What the Hell Just Happened?

Understanding Saturday's wild shift in the betting markets

The betting markets went a bit haywire last night (the last Saturday before the election). Over the past 24 hours, Trump's price dropped from 59.5% to 54.0% on Polymarket, and from 55.2% to 51% on Kalshi. And they seem to be struggling to find a new equlibrium. Given that not much had happened to change the nature of the race (certainly nothing meeting 2024 standards of campaign chaos at least), many, myself included, have been left wondering: what the hell is going on?

Below are some possibilities.

1. The Iowa Selzer Poll

The consensus seems to be that the price change was caused by the release of the Iowa poll conducted by Ann Selzer for The Des Moines Register, which showed Harris leading Trump by +3 points. This was a notable contrast to the recent Emerson poll that had Trump up by +10, a figure more in line with his margin of victory in Iowa in both 2016 and 2020. The Selzer poll stands out due to its reputation for accuracy, raising questions about whether this could signal a larger trend beyond Iowa.

Why might this matter so much? If the Selzer poll is even slightly accurate, it could indicate a broader shift in the electoral landscape that favors Harris. The possibility—not certainty—that this poll reflects trends in other key swing states was enough to make traders take notice. Betting markets are highly sensitive to new data that might shift the perceived likelihood of an outcome. This explains why there was a notable, albeit not overwhelming, move away from Trump. The market's moderate reaction suggests that while this one poll isn’t enough to cause a full swing in favor of Harris, it does reflect some information that wasn’t already priced into the market.

The problem with this explanation. The above logic relies on cherry-picking one poll in one state that has been very lightly polled during this cycle. Yes, as stated above, if there is even a glimmer of truth in the Harris +3 projection then it signals something could be happening below the surface elsewhere that hasn’t been picked up in other polls. But by that logic, why do we adopt the Iowa narrative while ignoring the equal but opposite narrative that could be spun from polls in Minnesota, Virginia, New Hampshire, and New Mexico.

Compare the RCP average of polls in these states to what the RCP averages were in 2020 and the actual outcome in 2020.

2024 RCP Averages:

• Minnesota: Harris +4.8

• New Hampshire: Harris +3.5

• New Mexico: Harris +6.7

• Virginia: Harris +5.8

2020 RCP Averages (Actual):

• Minnesota: Biden +4.3 (Biden +7.2)

• New Hampshire: Biden +9 (Biden +7.2)

• New Mexico: Biden +12 (Biden +10.6)

• Virginia: Biden +11.5 (Biden +9.4)

In other words, the RCP Averages were reasonably accurate in these states in 2020 (within a few points) and Harris is polling 5-6 points worse than Biden in all except Minnesota.

So whatever narrative is being applied to this one Selzer poll could just as well be applied to the RCP averages in other non-swing states, and the narrative flips completely.

2. Leaked or Anticipated Final Polls

While there were few polls released on Saturday, a bunch of polls were expected on Sunday, including the final NY Times / Siena poll in the battleground states. Early Sunday morning, these polls were released and appeared to lean more favorably towards Harris.

What’s interesting here is the timing. On Saturday night, before any of these polls were officially released, I suggested that the shift in Harris’s favor was likely due to more than just the Selzer poll. The fact that the market didn’t budge after the public release of the NY Times poll, which was favorable for Harris, suggests that the market had already moved to incorporate that information. In other words, the market move on Saturday night might have been anticipatory, driven by hints, whispers, or even a leak of what was to come.

The problem with this explanation. While it is very common for information to leak early, and for it to be reflected in the betting markets before it shows up in the public domain, the problem with the explanation is that it lacks hard evidence.

3. Panic and Irrational Behavior

Finally, it’s possible that the market shift was due to sheer panic and irrationality. The NY Post headline after the Iowa poll release blared "SHOCK POLL," sparking a wave of reactions on social media platforms like Twitter, where political bettors and observers went into a frenzy dissecting its implications. Given that participation in these betting markets remains largely recreational, dominated by individuals rather than institutions or seasoned professionals, the potential for knee-jerk reactions is high.

Here’s how this might play out: If enough traders panic and start selling Trump positions, the downward price momentum can create a feedback loop. As the price drops, more traders may jump on the bandwagon, causing an outsized move. The subsequent partial recovery in Trump’s odds suggests that the market overreacted initially before stabilizing.

The problem with this explanation. Like the above two reasons, this one is possible, but can be neither proven nor disproven. It also takes the defeatist approach of dismissing any market moves that lack an obvious explanation as “irrationality”, when in fact markets tend to move for good reasons, even if those reasons don’t immediately present themselves.

So what actually happened?

Well, hard to know for sure. Almost certainly, it’s some combination of all of the above; the real trick is to discern whether fundamentals, insider info, gamesmanship, or emotional trading is in the driver’s seat. One way or another, the episode demonstrates that even relatively stable markets are at best one hiccup away from whipsawing bedlam.

As the next 48 hours are likely to confirm.