Skip to content

research: design a different Hyperliquid vault strategy - #62

Open
miohtama wants to merge 2 commits into
masterfrom
research/hyperliquid-c00-next-leads
Open

miohtama wants to merge 2 commits into
masterfrom
research/hyperliquid-c00-next-leads

Conversation

@miohtama

@miohtama miohtama commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor

A different vault strategy

Replace this branch's incremental C00 sell-band × RS01 recommendation with a proposed factor- and regime-aware portfolio, adding underlying vault-book information where point-in-time historical coverage qualifies.

The goal is substantially better net profitability with acceptable downside. C00 is the standalone benchmark, rather than the new strategy's protected core. No challenger has been implemented or shown to beat it in this PR.

Decision Proposed change
Signal Separate market exposure, persistent residual skill and conditional downside; examine funding, leverage and book changes when supported
Selection Compare executable opportunities across eligible vaults, incumbents and cash; remove C00's six-entry/eight-exit ranking gate
Sizing Joint scenario allocation targeting net growth with an explicit tail-loss budget; remove mandatory inverse-variance weighting
Breadth and cash Follow opportunities, shared risk, costs and capacity rather than six slots or a deployment target
Sparse/young history Pooled priors and actual-observation confidence; unknown data is not zero profit or safe risk

What earlier work changes about this proposal

NB21 already tried expected-return/loss forecasts and adaptive, basis-aware allocation unsuccessfully. A more elaborate optimiser alone does not solve misleading forecasts. NB08 also examined BTC/ETH-adjusted outcomes; another beta chart is insufficient novelty.

The proposed next experiment isolates return-source decomposition, market conditioning, and new underlying information, with the same joint allocator across new-strategy arms:

Arm Signal and construction
A Exact C00 benchmark
B Shrunk residual skill and factor exposures; joint allocation across vaults/cash
C B plus market-conditional skill/exposure and downside scenarios
D, gated C plus historically supported funding/book/leverage information

The research note defines the return-source model, expected-log-growth objective, scenario CVaR constraint, chronological forecast controls, coverage gate, sparse-data handling, common accounting, 2026/full-history comparisons and economic reporting. Book-history truncation and reconstruction limitations are explicit. A shorter covered cohort requires matched controls, not comparison against a different full-history account.

These are proposed directions, not executed results. Current archive periods are repeatedly researched; any surviving policy still needs later confirmation.

Changes

  • Rewrite the research direction.
  • Update README/summary to match the new direction.
  • Retain the README's factual F01 correction to USD 116, from the saved USD 115.74 account difference.

PR #61 has no posted closure comments or reviews at the time checked; its concluding findings are the closure record used here. Changes are restricted to three Markdown files. No artefacts, binaries, datasets, notebook outputs or strategy code are added or changed.

@miohtama miohtama changed the title research: prioritise C00 follow-up leads after NB23 research: design a different Hyperliquid vault strategy Oct 14, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant