An allocation to ELM is an allocation to its underlying stocks, bonds and cash. This October 8, 2026 experiment estimates those exposures and asks how they affect a portfolio alongside bitcoin, gold and commodities.
The fixed 60% ELM / 30% GLDM / 10% FBTC example assigns only 10% of capital to bitcoin, yet bitcoin contributes about one-third of modeled portfolio variance at the saved ELM target mix. Fund weights alone give an incomplete picture of the risks being held.
The dated allocation used in the experiment
The ELM issuer describes a 75% global-equity and 25% fixed-income baseline, with allocations adjusted using valuation and momentum signals. Its stated 0.24% expense ratio includes underlying fund costs.
The study freezes the issuer’s October 6, 2026 target table, retrieved October 8. The live issuer page changes; the saved research inputs preserve the dated approximation used here.
| Underlying exposure | October 6 target approximation | Baseline approximation |
|---|---|---|
| U.S. equities, mapped to SPY | 26.76% | 45.50% |
| International equities, mapped to VXUS | 26.68% | 29.50% |
| TIPS, mapped to SCHP | 9.31% | 10.00% |
| Nominal bonds, mapped to IEF | 4.66% | 5.00% |
| Internal cash, mapped to SPAXX | 32.59% | 10.00% |
These are grouped target exposures. They are not a record of ELM’s actual holdings or an estimate of its future average mix.
The mapping has material compromises. U.S. value, small-cap and REIT sleeves share SPY’s assumptions. International regional weights share VXUS’s assumptions. U.S. aggregate bonds use an IEF Treasury proxy, leaving differences in credit, mortgages and duration unmodeled. The TIPS sleeve uses broad SCHP exposure.
How the model treats a fund of funds
The experiment transforms a saved Merton 3.0 model into four eligible risky funds: ELM, FBTC, GLDM and PDBC, with SPAXX as external cash. It aggregates the existing return forecasts and covariance using the fixed ELM mix. Covariance measures both each asset’s volatility and how assets move together.
Mapped proxy expenses are restored before deducting ELM’s total 0.24% expense once. Internal cash uses the model’s gross strategic short-rate anchor; external SPAXX retains its existing net return. Both have zero modeled return volatility.
The transformation carries over the same joint forecast-uncertainty draws. Bitcoin, gold, commodities and external cash retain their saved assumptions. It adds no assumed return advantage from ELM’s future allocation decisions and no extra manager-risk estimate.
Under these assumptions, ELM’s modeled compound return is about 5.05% with the October target mix and 4.95% with the baseline mix. Modeled annual volatility is 9.07% and 12.71%, respectively. These are conditional estimates for frozen exposures, rather than measured ELM performance.
What the optimization selected
The experiment retains the saved model’s risk-aversion and minimum-cash settings. Each fund can receive up to 100%; ELM has no required minimum weight. The central calculation uses the saved forecasts directly. The uncertainty-adjusted calculation also penalizes exposure to uncertain relative-return estimates, using the model’s existing uncertainty settings.
| Modeled capital allocation | Central, either ELM mix | Uncertainty-adjusted, October targets | Uncertainty-adjusted, baseline |
|---|---|---|---|
| ELM | 0.00% | 13.28% | 2.54% |
| FBTC | 28.58% | 19.08% | 19.41% |
| GLDM | 28.91% | 15.60% | 16.39% |
| PDBC | 22.41% | 22.97% | 23.40% |
| External SPAXX | 20.10% | 29.07% | 38.26% |
Rounded weights may not sum to exactly 100%. The full experiment results preserve the assumptions, calculations and independent solver checks.
The central calculation assigns ELM zero weight under both approximations. The uncertainty-adjusted calculation admits ELM, with a larger weight under the more cash-heavy October targets. Eligibility alone does not guarantee that an optimizer will select a fund as its portfolio core.
Internal cash also changes the interpretation. In the October uncertainty-adjusted case, ELM contributes another 4.33% of total portfolio capital in cash, on top of 29.07% external SPAXX. Total cash exposure is about 33.40%.
These allocations are diagnostic outputs from permissive constraints. In particular, the central calculation’s 28.58% bitcoin allocation reflects the saved 20% compound-return assumption. The bitcoin liquidity study did not validate that forecast or establish a replacement. The optimizer’s answer inherits that unresolved input.
A fixed 60/30/10 policy has different exposures
A separate calculation holds 60% ELM, 30% GLDM and 10% FBTC. It does not optimize those capital weights. Using the same fixed ELM approximations gives:
| Total portfolio exposure | October 6 ELM targets | ELM baseline |
|---|---|---|
| Global equities | 32.06% | 45.00% |
| Nominal bonds and TIPS | 8.38% | 9.00% |
| Internal cash | 19.55% | 6.00% |
| Gold | 30.00% | 30.00% |
| Bitcoin | 10.00% | 10.00% |
At the baseline, bonds plus cash total 15% of the whole portfolio. At October targets, almost one-fifth of total capital sits in ELM’s internal cash. That exposure can change as the fund changes its allocations, even if the investor keeps ELM at 60%.
The fixed-policy diagnostic estimates annual volatility at 11.85% for the October mix and 13.42% for the baseline. At October targets, contributions to modeled variance are 32.76% ELM, 32.78% gold and 34.46% bitcoin. At the baseline they become 44.46%, 26.67% and 28.87%.
Variance contributions include covariance with the other holdings. They are neither loss limits nor permanent equal-risk weights. The near-equal October split depends on the saved assumptions and allocations.
For a simple stress example, simultaneous declines of 25% in ELM, 20% in gold and 75% in bitcoin would produce a 28.5% portfolio loss before rebalancing. This is weighted scenario arithmetic, with no probability estimate or claim that it bounds the possible loss. An annual volatility estimate cannot establish a maximum drawdown.
What remains unresolved
The study contains no historical backtest of ELM’s changing strategy. Its risk estimate transforms the saved Merton covariance, including existing floors and shrinkage; it does not measure ELM’s own return history. Future allocation changes, manager risk and uncertainty in the proxy mappings remain outside the calculation. Investor taxes and withdrawals are also absent.
The next useful checks are actual underlying holdings, changes in internal cash, duration and credit exposure, and portfolio risk under alternative bitcoin return assumptions. Those checks would clarify whether a chosen capital allocation still expresses the intended risk as ELM’s holdings change.