Source: task4-all-weather/output/task4_v0pp_cap_sensitivity.md
Last modified: 2026-05-17 15:11:36 · 1,949 bytes

Task #4 v0++ Single-Asset Cap Sensitivity

Scope: quick patch on v0+ funded baseline. Stage 4 cash/short-bond zeroing remains unchanged; after redistribution, cap single asset at 25% / 30% / 35% and spread excess to same-sleeve assets first, then cross-sleeve assets by inverse-vol capacity.

IS 2015-2023 Results

Variant Ann Return Vol Sharpe Max DD Latest Max Asset Historical Max Asset
v0+ funded uncapped 6.01% 5.77% 1.04 -12.78% 53.52% 55.34%
25% cap 5.36% 9.61% 0.56 -24.98% 25.00% 25.00%
30% cap 5.47% 8.21% 0.67 -20.61% 30.00% 30.00%
35% cap 5.74% 7.19% 0.80 -18.76% 35.00% 35.00%

Latest 30% Cap Snapshot, 2026-05-15

Asset Weight Risk Contribution
511220.XSHG 城投债ETF 30.00% -0.37%
511020.XSHG 活跃国债ETF 19.25% 0.70%
510300.XSHG 沪深300ETF 17.34% 44.19%
511260.XSHG 10年国债ETF 15.76% 0.64%
510500.XSHG 中证500ETF 11.19% 44.03%
518880.XSHG 黄金ETF 2.09% 8.33%
159985.XSHE 豆粕ETF 2.02% -0.48%
510880.XSHG 红利ETF 1.47% 0.76%
159980.XSHE 有色/商品ETF 0.87% 2.20%
511990.XSHG 货币基金ETF 0.00% 0.00%
511360.XSHG 短债ETF 0.00% 0.00%

Latest 30% cap trailing 60d realized vol: 5.10% annualized.

Interpretation

The cap solves single-asset concentration but worsens portfolio risk because the excess from 511220.XSHG is mostly absorbed by R1 same-sleeve equity assets (510300, 510500). This raises equity risk contribution and pushes MDD above the original target zone.

Recommendation: do not lock 25% or 30% cap as final deployable baseline. If a quick cap must be used before structural fix, 35% is less damaging, but the better answer is still the structural fix: remove cash/short-bond from regime sleeves before Stage 2 and rerun ERC.