Portfolio optimization software for real-world constraints
Portfolio optimization software turns a list of candidate assets into weights, under the limits you set. Acubic does that in three explicit steps (screen, rank, allocate), lets you choose among three optimisers and several rule-based weightings, and tests the result walk-forward before anything is traded. It should do more than maximize a formula: the goal is an allocation that is explainable before it is tradable.
What should portfolio optimization software actually optimize?
The real job is not simply to maximize expected return. Strong portfolio optimization software balances return targets against concentration limits, turnover budgets, drawdown tolerance, liquidity boundaries, and practical rebalancing rules. When those constraints are ignored, the output may look elegant but fail operationally.
Key terms
- Optimiser: an algorithm that sets weights by solving for a goal, such as the best return per unit of risk, under constraints.
- Covariance matrix: how each pair of assets has moved together. Most optimisers estimate it from past returns, so it carries estimation error.
- Hierarchical risk parity (HRP): groups assets by how they move together and spreads risk across those groups, without inverting the covariance matrix.
- CVaR (conditional value at risk): the average loss in the worst share of outcomes. Optimising against it targets the tail rather than ordinary volatility.
- Rule-based weighting: weights set by a fixed rule, such as equal weights or company size with a cap, rather than by an optimiser.
Which portfolio construction approaches does Acubic support?
Acubic supports multiple quantitative portfolio construction styles: mean-variance optimisation, hierarchical risk parity, and defensive CVaR-oriented workflows, plus rule-based weightings. That matters because different investors need different tradeoffs between return-seeking, diversification, and downside control. Nothing is preselected; Beginner and Intermediate users see a recommended option for their risk level, and the choice stays theirs.
| Name in the app | How weights are set | Watch out for |
|---|---|---|
| Balanced Allocation | Hierarchical risk parity: risk spread across clusters of assets that move together | Uses no return forecast, so it does not tilt toward expected winners |
| Aggressive Allocation | Mean-variance optimisation: the best trade-off between expected return and volatility | Concentrates hard when return estimates are noisy |
| Defensive Allocation | CVaR-focused: limits the average loss in the worst outcomes, traded off against return | Can give up upside to limit the tail |
| Rule-based weightings | Equal weights, by rank, half rank and half equal, or by company size (alone or mixed with rank) with a 10% or 20% cap per asset | Simple and transparent, but blind to how assets move together |
How does the optimization workflow run?
- Screen: keep assets with enough price history and, for strategies created or regenerated since October 5, 2026, enough trading volume (a median of at least $1 million a day over the last 60 trading days).
- Rank: order the eligible assets by the method you choose, such as market capitalization, Sharpe ratio, twelve-month momentum or a composite score.
- Allocate: set weights with the optimiser or rule you picked, within your risk level's per-asset cap for an optimiser, or the rule's own cap, unless you set your own maximum.
- Validate: run a walk-forward backtest that repeats the same three steps at every past rebalance date.
- Rebalance: monthly, quarterly or annually, update the targets from your strategy and, on a connected broker, trade only the difference.
How does Acubic reduce optimisation fragility?
Optimisation inputs are noisy. Acubic addresses that by combining explicit constraints, risk profiling inputs, and walk-forward backtesting instead of relying on a single in-sample result. For the three optimisers, unless you set your own maximum, each asset is capped by default at 10%, 15% or 25% for conservative, moderate and aggressive risk levels, widened for small portfolios (with eight holdings: 18.75%, 25% or 37.5%). These caps apply to strategies created or regenerated since October 5, 2026. Rule weightings use their own 10% or 20% cap, or none. Impossible sets of bounds are relaxed to the nearest feasible value rather than failing. The goal is to judge robustness, not only headline return metrics.
Why does validation matter as much as allocation design?
A portfolio that looks attractive in one period can break down when costs, slippage, or regime shifts appear. Acubic pairs allocation design with walk-forward validation so users can compare how a portfolio behaves under more realistic assumptions before moving closer to execution.
Who is this portfolio optimization workflow built for?
Acubic is built for self-directed investors, financial analysts, and research-driven users who need more structure than spreadsheets but still want to understand the process behind each portfolio decision.
What happens after a portfolio is generated?
A usable workflow does not stop at weights. Users still need rebalance logic, implementation review, methodology context, and broker-connected execution paths. The U.S. Securities and Exchange Commission's investor guide defines the step plainly: “Rebalancing is bringing your portfolio back to your original asset allocation mix.” Acubic is positioned around that full lifecycle rather than a single optimization button.
What are the limits of portfolio optimization?
Every optimiser works from estimates of the past. Correlations shift, and they tend to rise together in the drawdowns diversification is meant to soften. Rebalancing has costs too: the same SEC guide advises considering whether a rebalancing method “will trigger transaction fees or tax consequences.” Acubic's backtests charge a flat transaction cost but do not model taxes or each broker's own fees. Results are hypothetical, past performance does not guarantee future results, and nothing here is financial advice.