Note that once you include the MoE, 538 was still predicting a maximum for T of about +9, while we had +12.6, so we were quite close.

Election day is over and Trump won, as expected. In a few days, allowing the time to get final results in, I will post a complete analysis about how wrong pools were. Look at Florida for example: 538 predicted T+6.6, we had an average of T+8.2, was T+13.1.

And this is the winning chance from pools in the last 2 and half months. One thing is certain, in this round Trmp was pooling better than ever. He won in 2016 with much worse pooling. We'll see tomorrow.

This is a simulation of the possible outcomes. About 50% chance that Trump gets betwwen 270 and 295 electoral college votes tomorrow.

And these are the corresponding winning chance for the electoral college. Please note that election results have been very different from the pooling predictions both in 2016 and 2020 (despite Biden victory, Trump got much better results than in pools in all swing states).

These are the corresponding winning chances. AZ, GE and NC look likely. PA and NV close to a tossup. WI really a toosup, with the change slightly above or below 50% depening on the choosen poolset.

Today marks our last update, election day is tomorrow. Here are our final pooling averages for swing states. Note that some states like WI, but also NH, have very large Margin of error. TX and FL look sure. MN and VI look our of reach. NH is possible, though unlikely.

This is a simulationg of the electoral college outcomes based on the latest pools. Bar heights represents the fraction of times the Trump got a certain number of electoral college votes in the 1000 simulations run. Red: >=270, Blue <=268, Yellow=269.

Electoral college, Winning chances by dates. One important thing to note is Trump's momentum. Every model will predice a different winning chance, but every model that has the minimum sense will predict a larger chance only if pools are improving. And they are.

Electoral College winning chances, as of today. New record high for Trump!

Update on Trump winning chances in battleground states. Trump finally up in Nevada as well.

Not frequently updating cause not much has changed lately. Trump is stable at 60-40 lead.

Today we also made a simulation to show possible election outcomes. We simulated 1000 events. here are the final results for Trump electoral college votes (yellow=tie):

This is how the hystorical looks like using the more conservative approach:

Translating pool averages to probabilities is not easy to be done in a meaningful way. We updated our model to be more conservative on Trump chance's for victory. here is the result (in line with @Polymarket ):

Here are the updated pool averages in battleground states. Pools from @NateSilver538 Silver Bullettin, we include now Margin of Error for better significance of these numbers.

We find however that this approach might be optimistic. In particular, the differences with other models is minimal except that for WI, where pooling was very off both in 2016 and 2020. We are now implementing some corrections for a more conservative estimate of Trump’s chances.

In our model, we base our weights on the 538 poolster rankings that weight in partisan pooling and expand the MoE in the direction of the past bias of the poolster, but we do not apply arbitrary shifts to correct the pools.

Other models try to address partisan pools by shifting the results by a certain number of points in the other direction. On top of that they also apply a lower weighting, doubly-suppressing the result of such poolsters.

Today, Trump reaches an all-time-high chance of 66.9% of winning the election since Biden Dropout in our model based on 538 pooling database.