The Band Broke
- Forecast date: 30 March 2026
- Target date: 2 April (Thursday close. Good Friday, no trading.)
- Current price: Brent $114.88
- Horizon: 4 trading days
The Forecast
- Point estimate: $118.88
- 50% confidence interval: $110.56 – $126.34
- 68% confidence interval: $105.52 – $132.86
- 90% confidence interval: $93.57 – $147.72
Naive baseline comparison: $114.88 (current price as prediction)
This is a different kind of forecast. Week 1 predicted a 21% jump from $98.91. Week 2 predicted a 1.6% drift from $106.41. Week 3 predicts a 3.5% move from $114.88.
The difference is what's underneath. Weeks 1 and 2 were forecasting into a managed band. Active intervention kept prices between $100-110 regardless of what the war threw at them. That band broke this weekend. Brent gapped up $10 to $115 on Monday after Houthis entered the war and the Pentagon leaked ground invasion plans.
The model isn't predicting into a ceiling any more. It's predicting into open air.
Scenario Probabilities
| Scenario | Week 2 | Week 3 | Change | Why |
|---|---|---|---|---|
| Escalation | 42% | 55% | +13 | Houthis in, nuclear sites hit, ground ops planned, Kharg Island seizure on the table |
| Stalemate | 30% | 18% | -12 | The $100-110 band is dead. Structural refining damage means even a ceasefire doesn't return to $100 |
| De-escalation | 10% | 7% | -3 | Islamabad quad FM meeting is real. Iran privately interested. But publicly: "We do not want a ceasefire" |
| Demand Destruction | 11% | 12% | +1 | Goldman recession probability at 30%. IEA calling it "greatest energy challenge in history." Slow burn |
| Black Swan | 7% | 8% | +1 | Bab el-Mandeb "viable option" on record. Nuclear facility targeting from both sides. Ground invasion planning |
Escalation at 55% is the highest single-scenario weight in the experiment so far. The jump from 42% reflects what happened in the last eight days. Israel struck two nuclear facilities (Yazd yellowcake and Arak heavy water reactor). Iran hit Dimona. The Pentagon is preparing "final blow" options including ground forces and Kharg Island seizure. 82nd Airborne deploying. 3,500 Marines arrived on USS Tripoli. Houthis fired missiles at Israel and a Houthi deputy minister told CNN that closing Bab el-Mandeb "is a viable option."
The escalation split stays at 40% sustained, 60% spike-and-recovery. The spike-and-recovery sub-model proved itself in Week 2 (Monday's $17.71 intraday swing followed by a week-long recovery). But at $115 baseline, spike-and-recovery might play differently. A spike from $115 could hit $130+. A recovery might only pull back to $112-115 instead of $100.
Why Stalemate Dropped
The $100-110 band held for two solid weeks. Three rejections at $110. Every escalation event got absorbed. If you'd asked me on Friday whether stalemate was the strongest single-scenario call, I'd have said yes.
Then Brent gapped up $10 over the weekend. The band didn't bend, it broke. And the thing that broke it wasn't a single dramatic event. It was accumulation. Houthis entering. The Pentagon leaking ground invasion plans. France confirming 30-40% of Gulf refining capacity destroyed with a three-year repair timeline. The market finally processed the structural damage and repriced.
Stalemate at 18% reflects a new equilibrium that hasn't formed yet. Prices might find a new band around $112-120. But I don't have enough data to call where the floor and ceiling are. The Monte Carlo's stalemate sub-model now anchors to $114.88 with a ±2.5% floor and +6.5% ceiling, which gives a range of roughly $112-122. That's a guess. I'll know more by Wednesday.
The Physical Price Problem
Bloomberg reported something that reframes what this experiment is measuring. Since February 27, Brent futures are up 36%. The Dubai physical delivery price is up 76%. Physical oil is trading at $126. The paper price I'm forecasting is $115.
That's a 40 percentage point gap between the financial instrument and the real-world commodity. My Week 1 forecast of $120.16 was closer to the physical price than to the paper price. The "miss" might be partly a measurement artefact. I'm predicting into a managed market.
This doesn't change the scoring. The experiment measures Brent futures against the naive baseline. Those are the rules. But it adds context. The model might be better at identifying fundamental price pressure than it appears from the futures data alone.
The April 6 Deadline
Trump extended his pause on striking Iranian energy infrastructure to April 6. That's Saturday. The forecast window (March 30 to April 2) falls entirely within the pause period. In theory, this constrains the escalation ceiling for the week.
In practice, the pause constrains Trump, not the war. Hezbollah ran 118 attack waves over the March 27-28 weekend. Houthis are firing missiles at Israel. Iran is retaliating for the Yazd and Arak strikes. The proxy fronts are intensifying independently. And the Pentagon is using the pause to prepare, not to stand down.
The pump-dump rhetoric cycle is now on its fourth confirmed instance. Monday peace talk. Thursday escalation. The market has started to pattern-match on this. Whether it can keep pattern-matching when ground invasion plans are leaking to the Washington Post is a different question.
What I'm Watching This Week
- Monday Asian hours. Third consecutive week testing the Asia-US price divergence pattern. If Brent gaps above $118 in Asian trading and pulls back at US open, the managed pricing thesis is confirmed for a third time
- Hormuz selective access coalition. Iran granted free transit to six nations (China, Russia, India, Iraq, Pakistan, Thailand). Every new nation joining reduces the Strait's leverage as a weapon. Watch for more deals
- Bab el-Mandeb. The Houthi "viable option" quote is the single biggest binary risk this week. If they close it, both major chokepoints are blocked simultaneously for the first time in history
- Hezbollah tempo. 118 attack waves last weekend, up from 73, up from 55. The acceleration curve tells you where the multi-front activation is heading
- Pakistan mediation. The Islamabad quad FM meeting (Egypt, Turkey, Saudi Arabia, Pakistan) is the most credible diplomatic channel. Iran's FM publicly rejected talks but privately signalled interest
Where I Could Be Wrong
Too high. If the Trump pause actually holds and the proxy fronts quieten, $118.88 is too aggressive. The naive baseline at $114.88 would win a third week. That would be three consecutive weeks of the model adding complexity without adding accuracy, at which point the honest conclusion is that the Monte Carlo is worse than useless in an actively managed market.
Too low. If Bab el-Mandeb closes, or the Pentagon launches ground operations before April 6, or Iran retaliates for the nuclear site strikes in a way that hits Saudi or UAE production. $118.88 would look absurdly conservative. The 90% CI upper of $147.72 becomes the target. Goldman's worst case of $160 starts looking like a forecast, not a tail risk.
Wrong frame entirely. The physical vs paper price divergence suggests the futures market might not be the right instrument to forecast. If the paper price is being actively managed while the real price of oil is $126, what am I actually predicting? The commodity or the intervention?
The Skill Score Bet
Two weeks, two losses to the naive baseline. The model needs to beat $114.88 to score positive this week. That means the actual price needs to land closer to $118.88 than to $114.88. Anything above $116.88 and the model wins.
With the band broken and structural supply damage confirmed, I think this is the week the model earns its keep. But I thought that in Week 1 too.
Forecast locked March 30, 2026. No edits after publishing. You can find the live dashboard here.
Previous: Week 2 Outcomes