November 10, 2020 - Who to believe about the US election results and the allegations of fraud? Meet Oraclum, the Croatian pollster who correctly called Trump in 2016 and a narrow Biden win in 2020, including the chaos which followed.
I learned a LOT about American geography last week.
Give me a map of America and I could probably have placed about 15 states in their correct place. After last week, I know exactly where the Blue Wall is, the neighbours of Nevada and Arizona, and Georgia is on my mind.
I also learned a lot about the US media and the US on social media. Moving between Fox News, CNN and Breitbart was as entertaining as it was scary - were we all watching the same election?
Social media was even more brutal. The conspiracy theorists were out in force, the accusations, the hate, the total division of a nation on full display. Any analysis seemingly had an agenda. In such a polarising vortex of fake news, which analysis to trust?
And then a post from Croatian brainbox Vuk Vukovic came into my Facebook feed. Aha!
Vukovic is one of the three co-founders of Oraclum Intelligence Systems. As the Oraclum website says, Oraclum is a data science company that builds prediction models. We created a disruptive innovation that will be the future of market research, and we used it to successfully predict both Brexit and Trump in 2016.
Yep, that's right - while every other poll was predicting a Clinton victory, this small Croatian startup correctly called Trump four years ago. And also, perhaps even more impressively, Oraclum also predicted the correct Brexit vote. How did Vuk and his colleagues do it? He explained a little in this 2016 TCN interview.
So how did Oraclum call the 2020 race? A narrow win for Biden. This is from their pre-election report. Interestingly, this year they decided to charge for the report to see if there was a market for such accurate information in these volatile times. And it seems that there is a very nice market indeed. You can read the pre-election report here.
If the results stay this way – a closer than expected Biden victory – we are looking at a high probability of post-electoral uncertainty and a contested election scenario. In fact, 64% of our users anticipate a contested election. The reason is simple: if Trump holds on to wins in FL and AZ (with NC also borderline), this will be known already on Election Night which means waiting for vote counts in PA and WI to confirm the winner of the race. In both of these states Biden is in front (and has been continuously since the start of our polling), however it will take time before this is confirmed which could result in continued market uncertainties next week.
If the situation remains as it is under our most likely prediction, where Trump wins FL and AZ, in addition to OH, GA, TX, IA and all the other states going as expected then there will be no clear winner on Election Night, and the results will depend on late vote counting in PA and WI.
If Biden however wins FL and AZ, while keeping all the others based on our projections, then he will have 270 electoral college votes already on Election Night (accounting for all the states going in the expected direction except for swing states).
The scenario with a higher likelihood after this week’s polling numbers is the first scenario, where the country will be engulfed in weeks of post-electoral uncertainty over who the winner is while votes are being counted in PA and WI.
Currently our method is predicting that Biden is leading in both of these states. Were this a normal election this result would be obvious already on Election Night and there would be much less postelectoral uncertainty. But given that this election depends on mail-in ballots, where half of the country already voted this way, the significance of PA will be even bigger than it usually is.
To conclude, Biden still has a much clearer path to victory, and if keeps PA, even if he loses FL and AZ (and even NC), he will come on top. But this might not be obvious on Election Night.
Oraclum also published a post-election article 5 days ago, which is a very interesting read - Post-election uncertainty: exactly as foretold.
Oraclum was not alone in calling the Biden win, as it had been with Trump in 2016, but the closeness of its predictions compared to other pollsters - both in electoral votes (294 v 244) and percentage of votes in crucial states, was impressive. Polling error in some states was 7-8%, while Oraclum got MI, WI and PA within 1%.
"For us the most important thing was to guess Florida," said Vuk, "as this was a big signal of whether the outcome would be known already on election night. We were clear that Trump is winning FL, but we never saw our method delivering PA, WI or MI to Trump. That gave us the confidence to call the race for Biden. Despite the fact that Trump won both OH and FL - usually two crucial states."
I asked Vuk about all the allegations of electoral fraud:
"I don’t think so. Any campaign has a right to ask for a recount, but with the entire world watching the process, I highly doubt that there was any fraud involved.
"After all, in many nondemocratic countries when people feel that there was election fraud, they spontaneously challenge this on the streets. This did not happen in the US. Quite the contrary actually.
"Plus there is no evidence of fraud, just like there was no evidence of fraud back in 2016. Trump even had an inquiry into this which found nothing.
"The confusion were mail-in ballots. There were simply too many of them this time and some states were not allowed counting before the election. That’s why we had to wait until Saturday - also something that we anticipated (see our report).
"As for the polling error — we explained this 2 months ago. It’s not the Shy Trump voter that’s the problem, it’s the non-response bias. And you cannot fix this by adjusting models (e.g. towards lower-educated voters) which are wrong to begin with."
Fascinating stuff, and congratulations to the Oraclum team. If you need some prediction services, you can find out more about what Oraclum offers on the official website.
