurdities, and tired of having him interjected into every aspect of American life. This is a referendum on Trump and despite his unimaginable win in 2016 this will be a short chapter in US history that is quickly replaced by the status quo. Trumpistan will be relegated to the golden dustbin of populist history

So with all this in mind here are my predictions that I will be wagering heavily on:

  1. Biden wins 350-187 on the electoral college

  2. Biden wins the popular vote with 69.1 Million votes

  3. Biden wins the Battlegrounds of Wisconsin, Pennsylvania, North Carolina, Nevada, Michigan, Georgia and Florida

Regardless of who wins please be civil, be safe, and pay up if you lose.

Cheers to everyone winning bigly on election night!