Monday, October 12, 2020

My Prediction For 2020 Election

The case for an epic landslide Democratic win

The 2020 Presidential election is only a few days away. Many questions are being discussed. Will there be a peaceful transfer of power if Biden wins? Will the polls be wrong again for predicting a Trump loss? And so on.


What is rarely mentioned is the possibility of an epic loss for Trump. Some believe that Teflon Trump will emerge victorious as he did in the 2016 primaries and eventual election. Some believe that his die hard supporters (or cult, as it’s often called) are the most loyal of all. Some believe polls are no longer reliable. And so on.


Almost an year ago, in late 2019, in my friend circles, I went out on a limb and outlined my prediction for a “yuuuge loss” for Donald Trump. At that time, it was even more of a contrarian opinion. Incumbents have an advantage. A good economy is a virtual guarantee that the incumbent will win again. Still I had my reasons to make that prediction.


But then the pandemic happened. Since I found it completely out of the “analyzable” space, I retracted my prediction, saying, I have no idea how this will play out. Especially with the rapid spread, and both the candidates being over 70 and in the high risk group, in April I had no idea what to expect going forward.


Now, I am again comfortable to reissue my prediction with an updated probability. Trump is going to suffer an epic loss.


Before I give my reasons, let’s define what my prediction means.


What’s in epic loss? I consider getting less than 200 electoral votes as epic loss. Or a blowout. So my prediction is Trump does not win more than 200 electoral votes.


How confident am I in making this prediction? Since the future is unknowable, we can only talk in terms of probabilities. I give it a 50-50 chance. Now before you smirk, let me restate. This is my estimate for the probability that Trump loses the election AND fails to win over 200 electoral votes. The probability for a general Biden win is much higher in my estimate. In other words, I am willing to have a bet with a friend, where if my prediction comes true, I win a dollar and if Trump gets more than 200 electoral votes I give them a dollar even if Trump still loses the election. For a general election result, I am willing to give a 1 to 3 payout ratio. Meaning, if Trump wins, I give them 3 dollars, and if Trump loses, I get only 1 dollar.


What are my reasons for such a high level of confidence?


Here is how I divide the eligible voters into distinct subsets.

  1. People who always come out to vote, and vote Democratic.

  2. People who always come out to vote, and vote Republican.

  3. People who always come out to vote, and vote based on their views on that specific election - the Independents.

  4. People who need motivation to come out to vote and if they do, they vote Democratic. 

  5. People who need motivation to come out to vote and if they do, they vote Republican.

  6. People who need motivation to come out, but may vote either way depending on the election.

  7. Net new additions to the voter pool based on old people dying, and young people becoming eligible to vote and will vote. Some may vote for Republicans, some may vote for Democrats.

  8. People who never come out to vote. Unfortunately there are such irresponsible people in every democratic country.


Which groups are more important in deciding the election? Without knowing the sizes, it’s hard to say. For example, if group 1 or 2 is 90% of the country, it really doesn’t matter what other groups feel or do. Thankfully, that’s not the case.


Since we don’t know the absolute sizes of each subset, we can then focus on estimating the change - the delta - between the last election and this. We can ask how do we expect these groups to behave differently in this election than what they did the last time. It’s a valid strategy to ask that question, as at least one candidate is the same in both the elections. So we can frame this question with respect to Trump. 


Let’s examine one at a time and focus only on the delta.


Group 1. By definition this group will come out and will vote Democrat, so no delta.

Group 2. By definition this group will come out and will vote Republican, so no delta.


Group 3. This group is more interesting to speculate about. Remember, this group will come out and vote. If they vote the same as last time, there is no delta. I am very certain that’s not the case. This is a referendum on the incumbent. Last time, some from this group voted for Clinton, and some voted for Trump. If someone was willing to vote for Clinton last time, I am sure they will vote against Trump again. Those who voted for Trump in this group, are more likely to change their minds. Not all. But some or many. Remember again, this group doesn’t represent the cult of Trump in any meaningful way. All of that is accounted for in group 2. We are only interested in the delta. So, I am arguing, this group represents a big negative delta for Trump. In other words, far more people in this group will change from Trump to “no on Trump”. Please note, not all have to change their mind this way. Some might actually change their vote from Democratic in 2016, to Trump in 2020. I am speculating about the aggregate, not the individual. So “net negative” is the operative word.


