Forecasting newsletter #1/2025: The calm before the Trump
Highlights
Lull after US elections and winter vacations.
With the new Trump administration, regulators will shift their approach to crypto and prediction markets.
Crypto.com adds an event contract for the SuperBowl, using its previous acquisition of Nadex.
Index
Prediction markets and forecasting platforms
Regulators
Research and articles
Odds and ends
Prediction markets and forecasting platforms
US Democrats’ ability to wage damage on the prediction markets sector wanes with the end of the Biden administration. Industry giants like Interactive Brokers, Crypto.com, and maybe Robinhood, are offering or looking to offer prediction markets, and they have been getting the regulatory permits related to that.
Polymarket’s regulatory problems could be solved with e.g., a CFTC no action letter and a pardon, with a regulatory framework that unshackles crypto. Trump’s team could put an end to the FBI investigation of Polymarket. But will the Trump administration make undoing damage to prediction markets specifically enough of a priority? That remains to be seen.
Shayne Coplan makes the Forbes 30 under 30 (a) in the finance category—he’s just 26 (my own age) joining such luminaries as…. Congrats! Polymarket had a particularly neat market (a) on whether a particular chain would get hacked. It partnered with Packy McCormick to cover its markets in Boring News.
Total value locked (a) seems like it has stabilized a bit at $100M, as Polymarket added new markets making sense of Trump appointments. The significance of this is that the Polymarket user base hasn’t cratered, but it also hasn’t yet found continued growth after the US elections—yet.
Decrypt.co (a), an outlet covering crypto, started myriad.markets (a), a prediction market that aims to complement its reporting and increase engagement. So far users just trade points.
Manifold added loans (a) back.
Crypto.com acquired Nadex (a) back in 2021. Earlier yet, in 2012 (a), Nadex had tried to offer political contracts, but was stopped by the CFTC of the time. Using its know-how and bureaucratic muscle, Crypto.com self-certified (a) Super Bowl (“title event”) event contracts. Nadex/Crypto.com traders have begun trading (a) those contracts.
I am however, saddened to see that these are sports contracts, which are socially useless and extractive, rather than anything more hedge-based, futarchic, or broadly cooperative and socially useful.
Kalshi added a few markets on the Luigi Mangione case, but soon withdrew them
Fatebook has a predict your year (a) page readers might want to forecast on.
Metaculus has a tournament with Vox’s Future Perfect (a).
DraftKings has a $20/month subscription plan to get better odds.
Play Money, a Manifold competitor aiming to be more wholesome and have fewer dark patterns, has an API (a); here (a) is a worked example.
Cultivate Labs CEO Adam Siegel looks back at 2024 (a). I didn’t know about their integration in Fox news (a). They are also continuing to develop a research analyst companion for forecasting and analysis (a)
PredictIt added new markets. Brave move.
Regulators
CFTC is refusing (a) court required mediation with PredictIt. The CFTC, perhaps spurred by Kalshi, has been going after PredictIt and Polymarket. But the Trump administration will be much more friendly to crypto and financial innovation, and the general stance of the CFTC (a) will change withe new US administration.
Brazil authorizes 81 online betting operators (a). Goes to show that online betting is a fast growing business. But also, in some sense Polymarket might now be illegal in Brazil, since it doesn’t have that particular license.
Ohio regulator reports (a) over $1B in sports betting wagers for November. It seemed like a poignant point of comparison when thinking about the relative sizes of sports betting and prediction markets.
A bill to ban election betting was filed (a) in the US House of Representatives, by Democrats. It seems unlikely to succeed.
Research and articles
Tom Adamczewski reanalyzes some AI expert survey answers, using CDFs instead of point estimates.
LLMs do well (a) at predicting time series when given written context about what they are forecasting—for instance, power capture by solar panels goes down at night.
A clever study (a) estimates the value of a cabinet appointment by looking at stock market prices and prediction markets. That is, they check how much the stock market moves when the probability of a confirmation rises as reflected in prediction markets. They estimate that “the true value of a fully surprising nomination is equivalent to between 5.4 to 7.6 percent”
My experiments on play money discount rates on Manifold are starting to resolve, starting with This question will resolve positively on the 1st of January, 2025 (a). Good to keep in mind for longer markets.
Diary of a Quant has an updated chart (a) of interest rate predictions:
A short letter on Challenges in Forecasting Antimicrobial Resistance (a) discusses updating short-range models in situations where the underlying inputs are hard to predict.
Here (a) is a short and sweet introduction to forecasting time series in R.
The CDC is testing (a) a disease prediction model, as opposed to the more ad-hoc models during COVID, aiming to make it more like weather forecasting.
Jessicata collects some AI predictions made in 2024 (a).
Back in November, the Forecasting Research Institute released a few papers (a) looking at cognitive tests related to forecasting abilities and some ideas to improve judgments of low probability events.
On my side, I wrote a memo on the grain of truth problem (a), where your hypothesis space doesn’t include the correct hypothesis, updated my fermi (a) command line calculator for distributions to add mixtures, reviewed what I’ve done this past year, helped Scott Alexander out with an H5N1 piece (a), and continued running the Sentinel foresight team, which posts weekly minutes (a) looking at big potential risks each week.
Odds and ends
The Long-Term Future Fund has a new grant round (a); the application deadline is February 15th. They have previously funded some forecasting projects, so it might be worth it for forecasting projects looking for funds to apply there.
The Rootclaim vs Steve Kirsch debate and bet on whether vaccines killed more people than they saved is slowly proceeding. Both parties have now deposited $1M each in escrow. The two submissions are here for Rootclaim (a) and here for Kirsch (a).
“The future will be better tomorrow.”
—Dan Quayle (a), US Vice President under George H. W. Bush
This newsletter is sponsored by the Open Philanthropy Foundation.