I think he explained it pretty well in the thread. I clicked through to the PA averages and it did seem to offer examples of what he said, Biden +6 and +5 polls that include data as recent as 11/1 being cut off and recent Trafalgar/Rasmussen polls being in the average.

I've followed Cohn a while and I do give him the benefit of the doubt that he's being honest in assessing RCP as using a shifting-goalposts model in offering current averages.

The political betting markets are useless in terms of predictive ability. Whatever the conventional wisdom is will be favored. They have no special insight nor are they any more accurate than any cable news pundit just because people are putting money on the line. And if you are talking about Predictit, the odds there are often different because there are betting limits and high fees so many people who would otherwise use it, won't.

I don't, started paying close attention to them after RP was gone. Were they 20+% off polling models with high volume across all sites for weeks? Was this on InTrade?

Natamus,
That's quite the tame follow-through for "If I get banned, I get banned, YOLO." The blatantly incorrect hyphen was the most offensive part.

That was just for you my guy

My impression is that in 2016 the betting markets were better than the pundit conventional wisdom in giving Trump 15-25% to win.

Prediction markets are bucking conventional wisdom in giving Trump relatively high odds. So if they are usually tracking conventional wisdom, why not in this election?

"Conventional wisdom" is a tough concept to define, but I see hardly anyone (other than, go some extent, the modelers) acting like this election is a blowout.

Considering Trump out performed the PA polls by 2.6% in 2016, this late trend doesn't look good for Biden.

Here's the 538 moving average in PA.



The race did tighten a bit there leading up to today, but the significant RCP lurch doesn't look to be very representative of the picture.

These numbers are going to skew heavily male though.

Trafalgar literally said "we will find the hidden Trump voters!" when they announced they'd start polling the race. Paragon of objectivity!

Looks like Ohio is going to be the sole source of new attention at 7:30 p.m. ET, and North Carolina is joining the reporting during the 8:00 hour.

Biden -210 on bovada

I think it should be uncontroversial to say that, should the election go against their side, the left is more likely to do violence against windowpanes and the right is more likely to do violence against human lives.

We will disagree on that one.

The movement is towards Biden oh boy

Nah. Betting markets are in line with conventional wisdom at all, unless you look at them only as a binary, who'll win. The principle bias in betting markets is a severe overestimate of the probabilities of rare events. Some of that is due to the nature of the site's vig, but people are really bad at telling the difference between a 5% event and a 0.5% event. The second is that with the low betting limits, there are a lot of people who can move the markets based on their alternative realities, or even political agendas. This dates back to at least 2008, when there were several instances of massive buys on John McCain on InTrade that shifted the market. We never really got word whether it was someone completely loony or someone trying to create positive headlines, but it was probably some of both.

I guess there's room to quibble about whether assigning a 10% probability to a 0.01% probability event is a larger mistake than assigning a 60% probability to a 40% probability event, but I generally think the former is the bigger mistake, and Ron Paul fandom lead to a whole lot of the former.

The way this article reads is that the betting markets were giving Trump 15-25% to win, until after results started coming in that night. That's not especially remarkable.

The death count from BLM and antifa is at, what, I think 1, going back to murder of George Floyd.

Has Timex ever done anything non-optimally? Jeez Louise.

Let's not lose sight of the fact that the real terrorists are still out there, running around throwing milkshakes at nazis.

Fair enough, although I don't really expect prediction markets like predictit to distinguish between a 1% and 0.01% probability of an event happening in interesting ways.

Sure, not that remarkable, but right around where the Upshot and 538 model averages were, which from my experience of 2016 punditry was better than the norm.

David Dorm
Dave Patrick Underwood
Damon Gutzwiler
Shay Mikalonis (vegetable for life)
Max Brewer (was critical at one point, I don't know if he lived)

There are dozens and dozens of incidences of police being shot at, blinded, or otherwise seriously injured by BLM/Antifa protestors. You're not correct on your facts.