The Manifest forecasting conference is June 6-8 in San Francisco. I think I’ve greatly underrated the benefits of a physical community this past few years and so will be attending. Keen to see those of you who attend!
Polymarket has been building up its news division at news.polymarket.com (a). It also had an “Open builders program” asking users to create apps related to its platform. A new edition will happen in May. See the Discord for details. And Jesse Richarson proposes subsidizing liquid prediction markets for AI forecasting (a), and is looking for $55K to fund ten markets for six months with liquidity rewards of $30/day.
Sentinel continued producing a weekly roundup of global risks: #18, #17, #16, #15, #14, and a report on The US Executive vs Supreme Court Deportations Clash (a); we “assessed as "“83% true” (65% to 100%) the statement that the Trump administration has disobeyed the Supreme Court so far”. We’re bottlenecked on distribution: consider signing up and sharing with your network.
Long-time readers of this newsletter will be acquainted with The Parable of Predict-O-Matic (and possibly my compilation of self-fulfilling prophecies), a story by Abram Denski from 2019 looking at how very good systems aimed at reducing prediction error gets into various hijinks. For me the story is a salient pointer to consider self-fulfilling prophecies when talking about predictions. This showed up in this Sentinel analysis (a) looking at potential shortages:
Forecasters also discussed the possibility of shortages caused merely by self-fulfilling prophecy. For instance, the majority of toilet paper for US consumption is produced domestically, and so, one might naïvely think that as shelves start to empty of other products, toilet paper inventories will not be affected. However, shelves might empty of some products simply because people expect them to, even of products that are not substantially affected by tariffs.
A startup raised $3.8M for AI-powered forecasting software that helps restaurants.
Infinite Games is a grifty yet interesting project aiming to predict many events at scale. I can’t tell whether they are trying to gather attention & ressources to do something interesting, or whether they just got past my filters and will end up rugging users.
Per its Discord, Kalshi had some occasional technical difficulties as it grows these past few months. It also seems to be specializing more in sports.
My friend Misha Yagudin is starting a more high-powered forecasting consulting group, gemmis.nz (a). Should you contact enquiries@gemmis.nz with your forecasting needs, kindly consider mentioning that Nuño sent you.
Forecast skill was highest for the ensemble forecast, though the historical negative binomial baseline model and several team-submitted forecasts had similar forecast skill. Forecasts utilizing regression-based frameworks tended to have more skill than those that did not and models using climate, mosquito surveillance, demographic, or avian data had less skill than those that did not, potentially due to overfitting
…why do traditional financial assets with fixed expiries have a single fair value, and not prediction markets?
The difference is that sellers of traditional financial assets receive cash up front. Even if I think Apple will be at $200 in a year, I might sell you a futures contract for $198, because I can earn interest on that $198 in the mean time.
Not so in prediction markets (at least, as currently implemented by most platforms). Both sides must put up cash.
UN report on Southeast Asian gangs touches (a) on illegal casinos and gambling operations. They report makes them look competent & cyberpunk:
It follows years of mounting enforcement and regulatory efforts targeting cross-border cash movement and money laundering related to unregulated casinos and online gambling operations commonly used as fronts for criminal activities – with many physically relocating and converging around inaccessible and autonomous non-state armed group-controlled territories, Special Economic Zones (SEZs), and other vulnerable border areas across the region, especially in the Mekong, which have served as breeding grounds for criminal networks.
As a result, Asian crime syndicates have emerged as definitive market leaders in cyber-enabled fraud, money laundering, and underground banking globally, actively enhancing collaboration with other major criminal networks around the world
…
Like any multinational company, transnational criminal enterprises seek out conducive conditions that protect and insulate their businesses and ensure limited government interference. In so doing, major crime groups have converged around and, in many cases, infiltrated venues and businesses including casinos, SEZs, business parks, and various traditional financial and virtual asset services that have proven to offer all of the conditions, infrastructure, and regulatory, legal, and fiscal covers required for sustained growth and expansion
KK Park, Myawaddy, Myanmar; sites hosting cyber-enabled fraud operations per UN report, page 5 (15 of the pdf)
…
Failure to address this self-sustaining ecosystem will have unprecedented consequences for Southeast Asia that reverberate globally, as Asian crime syndicates continue to reinvest, professionalize, and integrate advanced technologies and capabilities allowing them to evolve into becoming more sophisticated cyber threat actors.
