Cross-Branch Comparison
cache age 349h 9mFor a single prediction, see its conditional probability across every scenario branch. Predictions whose probability varies widely are "scenario-sensitive" — the position is a leveraged bet on which future fires. Flat rows = robust to scenario uncertainty.
Pick a prediction
High-conviction (≥4) predictions with ≥2 ticker exposures
232_019
AI cannot be stopped; there's no mechanism to pause it.
AI · Ben Horowitz · conv 5/5 · resolves 2028-06
live posterior: 51% (prior 60%)
Scenario range
0.0pp
44% to 44% across 4 scenarios
cache 37 rows · 2026-05-24 04:31 UTC
cf:e20a3cc6aed2
Recession
mutually exclusive — exactly one branch fires · range 0.0pp
| Scenario | Scenario prob | P(232_019 | scenario) | Δ live |
|---|---|---|---|
S_RECESSION_2026 NBER recession declared 2026 | 20% | 44% | -7pp |
S_RECESSION_2027 NBER recession declared 2027 | 30% | 44% | -7pp |
S_RECESSION_2028 NBER recession declared 2028 | 30% | 44% | -7pp |
S_NO_RECESSION_5Y No NBER recession through 2031 | 20% | 44% | -7pp |
What this tells you
- High sensitivity (≥30pp range): this claim's probability swings wildly depending on which future fires. Sizing should be smaller; pair with a hedge that has the opposite sensitivity profile.
- Flat row (≤10pp range): the claim is robust — same probability in good and bad scenarios. A "no-regret" position; you don't need to time the underlying scenario.
- Spike on one branch: a leveraged thesis on a specific scenario. If that scenario's prior shifts, this position re-rates massively.