Optimizing with an Efficient Frontier
Running optimizations is as simple as selecting a date, and clicking anywhere along the efficient frontier. By clicking on a particular risk target (x-axis), a new optimization is generated and appears in the optimization scorecard to the right. The efficient frontier can be customized using a variety of optimization constraints, setting factor targets, as well as by uploading custom forecasts.
The Omega Point Optimizer was built with speed in mind and should perform all calculations within seconds. This way you can spend time iterating and refining the optimization to find the best configuration that delivers on your strategy.
Once a satisfactory optimization has been found, saving the optimization as an experiment takes the optimized portfolio into the Experiments Manager, where you can view a simplified performance backtest (i.e the positions are simply rolled-forward from the date of the optimization) and compare to any other saved optimization / experiment.
The Efficient Frontier
The efficient frontier provides a visual and interactive way to explore a range of optimized portfolios. Each point on the graph is generated using a mean-variance optimization for a given level of total risk (x-axis), and finding the highest aggregate portfolio return (y-axis) for this level of total risk (the risk target).
Although any point along the frontier is clickable, the graph highlights two points:
The dark blue dot corresponds to the current portfolio
This visualizes the current's portfolio proximity to the frontier and potential room for improvement. As a refresher, any portfolio that does not lie on the efficient frontier itself is inefficient and the optimizer works to find a portfolio on the frontier.
The light blue dot corresponds to the portfolio with the best risk-adjusted return
Hovering over the light blue dot provides a preview of the approximate forecasted return for this portfolio, and clicking on it performs the optimization needed to achieve this portfolio.
Anytime the efficient frontier is re-generated by changing constraints, forecasts, or factor targets, the optimizer will kick off a new optimization using a risk target that correspond to the same level of risk as the original portfolio (as visualized by the Risk Target line running through this dark blue dot).
The efficient frontier is generated by selecting a forecast, including selecting custom forecasts that have been uploaded into the platform. If no custom forecasts are available, the optimizer utilizes a default "Implied Expected Returns," described below.
Default: Implied Expected Returns
When no forecasts have been uploaded, the efficient frontier is calculated using the aggregate implied expected returns** for a portfolio, calculated using a position's predicted risk relative to its % equity. This method was pioneered by William Sharpe in his article from 1974 titled “Imputing Expected Security Returns From Portfolio Composition”. This expected return is analogous to the security's hurdle rate, where the expected return justifies the predicted risk. In the near future, Omega Point will allow you to enter your own price targets, in order to incorporate your own forecasts directly into this optimization.
Implied expected return are used to graph each optimized portfolio's total expected return, where the Optimizer automatically highlights the point which corresponds to the best risk-adjusted return.
**This feature employs an implied returns method where the resulting implied expected return is assumed to conform to a market model of predicted risk.
I.e When calculating implied returns, we use a model which only includes the market factor. So when using the optimizer, it re-optimizes against a model which includes other factors as well.
To update the efficient frontier to use custom forecasts, open the Forecast tab from the optimization configuration toolbar. Once this tab is open, the default forecast can be changed by clicking on the compass button and selecting a saved forecasts.
Only one forecast can be selected, please ensure the selected forecast includes all the securities in your portfolio. If a security is missing from the forecast, it will be treated as having 0 expected return.
The Omega Point Optimizer comes with many useful features that fully power the optimization experience. General constraints, factor targets, per-security trade restrictions and your target level of risk work seamlessly together, where adjustments to any combination of these are quickly visualized through an updated optimization scorecard. This real-time feedback allows for rapid iterations to help understand the impact of every change you require.
The general constraints available via the UI include constraints enabling adherence to common mandate & strategy guidelines. These include:
- Max Turnover (% equity) Restrict the percent equity that can be traded during the optimization.
- Max Trade Size (as a % of ADV) Constrains the trade size for a single security to not exceed this multiple of the average daily volume.
- Equity Change (% equity) Governs the amount that the optimizer is allowed to change the portfolio's total NAV
- Net Long The ratio between the portfolio's long book and short book, where 50% net long results in a market neutral portfolio.
- Trade Baskets If the portfolio includes any baskets or swaps, then the optimizer can either trade these are normal securities or leave these alone. By default, this is turned on.
Add minimum and maximum exposure targets for any individual factor in the risk model, targeting up to 4 factor targets at once. If these constraints are too tight or unachievable, the optimizer will send a notification to this extent.
General Constraints are automatically saved per portfolio in the browser’s memory after clicking the “Apply” button. Returning to the optimization screen will automatically load and generate a new frontier with these saved preferences. In instances where the saved constraints are too tight, a message will appear to help you debug which constraint might be causing no optimizations to return.
Analyzing Optimization Results
By clicking on the frontier, the optimizer runs a full analysis and displays the optimized portfolio's expected return, risk decomposition, and trades. The analysis is shown in the optimization scorecard, where customers can compare the optimized portfolio and the original portfolio side-by-side.
Optimization Next Steps
A trades panel displays the trades performed and includes a link to instantly download to CSV. Or save as experiment to view the results of the optimization, including automatically populating performance, risk, and exposure metrics forward from the date of the experiment to today.
Trades & Trade Restrictions
Each optimization consists of a series of trades to achieve the desired total risk target with the expected total return. These trades can be viewed by clicking on the Trades button inside the optimization scorecard.
In the trades panel, trades are organized by buy, short, sell, buy to cover (shorts that are reduced in value). The panel also provides the ability to download the trades into CSV format for easy exportation into an OMS.
Each security listed in the trades panel can be provided with trade restrictions, limiting the optimizer’s ability to trade in / out of the current position. By clicking on a security’s row, an expansion panel appears that includes three options:
- Unrestricted This lets the optimizer trade the position with no limitation. This option is selected by default for all securities.
- Fix to Base % Equity This pins the trade to the current weight in the original portfolio, meaning the optimizer cannot trade this security.* *Please note: since the trades panel only displays securities that have moved, any security that was fixed to its base % equity will no longer appear in the trades panel.
- Set Trade Range A target minimum and maximum trade range can be provided, which means the optimizer must find a weight for this position that honors this restriction
After setting trade restrictions, clicking on the apply button at the bottom of the panel will instantly apply these limitations and generate a new frontier, or provide a message that no optimizations can be found. In this case, it is recommended to apply trade restrictions one at a time, where real-time optimizations can provide guidance on the impact of any one trade restriction.
Some portfolio managers may want to run optimizations in relation to a benchmark. Using the “Compare” button in the top right corner allows optimizations to take a benchmark or portfolio into account as a base.
To run active optimizations, it is as simple as clicking on the “Compare” button. The screen will reload with active metrics compared to the selected benchmark. Clicking on the benchmark name presents a list of benchmarks and portfolios that can be used and once selected, the screen will reload with the new active metrics.
Errors & Messages
Running optimizations through the UI comes equipped with built-in error prevention and useful messaging. These feedback mechanisms ensure smooth operation of the optimizer, while pointing out adjustments that lead to a successful optimization These are provided in real-time as errors are encountered and allow for rapid optimization iterations.
Some types of errors that may be encountered include:
- Turnover is smaller than the number of mandatory trades need to return any feasible optimization
- Turnover is smaller than the minimum trade size
- Turnover is too small to satisfy net long constraints
- The factor exposure constraints are too tight
More rarely, internal errors may occur that are outside normal working conditions. These require investigation from Omega Point.