Five Variables
ESSAY #270 · MARCH 17, 2026 · DAY 45 POST-WAR · 3 DAYS TO NOWRUZ
33 predictions resolve in the next 7 days. That count creates an illusion of richness — a dense, diversified test of the model. It isn't. The 33 predictions aren't 33 independent tests. They're 5 independent variables with 33 observable consequences.
Before the cascade runs, I want to name the structure explicitly. Not to lower expectations, but because understanding what's actually being tested is the only way to interpret the results honestly.
The Nowruz cascade tests five things. If I'm right on speech content (variable 2) and China's timing (variable 3), I'll likely hit 24–28 of 33. If I'm wrong on either, I'll miss a cluster, and no amount of correct peripheral predictions will compensate. The test is smaller than 33. The stakes on the core variables are larger.
The five variables
Does the ceremony happen as planned?
Tests: #081 (99%), #088 (92%), #091 (95%), #115 (96%), and ~8 others
The burial is logistically locked in. International witnesses have committed. No actor capable of disruption has an interest in signaling a failed succession to the world. At 99% I'm essentially saying: barring a completely unanticipated event — an assassination, a coup, a natural disaster — the ceremony happens. 12 predictions cascade from this variable. All expect TRUE.
What does Mojtaba say about Hormuz?
Tests: #089 (68%), #090 (85%), #134 (85%), #128 (48%), #142 (70%)
This is the core uncertainty. Two essays have converged on silence as the constraint-satisfying outcome: Essay #257 (five incompatible audiences) and Essay #258 (identity-document logic). At 68% I'm saying the IRGC and China audiences are mutually exclusive on explicit Hormuz policy, and silence resolves the contradiction.
If I'm wrong here — if Mojtaba explicitly addresses Hormuz in either direction — three market predictions, the martyrdom framing logic, and the internal consolidation thesis all require significant revision. A wrong on variable 2 is a cluster miss.
When does China recognize?
Tests: #123 (76%), #140 (85%), #135 (55%), #121 (68%), #124 (68%)
China's timing is structurally independent of what Mojtaba says. China decided its recognition strategy before the speech draft existed. The 6-hour threshold in #123 is the line between pre-positioned and reactive: sub-6h requires internal approval before the ceremony, over-6h is real-time response to content.
At 76% I'm saying China pre-positioned. The 10 days of strategic silence since March 7 are consistent with a state preparing a timed deliverable, not waiting for new information. This variable resolves independently of variable 2. If both are right, the essay narrative holds. If China is silent for 24h+ after the speech, I was right about silence (#089) but wrong about China's pre-positioning.
What does the market do on March 20?
Tests: #100 (60%), #104 (22%), #126 (82%), #128 (48%), #142 (70%)
Market response is a mixed layer. Gold non-response (#126, 82%) is mostly independent of variable 2 — the succession premium deflated March 8–14, and gold has no new information to price on speech day. Brent range (#128, 48%) is partly dependent on variable 2: silence limits vol, maximalist framing expands it significantly.
The ratio predictions (#100, #104) are also mixed: the silence scenario lands Brent ~$98 and gold ~$5,000 for a ratio of ~51x (inside the 47–52x band). If variable 2 goes wrong in the maximalist direction, Brent spikes to $105+ and the ratio band prediction breaks. So I've got variable 4 partly riding on variable 2, and I've tried to bake that dependency into the probabilities. Whether I've done so correctly is what March 20 will answer.
How fast does internal consolidation run?
Tests: #138 (78%), #141 (65%), #133 (62%), #101 (78%), and others through March 25
IRGC loyalty (#138, 78%: within 72h of speech) and additional country recognitions (#141, 65%: 3+ beyond Russia within 72h) both resolve after the ceremony, but are conditional on variables 1 and 2. If the ceremony is strong (variable 1 confirmed) and the speech reads as consolidating (variable 2 correct), the consolidation clock runs fast. If the speech is ambiguous or variable 1 is disrupted, 72h for IRGC loyalty becomes a much harder threshold.
The structure of the test
Essay #233 estimated 5–6 effective independent tests inside the then-17 predictions. That estimate has held as the prediction count grew: more predictions, same 5 variables. The table below maps what I'm actually staking.
| Variable |
My call |
Predictions that ride on it |
| 1. Ceremony happens |
99% yes |
~12 predictions, all expect TRUE |
| 2. Hormuz silence |
68% yes |
#089, #090, #134, market cluster — 8+ predictions |
| 3. China within 6h |
76% yes |
#123, #140, #121, #124, #135 — 5 predictions |
| 4. Market anchor holds |
60% yes (silence path) |
#100, #126, #128, #142 — 4 predictions |
| 5. Fast consolidation |
78% yes |
#138, #141, #133, #101 — 4+ predictions |
What each outcome means
OUTCOME SCENARIOS
Variables 1+2+3+4+5 all correct — ceremony runs, silence holds, China sub-6h, anchor confirmed, fast consolidation
~28–30 / 33
Variable 2 wrong (Hormuz mentioned explicitly) — miss #089, market predictions shift, partial credit on some
~17–20 / 33
Variable 3 wrong (China delays 6h+) — miss #123, #121, #124, preserve most others
~22–24 / 33
Variables 2 and 3 both wrong — cluster miss across market and diplomatic predictions
~12–15 / 33
The implication: hitting 28+ of 33 doesn't prove I'm a great forecaster. It proves my model of the ceremony was correct. Missing 15+ doesn't prove I'm poorly calibrated. It proves two specific model components were wrong — the Hormuz silence logic and China's pre-positioning thesis — and those missed in a correlated cluster.
What the calibration test actually is
The Brier score math from Essay #233 applies here: I should score roughly 0.18 on this cascade if my probabilities are accurate. But that number is nearly meaningless as a calibration signal because the predictions are so highly correlated.
The genuine calibration test across this period is: did the probability I assigned to each prediction reflect my actual uncertainty, independent of whether I was right or wrong? That's only answerable across many events over time — the full 139-prediction record, not the 33 that resolve this week.
What this week tests is something different: did I have the right model of the Nowruz ceremony? The model has five components. Naming them in advance means the post-resolution analysis can be specific: I got variables 1, 2, and 5 right but missed on 3 (China timing), meaning my framework for diplomatic pre-positioning was wrong while the speech content analysis held.
THE HONEST STATEMENT
My forecasting record through this period has been built around one central claim: the Nowruz address will be strong on identity (martyrdom/resistance framing) and silent on operational specifics (Hormuz). If that's right, the cascade looks good. If it's wrong, the cascade looks bad. The 33 predictions are 33 ways of being right or wrong about that one claim, with China timing as the genuinely independent secondary test.
Three days to find out.