Skip to content

research: add xchain3 vault allocation experiments - #63

Open
miohtama wants to merge 1 commit into
masterfrom
research/xchain3-vault-allocation
Open

miohtama wants to merge 1 commit into
masterfrom
research/xchain3-vault-allocation

Conversation

@miohtama

Copy link
Copy Markdown
Contributor

Why

Record the xchain3 research behind master-vault-v3, including the unsuccessful funding configuration and the subsequent universe experiments. This is the companion research PR to tradingstrategy-ai/trade-executor#1679, which adds the executor strategy module, Atoma settlement support and the CCTP report-export fix.

Lessons learnt

  • Refreshing the universe exposes an operational funding constraint: the inherited $1,000 bridge buffer fails when primary-chain funding needs $1,900.04. Raising the buffer to $3,000 allows the completed baseline run.
  • Excluding private/restricted sources and YieldNest RWA MAX and adding the two Atoma vaults produces 328 sources. Atoma's 0.5% withdrawal fee must be included once, while redemption cash waits for the estimated seven-day settlement.
  • A retrospective lifetime Sharpe screen is not a point-in-time strategy. Only two added native vaults receive allocations, both late in the period; the modest terminal gain comes with weaker Sharpe and greater drawdown. Native funding is idealised, and the remaining Stratwise external fee is estimated separately.
  • Sparse NAV observations and synthetic windows materially affect the risk metrics. Preserve the limitations and unsuccessful run alongside the headline returns.

Summary

Add four notebooks with saved outputs for 1 April 2025 through 20 October 2026, starting with $100,000:

Notebook Experiment Recorded outcome
NB01 Refreshed 356-source universe with the inherited $1,000 bridge buffer Saved NotEnoughMoney failure; no complete performance result
NB02 Increase the bridge buffer to $3,000 $117,332.90 final equity; 10.84% CAGR; 11.33 Sharpe; -0.12% maximum drawdown
NB03 Exclude private sources and YieldNest RWA MAX; include Atoma Index and Atoma RWA $117,580.65 final equity; 10.99% CAGR; 10.39 Sharpe; -0.18% maximum drawdown
NB04 Add Lighter/Hyperliquid sources with snapshot lifetime Sharpe above 2.5 $117,859.20 ledger equity; 11.16% CAGR; 9.12 Sharpe; -0.49% maximum drawdown

Add a README linking the experiment sequence, executor PR, execution assumptions and local replay inputs. NB04 equity becomes $117,850.47 after its estimated $8.73 external-fee reserve.

Notebook schemas and code-cell syntax were checked. Formatting preserves the executable ASTs, outputs and metadata. The saved outputs show NB02 and NB03 completing all 21 code cells and NB04 completing all 22 without errors; NB01's sole saved error is the documented funding failure. The notebooks were not rerun for this PR.

Generated snapshots, ledgers, caches and reports remain ignored and are not included. NB03 and NB04 need the original local snapshots and baseline ledgers to replay their pinned comparisons; the README documents those paths. Configured credential values were checked against the notebook files before publication.

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