AI Portfolio Intelligence
lumeportdex analyses risk-adjusted return data across thousands of instruments and configures a diversified portfolio in under 60 seconds, with automated rebalancing applied on an ongoing basis.
The Decision Engine
Every recommendation is generated from historical volatility, correlation data, and forward-looking risk metrics, not from a single market forecast. The engine cross-checks its own output against three independent models before presenting an allocation.
Workflow
The process is deliberately compressed: three data inputs, one calculation cycle, one confirmation step. Nothing is hidden — each stage below produces an output you can review before it is applied.
You specify the objective (school fees, a house deposit, retirement) and the number of years available. This sets the boundary conditions for the model.
A gradient-boosted forecasting model and a mean-variance optimiser run in parallel, then reconcile differences before proposing a single allocation.
Threshold-based rebalancing triggers are set at account opening, so drift beyond your risk band is corrected without requiring manual sign-off each time.
Model transparency note: the predictive layer combines a supervised forecasting model trained on long-run asset-class returns with a classical mean-variance optimiser. Outputs are constrained by pre-set risk bands, and the household retains the ability to lower the risk tolerance at any point.
Risk Management
The methodology treats risk as a constraint to be managed continuously, rather than a figure disclosed once at account opening.
Capacity for loss is estimated from time horizon and stated liquidity needs, not from a single questionnaire score.
Each proposed portfolio is tested against historical downturn periods to estimate plausible worst-case movement.
Holdings are screened for overlapping exposure so a single sector shock cannot disproportionately affect the outcome.
Allocations are reviewed daily; rebalancing is triggered only when drift exceeds the agreed threshold, limiting unnecessary trading.
Calendar-based rebalancing (e.g. quarterly) can leave a portfolio exposed for weeks after a market move. Threshold-based rules respond to actual drift in allocation weight, which data suggests reduces the average time a portfolio spends outside its intended risk band.
It does not attempt to time short-term market movements or predict individual security prices. The scope is limited to long-run asset allocation and risk containment, which is where the available data supports a defensible model.
Applied Scenarios
Below are three illustrative configurations. Figures are model outputs based on stated assumptions, not guaranteed outcomes.
With a long runway before funds are needed, the model typically weights higher toward growth assets early on, then reduces equity exposure gradually as the horizon shortens, following a pre-agreed glide path.
Shorter horizons carry less tolerance for drawdown, so the model constrains growth-asset exposure and prioritises predictable, lower-volatility instruments to protect the deposit timeline.
With retirement decades away, the model can absorb higher short-term volatility in exchange for greater long-run expected return, subject to the household's stated risk tolerance.
Behind the Interface
lumeportdex was built on the assumption that time-poor parents still want to understand what they are agreeing to. Every allocation decision, rebalancing trigger, and fee is disclosed in plain terms before it is applied, so the speed of setup does not come at the cost of clarity.
Transparency
These answers are written for readers who want the mechanism, not the marketing summary.
Account and transaction data is encrypted at rest and in transit. Access to raw data is restricted to systems required for portfolio calculation and regulatory reporting; it is not used to train models for other households without prior aggregation and anonymisation.
Model outputs are constrained by pre-set risk parameters and reviewed periodically by our investment operations team. The model does not have authority to exceed the agreed risk band for any account without a separate confirmation step.
An annual management fee is charged as a percentage of assets under management, alongside underlying fund costs where applicable. Both are itemised before setup is confirmed, and there are no undisclosed transaction charges for standard rebalancing activity.
Yes. Adjusting the stated risk tolerance triggers a recalculation of the target allocation, and rebalancing toward the new target follows the same threshold-based rules as ongoing management.
The portfolio is not moved to cash automatically. Rebalancing rules continue to apply, which in practice means buying relatively more of the assets that have fallen, consistent with the pre-agreed allocation targets rather than a reaction to short-term sentiment.
Provide your goal and time horizon, review the proposed allocation, and confirm. The calculation itself takes under 60 seconds; the decision to proceed remains entirely yours.
Initialize Portfolio No obligation to fund an account before reviewing the full allocation and fee breakdown.