This backtesting note compares two HKD 300,000 MPF allocation hypotheses using the same expected-return, risk, and correlation inputs that feed the MPF Modern Portfolio Theory chart.
Hypothesis
Compare two portfolios:
| Portfolio | Allocation | HKD amount |
|---|---|---|
| Americas-only portfolio | 100% Americas Equity Fund | 300,000 |
| Efficient-frontier optimiser portfolio | Long-only maximum-Sharpe optimiser using the MPF efficient-frontier inputs | 300,000 |
The optimiser is allowed to allocate across the MPF funds on the MPF page, but it cannot short funds or borrow. Under those assumptions, the maximum-Sharpe efficient-frontier solution selects the Americas Equity Fund at 100%, because it has the strongest expected excess return per unit of annualised risk in the current dataset.
Data Inputs
| Fund | ISIN | Expected annual return | Annualised risk | Market correlation | Role in comparison |
|---|---|---|---|---|---|
| Americas Equity Fund | HK0000936085 | 20.27% | 24.08% | 0.92 | 100% of Americas-only portfolio and 100% of unconstrained optimiser portfolio |
The optimiser also reviewed the full MPF fund universe used by the MPF page. The next most competitive growth funds, such as European Equity Fund and Global Equity Fund, have lower expected return-to-risk trade-offs under the same inputs, so they do not enter the unconstrained maximum-Sharpe solution.
Portfolio Estimates
| Metric | Americas-only portfolio | Efficient-frontier optimiser portfolio |
|---|---|---|
| Starting capital | HKD 300,000 | HKD 300,000 |
| Allocation | 100% Americas Equity Fund | 100% Americas Equity Fund |
| Expected annual return | 20.27% | 20.27% |
| Expected annual HKD return | HKD 60,810 | HKD 60,810 |
| Annualised standard deviation of risk | 24.08% | 24.08% |
| One-standard-deviation annual HKD risk | HKD 72,240 | HKD 72,240 |
| Risk-free rate | 4.50% | 4.50% |
| Estimated Sharpe ratio | 0.655 | 0.655 |
Efficient-Frontier Read-Through
On the current MPF Modern Portfolio Theory inputs, the efficient-frontier optimiser does not improve on the all-Americas portfolio. It converges to the same portfolio because the Americas Equity Fund has the highest expected return and the highest Sharpe ratio among the tracked MPF funds.
That means the comparison is not a diversified portfolio versus a concentrated portfolio. It is a useful diagnostic: without an explicit diversification constraint, the optimiser simply confirms that the Americas Equity Fund is the strongest point on the current MPF opportunity set.
Practical Diversification Sensitivity
If the investor wants a genuinely diversified portfolio, the model needs a concentration rule. A simple sensitivity is a 40% maximum allocation per fund. Under that additional rule, a practical capped optimiser allocates approximately:
| Fund | Weight | HKD amount |
|---|---|---|
| Americas Equity Fund | 40.00% | 120,000 |
| MPF Conservative Fund | 40.00% | 120,000 |
| European Equity Fund | 16.14% | 48,429 |
| RMB Bond Fund | 3.86% | 11,571 |
This capped version has an estimated expected annual return of 11.04%, annualised risk of 13.13%, and Sharpe ratio of 0.498. It sacrifices expected return, but it also reduces the one-standard-deviation annual HKD risk estimate from about HKD 72,240 to about HKD 39,399.
Recommendation
For a pure model-driven choice with no concentration limit, the recommendation is the efficient-frontier optimiser portfolio, but only because it is identical to investing all HKD 300,000 in the Americas Equity Fund. The all-Americas portfolio is therefore the simpler expression of the same result.
For a real MPF retirement account, the better choice depends on risk tolerance:
- Choose the Americas-only / unconstrained optimiser result if the priority is maximum expected return and the investor accepts high regional equity concentration and roughly 24% annualised volatility.
- Choose a capped diversified optimiser if the investor wants a more balanced retirement allocation and is willing to give up expected return to reduce volatility and concentration risk.
My preference for retirement planning is the capped diversified optimiser, not the unconstrained all-Americas result, because MPF capital is long-term retirement capital and concentration risk matters even when a single fund currently dominates the efficient-frontier chart.
Methodology
Expected return is the weighted average of each fund’s expected annual return from data/fund_metrics.json. Portfolio risk uses each fund’s annualised standard deviation and the same market-correlation inputs used by the MPF Modern Portfolio Theory chart. The optimiser is a long-only maximum-Sharpe calculation using a 4.5% risk-free rate.
This is a modelled allocation note, not investment advice. It excludes fees, taxes, contribution timing, MPF switching rules, fund platform constraints, and future changes in factsheet data.