On July 21st 538's model provided the following probabilities for the Biden-Trump matchup. Both Nate and 538 have since relaunched their forecast for Harris-Trump.

Nate Silver: 72.8% Trump, 26.9% Biden

538: 50.7% Trump, 48.9% Biden

The difference between the models was 72.8-50.7=22.10% for Trump and 48.9-26.9=22.00% for Biden, for an average difference of 22.05%.

On November 1st, 2024 at (approximately) 12pm PST I will do a similar calculation using the latest data available from both sources. All the options that are true will resolve to YES. So (for example) if the average difference would end up being equal to 7.95%, I will resolve 1%, 2% and 4% to YES and 8%, 16% and 32% to NO.

The exact formula for the calculation, to avoid any misunderstanding:

1. Take the odds for the top two candidates from both models.

2. Substract the probabilities, finding the absolute difference for both candidates.

3. Average the two numbers to arrive at the final result.

Edge cases:

1. I will use the numbers from the official forecast pages of both models. Numbers from Twitter or other social media won't count.

2. 538 haven't relaunched their model just yet. In the very unlikely scenario they won't relaunch it prior to Nov 1st, all the options will resolve to No. Unfortunately N/A's are no longer possible. => the 538 model was published on Aug 23rd, so this clause is no longer relevant.

3. The latest published model from both sources will count. So if (for whatever reason) Nate never updates his model again, I will still use whatever were the last published numbers. If 538 relaunches their model on August 20th and then never updates it, I will likewise use their last available numbers.

4. If one of the candidates drops out before either forecaster has a chance to update their models, I will do the comparison of predictions by party affiliation instead. So if Trump drops out last minute and only Nate has a prediction for Vance-Harris, I will use Trump's latest numbers from 538 as a placeholder for Vance's numbers.