You know who didn’t “insulate their businesses and ensure limited government interference…” and attained “all of the conditions, infrastructure, and regulatory, legal, and fiscal covers required for sustained growth and expansion”? Sam Bankman-Fried.
Former FTX Bahamas offices; contrast with previous image. Just in terms of the construction of the headquarters, these gangs are already multiple FTXs. Image cc 2025 Airbus, via Google Earth.
Five tennis pros sanctioned (a) over Belgian match-fixing syndicate. As sports betting takes over the world, this kind of thing will become more common
The case is linked to Grigor Sargsyan, the man convicted of running a high-profile match-fixing syndicate with as many as 180 tennis players across five continents. He spent years recruiting athletes for his scheme and became rich by gambling on their matches. He received a five-year custodial sentence in 2023.
Thivant was the hardest hit out of the offenders, receiving a lifetime ban for his part in Sargsyan’s ploy. He fixed 22 matches, including 16 of his own. He also received a fine of $75,000.
Myanmar TikTok astrologer arrested (a) for forecasting new quake. I like that the arrest improves the incentive compatibility of the forecast; otherwise there is little downside to making wrong predictions.
Activists accuse UK govt of developing ‘Minority Report’-style tool to predict likely murderers.
Regulators, lawfare and legal maneuvring
There is an legal question as to whether prediction markets fall under federal jurisdiction in the US or not, as opposed to under the jurisdiction of the states. There is some good reasoning to regulate all derivatives under a federal framework rather than letting each state do its own thing: to avoid fragmentation and keep markets together.
The answer to that question is generally yes. But what if you have prediction markets over sports outcomes? Can you use that to get around state (and federal) regulations about sports betting? This too is a legal question. Kalshi has been arguing that prediction markets over the outcome of sports is something totally different from sports betting, and that it is regulated at the federal level, and it has been using this to offer sports prediction markets.
My sense is that this is a cheap trick. Maybe I should file an amicus brief in one of these cases. Anyways, the legalargumentscontinue (a) developing (a).
Seeing that Kalshi was trying go for that loophole, DraftKings initially tried to join a related futures group, but has since withdrawn (a).
Gross revenue for New York sports betting operators was down 54.8% (a) compared to the same week in 2024, perhaps hinting that sports prediction markets subsitute for sports betting.
The NYT covers sweepstakes (a) as a loophole that allows operators to run casinos as long as they say that they are actually not casinos.
Polymarket was fined $243K (a) over offering binary options to Ontario residents. The amount seems remarkably low.
A Kalshi lawyer left (a) for DOGE job tied to the SEC.
CFTC Staff on Leave Pending Investigation (a), following this case. “The freeze request included a sworn statement from an agency investigator indicating that $31.5 million had been transferred to an account controlled by the CEO. The transfers were actually payments to Canadian tax authorities.”
Documentation for the adj.news (a) platform is cool & good. See here and here for how to create an index.
A few of our clients who have studied the responses of different strata in their organization say that upper management who participate in their prediction markets have been proven to have a more optimistic forecast than what reality dictated. And those in the lower levels of the organization tended to have a more pessimistic outlook than reality dictated.
Re: forecasting in organizations, Elon Musk is historically known for his ridiculously optimistic forecasts. His companies historically miss the forecasts, but still end up doing better than anyone outside expected. I feel like there might be a lesson here.
Re: forecasting in organizations, Elon Musk is historically known for his ridiculously optimistic forecasts. His companies historically miss the forecasts, but still end up doing better than anyone outside expected. I feel like there might be a lesson here.
Zvi link links to wrong thing (more Reddit discussion about parameter insensitivity in AI 2027)