I’ve heard stories (unverified) that hedge funds manipulated political betting markets ahead of Brexit. Whatever loses you made betting would be more than offset by even a tiny shift in govt bonds or whatever.
If something like this happened, and the inherent correction mechanisms of a market were insufficient, I would probably just focus on the forecasts of top users.
Thank you!
If memory serves, Nate Silver managed to dominate some news cycles with his predictive model that mostly just aggregated opinion polls
I suspect I could do something similar. Look forward to “This super sophisticated model that definitely isn’t just a weighted average of prediction markets predicated 49 out of 50 states!” coming soon to a mainstream newspaper near you
Appreciate you trying it out
Agree, the key next steps is marketing/distribution and user research. Also ruffling up some funding!
Reading a book called ‘The Mom Test’ on user feedback now :)
Eventually I would like to commission more high quality content from e.g. superforecasters.
I’m also working to improve the AI-summaries… eg wouldn’t it be cool if I could get AI to find historical data to contextualise each news story?
True! I’m just trying to keep it simple
Open to feedback for a better ‘catch all’ term
(I feel like ‘forecasts’ doesn’t capture the fact that this aren’t my forecasts and ‘wisdom of the crowds’ isn’t concrete enough)
Thanks! I’ll send out a newsletter poll as well (and open to any other ideas of places to poll users other than Twitter)
I mean I basically have to make a decision market, considering my project!
Thank you!
I’ve fixed the last headline now. I agree that it was being driven by the resolution date and therefore misleading.
I’m curious what you think of e.g. the Putin headline. Of the 8 Putin markets, some resolve by July, some by October, and most by 2024 -- but all of them show ~90% or better odds (well, Hypermind is at 88%). So even if technically correct, would including the resolution dates really add value?
Especially on an data-ink ratio basis?
Most prediction markets used on the site resolve by 2024 (unless specified otherwise on the chart). It would feel redundant to include “by 2024” in every single headline?
My subjective impression is that odds start collapsing as the resolution date gets near, more so than some linear decline as the market window slowly closes. Some markets do trend down (peace odds), but others sideways (Putin, Crimea), and others trend up (Russian territory gain). So my current sense is that as long as headlines don’t overweight markets closing within say ~2 months, it’s probably okay?
Anyway am still very open to feedback on this. Just trying to navigate the trade-off of keeping things simple on an already cluttered site...
Thanks again, it was a stimulating post and I think this is going to be an important issue for me to get right. Will keep thinking about it.
Oh, I’ve also added a Patreon button now. Warning: no benefits yet! Purely to support the mission
This is useful feedback, thank you.
Re headers and charts. I think what I will do is this: Click on headline --> separate page with full sized chart (and the news links, more info etc)
Re fonts. Yes, the design is ‘inspired’ by the NYT (and the motto by WaPo). It’s kinda meant to be little bit of a pretentious ‘performance art’ making fun of major newspapers. What if NYT was founded in the era of prediction markets? kinda thing
Re formatting. Thanks, I’ve fixed the spacing. The borders is a deliberate choice believe it or not! The idea is to elevate the highest level headers (e.g. Russia-Ukraine) above the borders, to make it clear that a new topic has started. I tried ‘fixing’ the borders, but at least to my eyes it looked worse to be honest—I’ll keep fidgeting...
Thank you again! It will continue
The goal of The Base Rate Times is to be a combination of prediction markets and news aggregator. Right now on the site there are links to the latest stories with AI-generated bullet-point summaries.
I’ve been trying to link news stories to prediction markets, but still figuring out a more systematic way of doing it.
I am currently testing how to make AI article summaries even better. My goals are for AI to:
find historical data to contextualise the story
advise forecast updates based on the story
make counter-arguments
translate articles from e.g. Chinese
detect bias & cross-reference claims
Yes, I love News Minimalist and check it everyday! But I don’t want to just directly rip them off!
I’ve been thinking about how to use AI for news curation in a way that’s original and specific to The Base Rate Time’s mission (news via prediction markets).
Newsletter (coming soon) is going to include a weekly summary (changes in forecasts, latest headlines).
Ideally I’d like to do something more value-added/original than that as well. For example, I love the Nonrival newsletter (hat tip Nuno Sempere), which collects forecasts from its readers—check them out!
Re: RSS, thanks Garrett. Will add an RSS for the website as well, but the way I’ve set things up doesn’t make it straightforward (I don’t update via ‘new posts’ on a CMS).
Another way to stay updated without checking the site is my twitter (@base_rate_times), if that helps.
In my corporate jobs, simplicity was prized due to the high informational load. In fact, many colleagues appeared to absorb only sharp soundbites.
On the other hand, even if you had the answer, implementing it without ‘aligning stakeholders’ first would be deadly.
Re injury prevention, you may be interested in Stuart McGill, professor of ‘spinal biomechanics’ at Waterloo.
For example, you shouldn’t exercise first thing in the morning. Something about your discs being more hydrated --> swollen --> vulnerable after laying overnight.
Very open questions, but curious if you have any thoughts on Olympic lifting vs powerlifting. Or on kettlebells and hip-hinge exercises.