The 28 most important questions about Iran

Yann Rivière

I’m a professional forecaster for Metaculus, part of Swift Centre and RAND, and one of the top ranked forecasters and question writers on Metaculus in 2025. Writing good questions is one of forecasting’s main bottlenecks: it is time-consuming and more complex than it appears.

The Iran conflict is the most important global issue today. Many people turn to prediction markets like Polymarket or Kalshi because they have demonstrated impressive accuracy. Nonetheless, these platforms don’t provide a comprehensive view, due to gaps in market coverage and definitional issues.

Which questions would we ask if we were unconstrained by trading volume, settlement contracts, or any factor that shapes which forecasts get produced on these platforms?

Over the past few weeks, I’ve been working with Mantic to build a dashboard of 28 questions I consider most worth answering. Mantic normally uses AI to generate forecasting questions but I’ve adopted a centaur approach, combining my (human) question writing skills with Mantic’s AI forecasting capability. Although AI forecasting accuracy has caught up to professional levels, asking the right questions is an art that AI has yet to master.

This post walks through my questions on the Strait of Hormuz, the ceasefire, military engagement, regime change and energy prices among others, to show what unconstrained forecasting on this conflict looks like. I’ll compare throughout to Polymarket’s coverage.

Strait of Hormuz

The most important question is when traffic through the Strait returns to pre-conflict levels. The economic stakes are obvious, to the tune of over $10B per day. The less discussed concern is precedent: the longer this lasts, the more similar moves become possible at the Malacca, Turkish or Taiwan Straits. Likewise, a toll regime in Hormuz could open the door for every chokepoint actor. Here are the questions I asked, with Mantic's forecasts:

With a generous reading, 4 out of 6 of these questions could be considered as covered by Polymarket but the criterion differences make it closer to 2 out of 6.

Their most informative question is the average number of ships transiting by the end of May, but there is no equivalent for June. The same gap shows up on the max number of ships on a single day by the end of month. Its“traffic returns to normal” series has better granularity across dates (May 15, end of May, end of June, July 31, December 31). However, they’ve set a lower threshold than I would have: a 7-day moving average of 60 transits hit on any single date in the window. With around 2,000 ships currently stranded, a few well-organized days could clear that threshold before the Strait re-clogs; 420 ships across seven days is enough, while pre-conflict daily traffic ran above 100, sometimes up to 130. A resolution could happen quickly and satisfy the criterion before things get to a halt again. Finally, their toll market is both short horizon and resolves to Trump publicly accepting Iran charging fees for transit. Given how often his announcements have detached from reality, that’s an obvious case where a resolution could split from the question it appears to track.

Ceasefire or definitive peace agreement

Reaching a definitive ceasefire or a peace agreement is obviously important as it would impact all forecasts in the dashboard. Here are the questions I asked, with Mantic's forecasts:

These are obvious questions to ask, so Polymarket’s good coverage (3 out of 4) is unsurprising, with caveats.

The US-Iran ceasefire was agreed April 8 and extended indefinitely on April 21. Polymarket ran an extension market in the first weeks tracking specific extension dates, but the series stopped after April 22, leaving roughly a month with no live market on the most active diplomatic question of the conflict. A daily-resolution ceasefire market appeared on May 20, only five days before the ceasefire actually broke. In the intervening window, anyone using Polymarket to know whether the ceasefire was holding had no explicit live signal.

They were also focused on very specific attack mediums that would breach the ceasefire and did not have a general question about a breach of the ceasefire, something that is quite likely early on. A separate gap appeared in the first hours of the April 8 ceasefire: it was unclear whether it covered Israel or only the US, and no market addressed that until April 17, despite it being a very plausible reason for early breakdown. Polymarket now has an Israel x Iran permanent peace deal market and a US x Iran one, but the first market only exists for May 31 and June 30 and does not look at ceasefires. The newest ceasefire extension market resolves on a Trump announcement rather than on whether hostilities have actually resumed, which is the same appearance versus reality problem as the Strait markets.

Military engagement

The evolution of the military side of the conflict is a variable that shapes nearly every other forecast on the dashboard. The ceasefire has held the picture frozen for weeks but could break at any time. Plausible deniability for Gulf-state involvement, restraint on energy infrastructure, and the absence of ground forces have been the operative norms so far, but each could break abruptly.

Polymarket currently covers none of these questions.

Regime change

Regime change is one of the hardest topics for question-writing, and current markets illustrate why. The main Polymarket regime fall market requires that core structures of the Islamic Republic such as the Supreme Leader or the Guardian Council be dissolved or replaced by a fundamentally different governing system.

The problem is that Iran is already, functionally, an IRGC-dominated state. They control vast shares of the economy, hold an increasing share of senior political positions, and are by far the strongest military force in the country. They backed Mojtaba Khamenei as Supreme Leader despite his lacking the formal qualifications, and there is reason to believe they are using his current condition to consolidate further power. The most consequential outcome by far under this criterion is whether the IRGC loses power as the end of the Supreme Leader or the Guardian Council would not by themselves change the daily reality of the regime.

There is therefore only one question which matters:

Energy prices

Oil briefly moved above $120 a barrel and could go higher if the situation deteriorates again. This matters everywhere, but US midterms are particularly sensitive to gas prices. The LNG capacity loss will hit Europe and Asia hardest as winter demand is the next major load on the system; in economies already carrying heavy debt, the damage won't stay inside the energy market.

Polymarket only covers the US strategic reserve question. They do have some markets around oil and natural gas prices and oil production but all are constrained to the same short horizon with almost nothing extending past June. Mantic's coverage runs much further out, which is where most of the interesting questions sit.

Other questions

Polymarket covers 3 out of 5 of these questions, with the usual time horizon limitations.

---

The questions walked through here weren’t adversarially chosen against prediction markets. Still, even with a generous reading, only 12 of the 28 are covered by Polymarket at all. Once you factor in the resolution criteria and time-horizon issues, it drops to roughly a quarter being actually covered.

Polymarket’s coverage on the topics it does run is mostly fine but sometimes has structural issues such as the criteria for the Strait returning to normal, the regime change question or the Israel framing inside the ceasefire. It is lacking on long-horizon markets, as most don’t currently extend past June. Many decisions a reader might make about Iran would look different given a forecast six months or a year out.

This is not unique to prediction markets. Exchanging with other top forecasters is one of the best ways I’ve found to sharpen my read on the world but it also runs into its own limits: the pool is small, and our time, interests and knowledge are all limited too. It can be hard to get that essential back-and-forth once the questions get niche enough, even on topics as debated as the Iran conflict.

With Mantic now regularly competing with the best human forecasters, I was curious how much it could help as an iterative forecasting partner that goes well beyond what the frontier LLMs offer. So far the experience has been very positive. I haven’t exchanged any less with other pros, but I now have a very competent control group for every forecast I’ve wondered about, no matter how niche or complex the subject. I believe this means both sharper forecasts and significantly more of them.