Oops! I made the convention bounce adjustment disappear.

One neat trick — for today's newsletter only.

Let me let you in on a little secret of blogging — or, excuse me, newsletter writing.

You always want to write up the most time-sensitive stories first. The items with a shelf life. The evergreen stories can wait. This principle is so important, in fact, that it’s usually worth changing plans even if you’ve already teased at a particular story. So, readers, I’m going to offer you my apologies: yesterday, I said there would be a Model Talk today on whether prediction markets are biased toward Republicans. And, well, it turns out I lied; we’ll hold that item for another day.

Instead, you’re getting a story that won’t be so relevant after next week’s debate. I’ll show you what happens in the model if you toggle a single setting, turning the forecast’s convention bounce adjustment off. This was shaving a point or two off of Kamala Harris’s polling averages — but for purposes of today’s newsletter and today’s newsletter only, poof, it’s gone. 

And then, once that’s done, I’ll take one more try at defending the adjustment — while acknowledging that Harris’s late entry into the race complicates all of this.

What the model says with convenetion_bounce_mode off

OK, let’s cut to the chase. Harris is still ahead in our polling averages, albeit extremely narrowly in some cases, in states totaling 292 electoral votes. So without the convention bounce adjustment, she’d be winning, right?