QIS-Powered ETPs: A Better Product Stack Onchain

QIS-powered ETPs are one way to reconsider basket investing for institutional allocators: they combine the distribution advantages of an exchange-traded wrapper with the rigor of quantitative portfolio construction.

The underlying idea is straightforward: apply QIS methodology to improve diversification and risk-adjusted performance step by step, then deploy that logic through programmable onchain vault infrastructure. This note uses the 21Shares Bitcoin Gold ETP and the Vinter ByteTree BOLD1 methodology as a public reference benchmark (21Shares & Vinter methodology), then walks through how QIS-driven upgrades and onchain implementation each changed risk-adjusted performance in this sample.

Faithful Replication of the Public Benchmark

As a baseline, this note reproduces the reference inverse-volatility methodology using sample volatility estimation (Sharpe ratio: 1.22):

That baseline provides a clean reference point for attributing each subsequent change to a specific cause rather than to the basket as a whole.

Better Risk Estimation

The first upgrade keeps the same inverse-volatility construction but replaces naive sample estimates with skfolio-based routines and shrunk covariance estimation (Sharpe ratio: 1.36):

In this sample, shrinkage improved the robustness of risk estimates, particularly in noisier regimes, without changing the underlying weighting rule.

From Inverse Volatility to Minimum Volatility

The second upgrade addresses a structural limitation rather than an estimation issue: inverse-volatility weighting scales each sleeve by its own volatility and does not use cross-asset correlation information at all.

Moving to minimum-volatility optimization under the same risk estimation framework changed the result (Sharpe ratio: 1.59):

This is a shift in kind, not only in degree: from scaling each sleeve independently to optimizing the joint basket using the full covariance structure.

Onchain Yield-Bearing Implementation

The final upgrade in this note comes from implementation rather than methodology. By accessing BTC/XAU-denominated yield-bearing vaults onchain, the strategy kept its diversification profile while adding a further, onchain-native return component (Sharpe ratio: 1.74):

Why Onchain Vault Implementation Matters

These stepwise results are shown on nominal returns only; they do not yet include the additional operational efficiency (on the order of tens of basis points) that automating methodology execution, rebalancing, and controls through Orion's smart contracts could plausibly add on top.

That gap points to the real contrast between offchain basket products, such as traditional ETPs, and onchain vaults: the portfolio logic can be similar or identical, but the operating model underneath it is not.

In offchain structures, a meaningful share of cost and friction sits in manual or fragmented operational layers. In vault-based onchain structures, those same layers become programmable and continuously executable:

• benchmark maintenance;
• portfolio management operations;
• execution and rebalancing workflows;
• accounting and NAV tracking;
• reconciliation overhead.

As these functions are streamlined, a vault-based product can preserve the same basket exposure while operating with less manual overhead than its offchain counterpart.

Conclusion

In this reference construction, onchain vault implementation functioned as more than a distribution channel: it contributed to the measured performance difference alongside the methodology upgrades. For institutional allocators, the more durable implication is structural rather than incremental: combining quantitative methodology with programmable operations changes the iteration cycle for a product, tightens the risk-control loop, and can lower operational drag, though each of those depends on the specific methodology and infrastructure choices made, not on moving onchain by itself. Readers interested in how the operational layers referenced above, issuance, custody, administration, are replatformed onchain in more detail may find The ETP Issuer Stack, Onchain a useful companion piece.

References

• 21Shares & Vinter. 21Shares Vinter ByteTree BOLD Indexes Methodology.
• Nicolini, C., Manzi, M., & Delatte, H. (2025). skfolio: Portfolio Optimization in Python. arXiv.
• Ledoit, O., & Wolf, M. Honey, I Shrunk the Sample Covariance Matrix.