Making good decisions comes from assessing many different data points, including assessing how different trade strategies play out in the market. With the Omega Point Experiments Manager, you can quickly iterate various scenarios, save them as experiments, and compare each experiment to the original portfolio.

With experiments you can:

  • Evaluate how a particular rebalance — as generated from a simulation, optimization, or via our AI-created insights — impacts the performance, risk, exposure, and composition of your portfolio
  • Upload and organize alternative portfolios through the Experiments Managers
  • View the trades to go from one portfolio to the experimental portfolio for any date
  • Utilize forecasts to view an experiment's efficiency

Designing “What-If” experiments

Instantly evaluate what happens to a portfolio’s performance, risk, and exposures when certain trades are made using Rebalance: Optimization, or Rebalance: Simulation

Once you are satisfied with these rebalances, save the results and view these in Rebalance: Experiments to evaluate these changes side-by-side to the original portfolio to assess the experiment’s impact to a portfolio’s profile.

Experiments act as a comprehensive comparison tool to asses how different portfolios stack-up against each other. The following workflow makes it possible:

1 - Open a portfolio, and navigate to Rebalance > Experiments

2 - Click on Add Experiment

3 - Choose experiment type:

  • Optimize the portfolio using an efficient frontier
  • Simulate trades using a real-time rebalance tool
  • Upload a portfolio that already has the desired trades executed

4 - Make desired changes to your rebalance (trade simulation shown below). Then save & name the rebalance as an experimental portfolio.

5 - Experiments immediately populate in the experiment comparison screen

  • Experiments automatically populate forward from the date of the experiment to today. 
  • Evaluate performance, risk, exposure, composition, and forecasts between experiments
  • View the trades to go into the experimental portfolio
  • View an experiments efficiency by visualizing its placement along the efficient frontier

With the Experiments Manager, it is possible to quickly iterate on new ideas and hone onto the most impactful rebalance.

Utilizing Forecasts in Experiments with Frontier Placement

The efficient frontier in the context of experiments allows quick side-by-side comparison across forecasted metrics, given the selected forecast + horizon pair. 

When selecting Efficient Frontier from the experiments view tab (top-left dropdown), an efficient frontier is drawn using the selected forecast — as a default, this frontier uses the Implied Expected Returns forecast to create the same frontier found in On-Demand Optimizations

Once the frontier is drawn, the graph displays the placement of each experiment in relationship to the frontier line. Recall that modern portfolio theory
In addition, the forecast is used to generate Forecasted Performance, Risk, & Sharpe for all loaded experiments.


API: Developing “What-If” Experiments

As an API-first platform, the Experiments Manager is powered by the Omega Point Developers Platform. From a development point-of-view, experiments themselves are position sets that live within the context of a portfolio. The following endpoints allow customers to programmatically create, update, and manage experiments.

Create, Update, & Delete Experiments

Experiment objects are containers that connect the underlying experiment position set with a portfolio. Managing experiments and their metadata helps organize experiments.

mutation.createExperiment provisions the experiment object with a name, the type of experiment (optimization, simulation, or portfolio), and associates the experiment to an existing portfolio.
mutation.updateExperiment allows an existing experiment’s metadata to be updated, in particular, its name and various experiment specific options (like optimization constraints).
mutation.deleteExperiment will remove the experiment and all its associated data from the platform.

Upload Experiment Date Data

Much like a portfolio positionSet, experiment data is uploaded on a per-date basis in the form of a positionSet.

mutation.uploadExperimentDate  uploads holdings data into the experiment object. The platform and model calculate all performance, risk, and exposure metrics from these positions on a date-by-date basis, including automatically rolling forward holdings data

Reading Experiment Data

Experiments can be retrieved in two ways. The first manner includes retrieving experiment metadata:

query.portfolio(id).experiments returns a full list of available experiments on a per-portfolio basis.
query.portfolio(id).experiment(id) retrieves experiments metadata and position data

More importantly, a second method retrieves metrics as calculated from a risk model:

query.model(id).simulation(positionSet(portfolioId, experimentId)) configures the simulation endpoint such that all simulation sub-nodes — such as performance, risk contributors, grouped exposure contributors, composition, etc — will return modelled metrics as it pertains to the experiment.

As always, the Omega Point API returns experiment data in easy-to-use JSON format that mimics the request structure, and returns data within seconds.

Did this answer your question?