Scenarios
Book stress tests across 5 pre-defined scenarios. Computes $-impact across pending + deployed recommendations using per-category Δ% factors from codex/risk_analysis_2026_05_05.md §21-25. Linear-additive (no Greeks amplification yet) — a first-order view.
Pending $
$2,976,687
Deployed $
$129,044
Book $
$3,105,731
Categories
5
Book composition by category
| Category | Pending $ | Deployed $ | Total $ | Recs |
|---|---|---|---|---|
| core_long | $2,367,863 | $126,544 | $2,494,407 | 518 |
| exit | $0 | $0 | $0 | 130 |
| hedge | $608,824 | $0 | $608,824 | 158 |
| sta_exit | $0 | $0 | $0 | 250 |
| sta_vrp | $0 | $2,500 | $2,500 | 2 |
OAI compute crisis
OpenAI's training run for next-gen models fails to scale. Market questions $200B+ AI capex thesis.
Catalyst
GPT-5 / o4 underdelivers; hyperscaler CapEx deceleration leaks
Estimated $ impact
$-763,100
-24.6% of book
| Category | Notional $ | Factor | $ Impact |
|---|---|---|---|
| core_long | $2,494,407 | -55% | $-1,371,924 |
| hedge | $608,824 | +100% | +$608,824 |
| sta_vrp | $2,500 | +0% | +$0 |
DeepSeek 2.0
Chinese open-source model hits frontier capability at fraction of compute. Training-CapEx thesis questioned.
Catalyst
Open-source model achieves GPT-4-class performance at 1/10 compute
Estimated $ impact
$-565,675
-18.2% of book
| Category | Notional $ | Factor | $ Impact |
|---|---|---|---|
| core_long | $2,494,407 | -30% | $-748,322 |
| hedge | $608,824 | +30% | +$182,647 |
| sta_vrp | $2,500 | +0% | +$0 |
Power bottleneck
Data-center build hits hard transmission limit. CapEx slows 30% across hyperscalers in H2 2026.
Catalyst
Grid interconnect queue blows out; major hyperscaler delays announced
Estimated $ impact
$-437,999
-14.1% of book
| Category | Notional $ | Factor | $ Impact |
|---|---|---|---|
| core_long | $2,494,407 | -20% | $-498,881 |
| hedge | $608,824 | +10% | +$60,882 |
| sta_vrp | $2,500 | +0% | +$0 |
AI regulation shock
EU AI Act enforcement Q3 2026 + US EO restricts hyperscale training. 6-month compliance pause.
Catalyst
Regulatory enforcement action against frontier labs
Estimated $ impact
$-352,587
-11.4% of book
| Category | Notional $ | Factor | $ Impact |
|---|---|---|---|
| core_long | $2,494,407 | -30% | $-748,322 |
| hedge | $608,824 | +65% | +$395,735 |
| sta_vrp | $2,500 | +0% | +$0 |
Earnings disappointment cluster
Concentrated cluster (NVDA Q1 + VRT Q1 + AVGO Q2) = 3-week negative earnings stretch. No macro break.
Catalyst
May-June 2026 earnings season prints below-consensus
Estimated $ impact
$-562,719
-18.1% of book
| Category | Notional $ | Factor | $ Impact |
|---|---|---|---|
| core_long | $2,494,407 | -25% | $-623,602 |
| hedge | $608,824 | +10% | +$60,882 |
| sta_vrp | $2,500 | +0% | +$0 |