Small Steps to Accuracy: Incremental Updaters are Better Forecasters

During their recent episode of the VALUE: After Hours Podcast, Taylor, Brewster, and Carlisle discussed the research paper – Small Steps to Accuracy: Incremental Updaters are Better Forecasters. Here’s an excerpt from the episode:

Jake Taylor:
So this research paper came out this week. The author is, and I’m probably going to say this wrong, but it’s Pavel Atanasov. Atanasov, something like that. I apologize Pavel.

Jake Taylor:
Exactly. So he’s a PhD researcher in decision making science, and the name of the article or the research paper is Small Steps to Accuracy.

Bill Brewster:
Sorry.

Jake Taylor:
Incremental belief updates are better forecasters. So they took these four years’ worth of forecasting data where people are making probabilistic predictions in these prediction tournaments. And 400,000 observations, so it’s reasonably robust, but what they found was that the more accurate predictors made frequent small updates to their predictions, and the low-skill predictors, they confirmed their initial hypothesis and made large and frequent revisions to their. So they didn’t make little adjustments to what they thought the world was going to look like, it was like all or nothing in big swings.

Jake Taylor:
So the best quote in the whole thing was, “The best forecaster seemingly experienced the prediction task as a long sequence of slight surprise, rather than a short string of hard collisions with reality.” So all of us right now are dealing with a lot of what feels like reality changing. Perhaps we need to increase our dampening of prediction changes and maybe not move so quickly in one way or the other and stay a little bit more true to your original thesis, and slowly update rather than bounce off the guard rails, which I feel like I’ve been doing that more than I should be at this point.

Tobias Carlisle:
That’s great, isn’t it? That’s like a little bit of Bayesian updating to your original thesis, and the more little iterative moves you make, the closer you get to reality without overshooting. That’s so funny. I saw a tweet today, somebody must have read something like that and they said, “The better investors through these periods are just making incremental changes to their portfolios rather than tossing everything out and starting again from the very bottom.” It’s an interesting… I think it’s a great approach.

Bill Brewster:
I think that’s interesting. It sort of supports the idea that maybe the market is somewhat fairly valued here, right? I mean, maybe over the long term this is just a smaller change than feels like it right now. It’s very interesting.

You can find out more about the VALUE: After Hours Podcast here – VALUE: After Hours Podcast. You can also listen to the podcast on your favorite podcast platforms here:

Apple Podcasts