Source: task4-all-weather/output/phase_a_v01a/task4_phase_a_v01a_ic_handoff.md
Last modified: 2026-05-18 15:07:53 · 5,344 bytes

Task #4 Phase A v0.1a IC Handoff

Status: ready for IC pack v1 drafting.

Scope

Phase A implements the final-acked v0.1a universe:

- 27 instruments / about 25 ERC entities.

- ETF-only, unlevered Path X.

- Two grouped entities:

- R1 growth_innovation_group: 159915.XSHE 创业板ETF + 588000.XSHG 科创50ETF, fixed 50/50.

- R2 ai_compute_chain: 515880.XSHG 通信设备 + 512480.XSHG 半导体, fixed 50/50.

- Multi-stage covariance ERC:

- Stage 1: entity ERC inside each sleeve.

- Stage 2: sleeve ERC using sleeve sub-portfolio covariance.

- Cross-sleeve overlap handled by Option A: monitor underlying risk-source RC, do not cap or restructure in v0.1a.

Baseline Results

Segment Annual Return Vol Sharpe Max Drawdown
IS 2015-2023 4.29% 6.03% 0.71 -10.52%
OOS 2024 pre-9/24 -6.56% 8.12% -0.81 -9.22%
OOS 9/24-2025H1 21.20% 9.41% 2.25 -3.30%
OOS 2025H2-now 17.05% 7.57% 2.25 -3.50%
FULL 5.52% 6.58% 0.84 -13.65%

Primary Premium RC

Average risk contribution:

- Equity: 56.05%

- Inflation: 35.62%

- Rate: 5.68%

- Credit: 2.65%

Read: v0.1a remains equity/inflation dominated. Rate RC is structurally low despite latest notional duration weight being high, which is the expected unlevered risk-parity failure mode for low-vol assets.

Underlying Risk-Source RC

Average risk contribution:

- Broad A equity: 22.61%

- Commodity inflation chain: 19.96%

- R3 defensive equity: 18.40%

- Growth + AI compute combined: 17.40%

- Physical chain: 15.66%

- Green manufacturing growth: 12.44%

- HK internet + A growth: 10.08%

- Rate duration: 5.68%

- Productivity chain: 4.35%

- Credit carry: 2.65%

Read: cross-sleeve growth/AI overlap is present but not the binding issue. growth_ai_compute_combined_risk is below the 30%-35% escalation threshold, so Option A monitoring is acceptable for v0.1a.

Latest Portfolio Shape

Latest date: 2026-05-15.

Largest scaled entity weights:

- treasury_10y: 63.10%

- treasury_30y: 16.59%

- consumer_staples: 1.93%

- hs300: 1.84%

- utilities_power: 1.75%

- free_cash_flow: 1.25%

- Most other entities: roughly 0.4%-1.2%.

Read: the current unlevered implementation is duration-heavy in notional terms. This is not an implementation bug; it is the expected result of unlevered ERC allocating high notional to low-vol duration assets while still producing low rate risk contribution.

Main Findings

1. The framework direction is validated enough for IC review.

- Full-sample Sharpe is positive at 0.84.

- Post-2024-09-24 periods are strong.

- Vol is in the 6%-8% target range.

2. Path X is not the final framework.

- The unlevered version under-represents rate risk contribution.

- It uses high notional duration but does not create enough rate premium RC.

3. Stage 4 leverage is the main v0.2 engineering priority.

- The data supports Qian's point: unlevered risk parity over-allocates low-vol assets by notional and still under-delivers their risk premium.

- v0.2 should evaluate TL/T/TF/TS treasury futures or other funded duration implementation.

4. Cross-sleeve overlap does not require immediate restructuring.

- Growth + AI compute combined RC is 17.40% average and 12.94% latest.

- Keep monitoring in IC pack and production reports.

5. 2024 pre-9/24 stress remains a weakness.

- OOS pre-9/24 annualized return was -6.56%, MDD -9.22%.

- This supports the v2 roadmap for crisis-mode triggers and dynamic risk allocation.

IC Pack v1 Mandatory Disclosures

v0.1a is a China-localized stepping stone based on Qian / Bridgewater / Lin Yingying principles. It is not the complete final implementation.

Explicit gap disclosure should include:

- Stage 4 funded leverage not implemented.

- Single-level integrated ERC not implemented.

- Dynamic Risk Allocation overlay not implemented.

- TIPS / inflation-linked bond market gap in China.

- Crisis-mode hard triggers not implemented.

- Full sector RP not implemented.

- Simplicity gap versus Bridgewater's original broad-instrument all-weather design.

v0.2 / v2 Engineering Roadmap

Priority order:

1. Stage 4 funded duration leverage.

- Evaluate TL main contract and 10Y/30Y duration mix.

- Target: lift rate RC toward 25%-30% while controlling overall vol.

2. Crisis-mode hard triggers.

- Candidate inputs: M2, central bank rate moves, credit spread widening, global liquidity tightening.

- Goal: reduce 2024H1-style stress drawdown.

3. Dynamic Risk Allocation overlay.

- Inflation tilt, deleveraging, rebound/re-risking rules.

- Keep capped and auditable.

4. Single-level integrated ERC diagnostic.

- Run as research benchmark against multi-stage ERC.

- Do not replace the four-sleeve framework until IC sees trade-offs.

5. Full sector RP / sector-balanced equity sleeve.

- Evaluate after v0.2 leverage and crisis-mode work.

Output Files

- task4_phase_a_v01a_summary.md

- task4_phase_a_v01a_metrics.csv

- task4_phase_a_v01a_daily_returns.csv

- task4_phase_a_v01a_monthly_scaled_weights.csv

- task4_phase_a_v01a_risk_contribution.csv

- task4_phase_a_v01a_primary_premium_rc.csv

- task4_phase_a_v01a_audit_bucket_rc.csv

- task4_phase_a_v01a_underlying_risk_source_rc.csv

- task4_phase_a_v01a_price_metadata.csv