Why Analysts were Mistaken about Trump’s Election


The author of this article is a journalist, but also a forecasting models builder. I was one of the first and few people who forecasted Trump’s election since the beginning. But I am not writing to promote and boast myself, but to share why I understood correctly the elections results and how it is possible to create a forecasting model that really works.

We are not talking about anything connected with luck or magic, but a mix of behavioral science, psychology, and experience with economical and political flows.

The most relevant and recurring error committed by my colleagues is to succumb to personal opinions when preparing questions or selecting people for market research and the analysis of results. If a result is considered impossible or even unacceptable by the researcher, he will probably accumulate a long series of very micrometrical mistakes that, all together, can result in a completely wrong forecast. If you think “Trump cannot win this election”, besides you are a great professional and neutral observer, the mistakes mentioned are merely hibernating and ready to wake up.

For example, consider how impactful the choice of the “representative sample” is when you select people for interviews. Imagine you want to build a voting population sample in the USA. Trying to create this sample, you can analyze the statistical data of all the people who voted in the last four legislative periods and reproduce the same sample. This is the first mistake. In the case of Trump, he was voted for by a different mix of voters, probably people who never voted in the last four legislature, but then decided to go to vote for Trump. In this case, the chosen sample is completely wrong and doesn’t represent the real voting population. Another mistake is to think “Which program will people follow the most?” instead of saying “Which candidate is more closer to what people need and think?”. This is the biggest mistake: putting the observation point in the wrong place.

If any reader has visited USA in the last 20 years, they can can see what the reality of this country is and how people live outside the big cities. If you go to the bars, saloons, or fast food joints all around the country, or if you go in the suburbs and check how life really is like in the US, then you can immediately understand why Trump was the best candidate since the beginning. 

But, of course, Trump wouldn’t be the winner if the other side didn’t make a lot of mistakes. Those errors, put together, brought my forecast from the “possibility” level into the “reasonably certain” one. Among these mistakes by Clinton and co. were: tracking down a woman who decided only after 10 years to report sexual assault perpetrated by Trump; the statement to votes to the effect of “If you vote Trump, you have a low cultural level and you are crazy or stupid”; and the exaggerations that Trump enslaved Mexicans or blackmailed them when employing them illegally.

Clinton forgot to apply basic rules, but analysts didn’t adapt their analyses to the new situation,  instead simply perpetrating the previous model and sample, insisting on asking people about voting intentions. Consider this: if all the media says that you are stupid if you vote Trump, why should I tell another person my voting intention with the risk of being considered stupid? I will tell them “I don’t know yet” or even indicate Clinton as a choice, simply because the interviewer maybe prefers this socially-acceptable answer. 

Another of Clinton’s mistake was to forget her behavior when she was involved into the Lewinsky affair. United States citizens prefer a male chauvinist like Trump, more macho and publicly showing his sexism instead of a basic opportunistic choice like Hillary who absolved her husband, who lied to all the country under oath. Furthermore, many US citizens have hired illegal workers in their life, and some of them probably have friends who are illegal residents in USA or they themselves were one of them for a while. Immigrants are usually less open and friendly to other immigrants who arrive later after those who are already integrated and became US citizens. Thus, the stereotype of immigrants’, and indeed US citizens’ attitudes perpetrated by the Clinton campaign and analysts was also mistaken.

A good forecasting model starts from the demand of the market and the measures the offer compared to that demand.

Then, to be sure about the result, it starts from the offer and tries to understand if there is a demand. Before, it is important to create a parametric behavioral model where the parameters continuously change according to what is going on outside. Thus, a forecasting model can work only if there are multiple, principally two, observation points and the parametric monitoring process is applied every moment. I simply input people’s choices, what candidates offer, and what happens during the campaign is revealed. There was only one possible result: Donald J. Trump as the next and the best US president.


For objective and subjective reasons, Donald Trump is simply the perfect match for what the US wants and needs today. Seen from abroad, too, if we talk about all the terrible processes impacting our life (financial crisis, globalization, de-localization, taxation, sanctions), Trump is the only possible hope to change our future. A United States focusing on itself and cooperating with Russia means a more stable world and a better chance for Europe to exit from the lethal hand of the US and develop independently.