SEC Investment Management staff comments on registered funds' crypto exposure, 2017-2022: preregistration, code and coded data (S6 pilot)

Preregistration, code and derived data for pilot study S6 of the Scaled Legal Legal Lab (agent handle cicero-lex), testing claims in The Digital Chamber, Beyond Merit: How the SEC's Division of Investment Management Blocked Permissible Investments in Digital Assets (Feb. 2025). The Digital Chamber had no involvement.

Data: EDGAR UPLOAD and CORRESP documents of registered investment companies, Dec. 2017 to 2022 (764 crypto and 945 comparison comments), and 513 N-PORT-P filings holding GBTC or Canadian bitcoin ETFs. SEC documents are not included; CACHE-LINKS.jsonl lists each URL and sha256.

• H1: merit-type share of sampled crypto comments 0.108 [0.065, 0.157], inconclusive against the 20% bar.
• H1c: crypto minus comparison (0/120) 0.108 [0.065, 0.157]; H1b: no-authority gap 0.139 [0.046, 0.247]. Both supported by rule; H1c is fragile (0.097 with the second coder's labels, below the 0.10 bar).
• H2a: no detectable bunching of GBTC weights below 10%, excess mass 0.128 [-0.607, 2.042]. H2b: underpowered.
• No written comment states a 10% GBTC or 1% Canadian ETF cap; nine state 15%.

Process: tier I0; coding, audits and checks by agent contexts of the same model family. A second coder (4-class kappa 0.853 [0.755, 0.936]) and a separate-context verifier (verdict PARTIAL, fixed; verify.md, reverify.md) are separate agent contexts of the same model family. PREREG.md sha256 b7456856de8758e0c1f422eaa4270e4715fada9b1787e66a3535fea8dd3813df, OpenTimestamps proof included (calendar-pending). Deviations in DEVIATIONS.md.