Not sure what the political news coverage means? Forecasters have something to say about that
Written by Miles McCrocklin
Editor’s note: Forecast is a social prediction market: an exchange where participants use points to trade on the likelihood of future events. Community members pose questions about the future, make predictions, and contribute to discussions leveraging their collective knowledge. Download the app to join the community.
This post was originally shared on Substack on November 6.
We live in a time where news can become stale, or worse, misinform readers the moment it goes online. Forecast, an experimental project built by Facebook’s NPE group, can help sift through the noise and showcase an evolving community consensus.
Forecast is a social prediction market: an exchange where participants use points to trade on the likelihood of future events. Community members can pose questions about the future, make predictions, and contribute to discussions leveraging their collective knowledge. The market ‘prices’ (which are range from 0–100) indicate the community’s best guess at the probability of any given event.
Since launching Forecast, we have noticed something encouraging: price changes on Forecasts tend to closely mirror news events. As the news changes, the community surfaces it, reading through the details to make educated forecasts of what’s happening. It doesn’t hurt that Forecasters have a strong incentive to quickly break news into the community*, but only if it’s high quality**.
At a tiny scale, this means that Forecast is starting to function as a kind of time series aggregator of news+analyses+opinion. News and opinion go in, and out pops a snapshot of how the community has grappled with a complex issue over time.
Show me
A question that highlights this effect is: “Will the US government ban TikTok by the end of 2021?”.
This narrative has changed drastically day by day. Along with it, there has been a slew of quality reporting on the issue ranging from explainers to op-eds, to breaking news. Often this coverage has headlines focused on what was said (e.g: Trump issues orders banning TikTok and WeChat from operating in 45 days if they are not sold by Chinese parent companies) requiring people to dig deeper to understand the downstream impact.
Let’s compare that with Forecast, which poses the question to the community and then asks them to both choose to invest in what they believe the outcome will be and do some research or explain their rationale to back up their position. Here’s an example:
Since this question was posed to the community, we’ve seen the predicted outcome change frequently. Let’s dive into the trading activity:
Reason writing behavior is clustered around the biggest price spikes — these are the moments when the community is assimilating new information into the forecast. Note that there were two days where the forecast shifted drastically. Let’s look closer at one of them from July 30th to August 10
As you can see, Forecasters were pretty quick on the uptake in capturing what was happening in real time and using it to updated their positions. More recently, the consensus has equalizes to an expected outcome, but if the historicals are any indication, we may see more volatility before this question closes in December.
But is this a fluke?
Let’s look at a question that we already know the outcome for. Will a second presidential debate between Donald Trump & Biden take place on 10/15/20?
Prior to October 1, the consensus in the Forecast community was that this question was a given — there was no serious reason to doubt that the debate would happen. That changed significantly when the Trump campaign announced that the President had tested positive for COVID. In the days leading up to the scheduled debate time, this question it became heavily debated in the app until the debate was ultimately cancelled.
Looking at this question in this way, you can see a quick summary of what people thought would happen at a given moment in time. This is both useful in the moment as understanding is changing and after the fact as a sort of historical record. If you missed the twists and turns of a developing narrative, Forecast’s reasons can give you an aggregated sense of why the conversation shifted.
Some final thoughts: check our work
Forecast is still working at a small scale. We’ve noticed (unsurprisingly) that this news-driven volatility effect is much stronger on shorter-term questions (especially those closing within the next 3 months), likely because the incentive to make a forecast weakens as the time to close lengthens.
If you want to join and help push conversations like these forward, download the app! All the data — transactions, forecasts, and reasons — used to make these forecasts is available here. We invite you to take the data for a spin and tell us if you have any feedback!
