The Latest on the Election Horserace

While Harris has been trending higher as the post-debate polls are rolling in, the broad measures of betting markets, polling averages, and models are best described as "too close to call."

Forty-three days until the election.

Comment

We are fans of election betting markets. They aggregate all known information into one probability. Manipulating them is pointless. It will not change anyone’s vote. Even if bettors are partisan, they are there to make money, not express support for a particular candidate.

They are no more predictive than polls, but they are real-time, so when events like this month's debate occur, they can measure how much things have changed faster than polls.

Real Clear Politics averages the eight largest betting markets in terms of the probabilities of who will win.

In the wake of the debate, Harris’s (blue) and Trump’s (orange) positions flipped. Harris leads Trump 52% to 47%. This is best described as being too close to a call, with a slight but not meaningful advantage to Harris.

The largest of these eight prediction markets is Polymarket. As of September 22, Trump/Harris betting had a combined $980 million on the line.

It is the same as the average above Harris (blue) at 52% and Trump (orange) at 47%. Again, too close to call.

Aggregating Polls

Various websites aggregate all election polls into rolling averages. Below, we show two of them to illustrate that they give similar results.

First is the poll aggregation from the website FiveThirtyEight. It shows Harris (blue) leading Trump (orange) by 2.72% (green bars). Kennedy (red) is also shown until he dropped out and endorsed Trump.

The second is from Nate Silver’s Substack “Silver Bulletin.” It shows a similar reading of Harris (blue) leading Trump (orange) by 2.98% (blue bars).

Silver Bulletin Models

How should one interpret the polling data above??

First, it should be noted that the above are national averages. Democrats are believed to have about a 2.5% advantage in national polling due to California’s overwhelming support for Democrats. In other words, a 2.93% Harris lead nationally is considered a 0.43% Harris lead in the “other 49 states.”

Consequently, Harris is believed to be polling mediocre despite having a national lead. So, the Silver Bulletin model has Harris (blue) leading over Trump (orange) by a very narrow margin despite Harris' national polling lead.

Note that this model is a probability for who will win. Currently, Harris has a 54% probability of winning. This is very close to the betting markets above. However, like the betting market, anything less than a 66% probability is still best described as too close to call, with Harris slightly ahead.

For NFL fans, this is the equivalent of Harris being a pick-em to a one-point favorite.

We highlighted the word national above. But the election is based on the Electoral College. The Silver Bulletin models do take this into account.

The following chart aggregates individual state polling into rolling averages. The most crucial swing states are shown.

The shaded area is the margin of error. Note that all of the polling averages are within this margin.

Silver Bulletin state models show that Harris's winning probability has increased. But note that none of them are above 66% (or below 33%). So, again, these are best described as too close to call with a trend toward Harris.