If I managed a HF with billions at stake I would absolutely spend a few thousand on exit polls in early swing states. I would then have the best statisticians infer as much as possible from that sample.
You can probably do Pennsylvania with a sample of ~1000 voters. Similar for North Carolina. https://m.primal.net/LtbK.png A team of 10 surveyors can get you that sample in a few hours for 20$/hr. The cost of having a good early estimate is ridiculously low. Why wouldn't you do it? And if you have the info... why not use it? And if you are going to use it... Bitcoin is the best way to bet on the results.
By 12:00 EST / 17:00 GMT / 18:00 CET you are going to have a pretty good idea. https://m.primal.net/LtbZ.png