Group 4. This was a strong reason for Clinton’s loss. Many people who would have voted for a Democratic candidate, simply did not show up. This is a combination of die hard Sanders supporters and people who disliked Clinton. I am arguing they have a very strong motivation to come out and vote for the Democrats. They see Trump’s record. They see what’s happening with the Supreme Court. Biden is not as disliked as Clinton. Again, not all will come out to vote. We are just looking at the delta. This is a positive delta for the Democrats. No doubts about it.


Group 5. The reverse of group 4. This will be another big negative delta for Trump. How many are going to be really enthusiastic to come and re-elect Trump? A reminder once again. The cult of Trump is already accounted for in group 2. Coming back to this group of “likely Republicans”, will all of them come out to vote again for Trump? When there is no Clinton to defeat? Just like group 4 sitting at home in 2016, a section of this group will sit this one out. This is another negative delta for Trump.


Group 6. For someone who needs a strong motivation to come out and vote for someone (either Democrat or Republican), what’s their motivation this time? If they were not motivated last time to vote for Trump, they are not going to be motivated this time! In fact some may be motivated to vote against Trump. That’s why I am giving this a net negative delta for Trump.


Group 7. Another interesting group to speculate about. But maybe easier to get an agreement on. New young voters - which way will they go? Is there any doubt at all? Seriously! This is another net negative delta for Trump. And perhaps the biggest delta in percentage terms, when compared to the group size.


Group 8. By definition, there will be no delta here.


So there you have it. I do not know the size of any delta. I don't need to. Because none of the subsets is generating a positive delta for Trump. If there was a positive delta, then the question of the size becomes important. Because one positive delta can in theory override all other negative deltas. But I am confident that each delta is negative. Some mildly negative, to some wildly negative. Hence my prediction, and the confidence.


You are going to see this delta affecting “down ballot” candidates. This is not only going to go against Trump in a big way, but against Republicans in general. While I don’t think Lindsay Graham will actually lose, or that Texas will actually turn blue this time, both are a lot closer to happening than last time. As a side effect, Democrats have a good chance of getting complete control - the Presidency (high probability), the House (near certainty) and the Senate (slightly better than a 50-50 chance).


<TL;DR>

All this can be simplified into one question. Has Trump won more supporters than what he has lost? If you think he has converted more to support him than what he has lost, feel free to have a bet with me. Remember again, don’t just look at his cult. It’s the delta that’s going to decide the election, like it does in all the elections. His cult has not grown. But the size of his detractors has grown considerably.


Last time, he won with a razor thin margin. That margin got amplified, as it does in many elections, due to the structure of the electoral college. But that works in the other way too! If he loses that razor thin margin, the electoral map will look a lot different this time. I am betting, since he will lose a lot more than thin margin, it’s going to look like a blowout - less than 200 electoral votes.


UPDATE October 15, 2020
A couple of friends pointed out that I have not called out some of the data points clearly. So here is more explanation - no change in the forecast.

First, this way of doing analysis based on speculating about delta, is valid only for incumbents, which is the case here. Second all the deltas must be all negative, or positive else, the size of the deltas matter, and those are really hard to estimate.

Third, it’s valid only in cases where the margin was not high when the last election was won by the incumbent. I should have called out the third point very clearly. For example, President Obama did not get as many votes in his reelection bid, as he got the first time. Still he won the second time. In spite of negative delta. But his margin of victory was large enough to cushion it. That’s not the case with Trump.

Let’s look at the data. In Michigan, Trump won by an extremely small margin of 11K votes. That’s 0.23% of the votes. That margin is extraordinarily hard to defend, especially when you see that, in 2016 all other candidates (Green Part, Libertarian etc) got 5% of the vote (250K). So the chances of negative delta making Trump lose Michigan is very high. Note that Michigan had voted Democratic in the previous 6 elections. So 2016 with such a thin margin is not a secular change, but an outlier.

In Wisconsin, Trump won by 23K votes, which is about 0.75%. Third party candidates won 5%. And this state has voted Democratic for the previous 7 elections. This was again an outlier, which is very hard to defend for Trump.

In Pennsylvania, Trump won by 44K votes, which is also about 0.75%. Third party candidates got over 3%. This state has also voted Democratic in the previous 6 elections.

This is the basis for speculating about the deltas. A margin of less than 1% in a Democratic stronghold is too thin to stand on for a Republican candidate.

Similarly, Trump's margin of victory in other states was - Florida 1.2% and both Arizona and North Carolina less than 4%. None of these states is reliably Republican.

No comments:

Post a Comment

Related Posts Plugin for WordPress, Blogger...