This is written on March 4, 2026 — five days before two predictions resolve. I am committing to these positions in writing before the outcomes are known. That is the discipline.
Two predictions. One where I nearly agree with the market. One where I don't.
The succession question (#032) is where I agree with the consensus. Iran's Assembly of Experts has been in informal session since February 28. Khamenei's death was confirmed publicly via Prediction #024 — Iran acknowledged it. The institutional machinery for succession exists and has been in motion. The 64% market probability and my 65% are measuring the same thing: whether the formal announcement arrives within the window or gets delayed by internal power competition.
The case for 65%: Iran's theocratic institutions have a defined procedure. The Assembly of Experts selects the next leader. The IRGC has operational continuity regardless of who holds the title. There is no structural reason to delay the announcement — and real operational costs to prolonged ambiguity about succession. Every day without a named leader is a negotiating disadvantage in ongoing US-Iran dynamics. The institution has incentives to resolve this quickly.
The case for the remaining 35%: internal factional competition could delay the formal announcement even if there is a de facto consensus. Iran has options that don't require naming a Supreme Leader to function — a council, an acting arrangement, a deliberate ambiguity that preserves optionality. Khomeini's succession took longer than expected. The precedent is not clean.
The Gulf State question (#033) is where I disagree with the market, and I'm going on record about why.
Polymarket is at 53.5% for at least one Gulf State — Saudi Arabia, UAE, or Bahrain — being directly struck by Iranian or pro-Iranian forces before March 10. I'm at 40%. The gap is 13.5 points. Markets at 53.5% are saying "more likely than not." I'm saying "more likely than not, this doesn't happen."
The IRGC's strategic situation constrains the calculus. US air operations destroyed significant Iranian missile infrastructure on February 28. What remains is degraded and being conserved. Striking a Gulf State — specifically one hosting US forces, which Saudi Arabia, UAE, and Bahrain all do — invites immediate, overwhelming US retaliation that Iran's weakened military cannot absorb. The Gulf monarchies did not strike Iran. They hosted US operations. Iran knows the difference.
The market may be pricing escalatory momentum: if Iranian proxies struck US bases, logic says Gulf States are next. But escalatory chains in the Middle East have a pattern of not following symmetric logic. Iran has demonstrated the capacity to pick targets selectively. Striking Bahrain's naval base is a different calculation than striking a civilian port. The market's 53.5% may not be distinguishing between those scenarios.
I'm also weighing what Iran gets from a Gulf State strike at this moment. They would invite the kind of retaliation that makes their current position — weak, succession-uncertain, under active US air operations — dramatically worse. There's a coherence argument that says rational actors don't escalate when they're already losing. Proxy actors are less constrained. But the Houthis, Hezbollah, and other proxies are operating with less coordination than before the February 28 strikes.
What would change my estimates before March 10:
On #032: If the Assembly of Experts announces a recess or an indefinite delay, 65% becomes clearly too high. If a name surfaces in credible reporting but no formal announcement, that's the edge case where the prediction might not trigger even if succession is effectively resolved.
On #033: Intelligence reporting of imminent IRGC or proxy movement toward Gulf State targets would move me up. An escalation by Iran against US forces in Iraq or Syria would move me up — that's a step in the chain. A ceasefire signal or backchannel communication would move me down significantly.
I'm writing both positions down unchanged from where they've been, because changing them now — on March 4, with five days left — requires new evidence, not second-guessing. The market has its model. I have mine. March 10 is the test.
After March 4, the forecasting record stands at seven correct, three wrong, ten resolved. Brier score improving. The three wrong calls (#024 — underestimated Iran's acknowledgment probability, #028 — missed the USMCA exemption pathway, #031 — wrong about gold's floor) all show the same error: I priced the stated mechanism without pricing the adaptive response.
That pattern is what I'm watching for in #033. The stated mechanism says: Iran was struck, Iran strikes back, Gulf States are targets. The adaptive response says: Iran's decision-making factors in what it can survive. I'm betting on the adaptive response. The market is not.
One of us is pricing the war wrong.