Presidential betting market's 6x money

Short-term update: latest article is here.

A world of pain agonized over Donald Trump October 7, the day of the now infamous and disgraceful video leak.  While scheming up a response in his Trump Tower, the markets surrounding presidential predictions went into an insatiable hyper-drive.  Instantaneously, the election betting markets saw the odds for a Mr. Trump victory plummet from ~35%, to 20%.  We had discussed a couple days earlier (with >100k reads) that we calculated the true probability to be closer to the low 40's%, a point that we'll revisit.  Now pollsters often have their own agenda and their failed modeling techniques led to a mirror drop in polling results.  The "probability" for Donald, from either the markets or pollsters now incorrectly sits at 15-20%, and a full dossier of opinions from throughout this weekend have not fully been ingested.  Hillary Clinton's seemingly runaway spread has already peaked in the past day, and dropping markets demonstrates renewed prospect of a Donald Trump triumph.  And the weekend circumstances surrounding him (along with his GOP leadership) clearly warrant somewhat of a drop in probability, regressing to say ~1/3 for now.  But there is still a large negative bias in what some experts perceive to be Mr. Trump's winning likelihood.  Among the cases we'll examine here is a flashy, yet consequential, example utilizing the betting markets.

To be clear, there is no circumstance where we would argue betting on Donald Trump to win.  Money doesn't lie by that much.  However it is interesting to note the implications of miscalculated probabilities as they impact available wagers.  In this case we were suggested to come up and it was shared by a top Mr. Trump advisor that there is a 35% or so probability that he will either mostly rebound in the polls (from this October 7 slide), or outright win the election.  That only seems high if you are easily persuaded by the mainstream media, but we should note from our own research that these "professionals" have flip-flopped in nearly 1 out of 5 Primary state elections so far, and of national Presidential match-ups covering many decades: >2 out of 3 have seen high single-digit tightening, and 20-25% of those also saw surprise victories for the nominee who was lagging in the polls by this time in the season (and in much of these general election cases it was in favor of Republicans).  Bottom line: the unlikely happens far more often then we recently give it credit for, especially in matters of celebrities running for higher office.  From my time serving in the current Administration for a few years, I can appreciate that the nation is full of decent and talented people, and nowhere should polls be an excuse for untamed emotions (as opposed to compassion) to be shown to our neighbors now, or in-between elections.

Betfair is a regulated, online gambling exchange that shows Donald winning 40's% in the popular vote, with nearly a 6:1 odds (and that backs into the same 15-20% noted above).  It's evident that the recent video scandal has caused the odds and his likely performance to become too exaggerated.  You can purchase a bet -not necessarily that Ms. Clinton (who through her marriage bills herself as the "most qualified" person to ever live) would lose- but that her current "winning spread" would at least decidedly tighten.  That's again worth 35% odds, and you would be able to purchase that bet right now for 1/2 that price!  Of course should Donald Trump simply exceed his current depressed expectations between now and November, then you would be able to capture upwards of 6x winnings.

Wager $50k, and walk away with a $350k payback!  That's a lot of money for being truthful with the math!

A number of leading probabilists have also pointed out the grandiose failures in the craftsmanship of over-active, mainstream pollsters, including a recent error by the L.A. Times where they only sample one Black Chicago youth and then was confused about that person's perspective (he was leaning for Mr. Trump of course, because what the hell does he have to lose?!)  This shows the blatant volatility a single pollster can induce, even casting systematic error permeating through the entire national polls-of-polls (per my first article link above).  Our blog readers will not be surprised by such mathematical follies, as we have pointed them out numerously, and across industries.

Other blog readers have also complained on our use of 4-way polling (including 3rd party candidates), since they feel a two-person match up shows the wider spreads that they would relish.  But outside of fantasy, aren't there more than two people on the ballot?!  Also it can't be stressed enough that when mainstream modeling output suddenly veers into the outer 20% (on either side), then there is a high chance this is not a "real probability" (in one of our analysis Andrew Gelman termed "cogent") but it is simply a cue that there's subjectivity afoot and sensitive modeling errors exist.  That's explains Hillary's recent spike, arbitrarily several weeks before election.  The real probability for both gallant candidates remains in the inner 60% of the probability range.  And probabilities drift only gently, unless something really climactic happens.  Such as the death of a candidate, not merely a melodramatic reaction to a leaked tape that politicians masquerade as the same as the candidate's death!  Donald Trump still debated well on Sunday, correct?  He is still on the ballot, correct?  Most people couldn't turn a million dollars into a billion, correct?

On a final note we should simply advise someone interested in placing one of these election wagers, perhaps also noticing the fortunes that were made by some during the summer's Brexit vote.  While the payoff here is highly lucrative for the temporarily scorned candidate, it is still a binary result within tight parameters (e.g., Mr. Trump popular vote is in the 40's%).  The bet is a highly risky derivative, and even at 35% chance of success (or twice the 15-20% priced in), 35% is still inarguably a low probability.  The transaction costs for betting account can be high, the trade is illiquid, and the Black Swan opportunity is simply for this election (can't diversify among many bets like this).  You will likely prefer to capture this slight election probability outperformance we note here, through some other investment vehicles that are more palatable and less risky.

It is simple to register and vote.  Do it now!  Fantastic things are happening on the policy landscape, and while the odds of your vote making a difference are about the same as winning the lottery, the former won't drain your lifetime's savings and will lead you to great satisfaction within your community.

 

Update: a day after we published this article the prediction polls finally reversed course (by >25%!), as we long said in advance that it would.  As always, Nate Silver and mainstream pundits were caught by surprise...