quantmodel — Port + verify Kipnis Defensive Adaptive Asset Allocation vs AS metrics

Objective

Port + verify Kipnis Defensive Adaptive Asset Allocation vs AS metrics.

Notes: KDA replicated 2015+: corr 0.9965 CAGRΔ +0.40%. Spec/backtest/kda.py + runner. Fixed canary-only AGG sparsify/ffill bug. Findings: kda-scale-batch.md.

Approach

Reuses existing report research/findings/kda-scale-batch.md — tables/metrics preserved verbatim via source_report pointer. Body below summarizes grounded evidence from that report and adjacent artifacts.

Files / code / data changed

Associated commits (inferred from log grep t18):

ce7ca907 research: t18 Kipnis Defensive AAA replicated vs AS (#18)
ce7ca907 research: t18 Kipnis Defensive AAA replicated vs AS (#18)
Source report: research/findings/kda-scale-batch.md
Task-history entry: research/findings/task-history.json#t18

Results

Key table from source report (preserved):

metric	ours	AS	notes
monthly corr	0.9965	—	gate ≥0.85
overlap CAGR	5.03%	4.63%	Δ +0.40%
overlap bias (mo)	+0.00031	—	near zero
full-window CAGR	5.26%	10.9% (1989+)	AS headline is full history
Sharpe (ours / AS)	0.68 / 0.95	—	different windows

Conclusions and metrics are in research/findings/kda-scale-batch.md; see detail page for preserved tables/metrics.

Conclusions

KDA replicated 2015+: corr 0.9965 CAGRΔ +0.40%. Spec/backtest/kda.py + runner. Fixed canary-only AGG sparsify/ffill bug. Findings: kda-scale-batch.md.

Problems / follow-ups

See source report or PR discussion for follow-ups. No unsupported follow-ups fabricated.