Portfolio Tools¶
Portfolios turn an Idea into a weighted set of holdings you can analyze, backtest, and refine. Most portfolios start from an Idea (a saved shortlist), then get weights either manually or via the Asset Allocator.
1. Create a Portfolio from an Idea¶
- Open My Ideas
- Select an Idea
- Click Create Portfolio from Idea
- Name the portfolio and confirm
A portfolio inherits the Idea’s holdings as the starting universe, but becomes its own object with its own weights, constraints, and analytics.
2. Holdings and weights¶
A portfolio is a list of assets plus weights (allocations). When a portfolio is first created, weights may be empty or unset.
You can set weights in two ways:
- Manual weights: type in your own allocations (subject to min/max constraints).
- Asset Allocator weights: let InvestLens propose weights based on your selected settings.
3. Asset Allocator and backtesting¶
The Asset Allocator proposes several standard allocations and backtests them over a chosen historical window so you can compare tradeoffs. You can set a period for backtesting in the settings under Portfolios tab.
Typical allocations include:
- Max Sharpe (MSR): targets the highest Sharpe ratio given an input risk-free rate.
- Equal Weight (EW): allocates evenly across holdings.
- Global Minimum Variance (GMV): targets lower volatility given the estimated covariance.
- Custom: respects your current weights and constraints (and may be used as a baseline).
3.1 Constraints¶
You can optionally set per-asset constraints such as:
- Minimum weight (
w_min) - Maximum weight (
w_max)
in portfolio creation or edit screens.
These constraints are used when generating allocator outputs.
3.2 Backtest output¶
For each allocation, InvestLens can show a historical growth-of-$1 series (or equivalent cumulative return series) over your chosen backtest window.
This helps answer questions like:
- “Does the max-Sharpe portfolio outperform, or does it just look good on paper?”
- “Is the low-variance portfolio actually more stable during drawdowns?”
- “How different is my custom portfolio from the suggested allocations?”
4. Performance Report¶
The Performance Report is a statistical summary of portfolio behavior over the selected time window.
4.1 What the Performance Report measures¶
Below are the most common fields you may see:
| Metric | What it means |
|---|---|
| Portfolio To-Date Return | Cumulative return over the period of Data Collection (see Portfolio details), based on the portfolio’s weighted daily returns. |
| Annualized Portfolio Return | The portfolio’s return scaled to an annual rate (assumes 252 trading days per year). |
| Annualized Portfolio Volatility | Standard deviation of daily portfolio returns scaled to an annual volatility. Higher means more variability. |
| Annualized Sharpe Ratio | Risk-adjusted return: (annualized return − risk-free rate) ÷ annualized volatility. Higher is better (all else equal). |
| Annualized Semi-deviation | Volatility of negative returns only (downside risk), scaled to annual. Useful when you care more about losses than swings. |
| VaR (5%) | Value at Risk at the selected confidence level (commonly 5%): a threshold such that roughly 5% of returns are worse than this number (historical method). |
| CVaR (5%) | Conditional VaR (Expected Shortfall): average loss given returns are beyond the VaR threshold. More conservative than VaR. |
| Cornish-Fisher VaR (5%) | A VaR estimate adjusted for non-normality using observed skewness and kurtosis (a normal approximation). |
| Skewness | Asymmetry of returns. Negative skew means larger downside tail risk; positive skew means larger upside tail. |
| Kurtosis | Tail heaviness vs normal distribution. Higher kurtosis usually implies more extreme outcomes. |
| Max Drawdown | Worst peak-to-trough decline in the period (the lowest drawdown value). Useful for “how bad did it get?” |
Where applicable, metrics are computed over the full return history available in the InvestLens database for the assets in the selected portfolio, as of the report’s as-of date. The effective date range is constrained by data availability (for example, if the database has 3 years of history, the report uses 3 years; if an asset has only 1 year of history, the usable window may be 1 year). The exact date range used for the portfolio can be verified in the portfolio Details view.
5. Style Analysis¶
Style Analysis explains what kind of risk and return drivers your portfolio behaves like, using a standard multi-factor model.
InvestLens supports a factor-based style view referencing the Fama-French 5 Factor model (2×3) data library: Fama/French 5 Factors (2x3)
5.1 The five factors¶
| Factor | Common label | Intuition |
|---|---|---|
| Market | Mkt − Risk Free | Broad equity market exposure. Positive exposure implies you tend to move with the market; negative implies you tend to move against it. |
| Size | SMB | Small-cap vs large-cap tilt. Positive exposure implies a small-cap tilt; negative implies a large-cap tilt. |
| Value | HML | Value vs growth tilt. Positive exposure implies a value tilt; negative implies a growth tilt. |
| Profitability | RMW | Robust vs weak profitability tilt. Positive exposure implies a tilt toward more profitable firms; negative implies weaker profitability exposure. |
| Investment | CMA | Conservative vs aggressive investment tilt. Positive exposure implies a tilt toward firms that invest more conservatively; negative implies more aggressive investment exposure. |
5.2 What you get from style analysis¶
Style analysis typically reports:
- Factor exposures (betas): how sensitive the portfolio is to each factor.
- Alpha (if shown): return not explained by the factors over the period (a model residual).
- Interpretation: summarizes Fama–French factor tilts such as size (SMB), value (HML), profitability (RMW), and investment/asset-growth behavior (CMA), along with the portfolio’s overall sensitivity to broad equity movements (market beta).
6. Notes and limitations¶
- Portfolios and reports are decision-support tools for research and workflow, not execution instructions.
- Allocations are sensitive to assumptions (return estimates, covariance estimates, constraints, and lookback window).
- Backtests are historical and do not guarantee future results.
- Factor/style analysis depends on data availability and may vary across time windows.