Having studied, backtested, and worked in traditional markets for the better part of a decade, my conclusion is that an investor who focuses on the long term can generate alpha by adhering to time tested strategies if decisions are made using a disciplined, rules-based approach. Rules-based doesn't necessarily mean strictly quantitative, but subjective decisions need to be analyzed by looking at the data, comparing subjective decisions to actual outcomes. This helps us learn, leading to refinements in the investment decision-making process. It's through this constant review that we can improve over time. Typically these data-driven reviews help us refine either our investment criteria checklist or our quantitative approach to investing. Natural biases creep into the decision-making process, and for the most part, as a hedge fund consultant, I've seen that these decisions are typically more harmful than helpful to portfolio returns. Data-driven reviews help us iron out these biases, resulting in improvements over time.
Tracking Data in Organizations
The first step in utilizing data to help inform the portfolio process is tracking the data in the first place. This seems quite obvious, but it's something that's not really being done rigidly in the hedge fund industry. Analysts, for the most part, aren't developers, and developers typically don't understand the value-add for capturing certain data points within an organization. Most fundamentally based hedge funds typically don't have cross-functional resources, someone who understands investing with a background in technology and portfolio analytics. This is typically where I step in as an outsourced data scientist to help organizations define their data strategy. The data strategy is what yields exceptional rewards over time, helping managers analyze past results that help refine their portfolio construction process, typically leading to enhanced alpha generation. For a $1 billion hedge fund, every 1% additional alpha with a 20% performance fee can add $2 million a year in revenue for the management company, making it easy to see why an upfront investment in data can be quite lucrative. But it goes beyond this, these small alpha differences can also decrease redemption risk, resulting in lowered business risk.
Defining a data strategy
A typical data strategy for a hedge fund may look something like this:
- Define the goals of the data strategy. Typically this is as follows:
- Automate manual processes
- Enable data availability across the organization
- Build reporting capabilities
- Research the aggregated dataset to refine the portfolio construction process. This typically means analyzing idea generation, sizing, risk, forecasting, and general portfolio management decision making.
- Backtest additional strategies to add additional investment themes to the portfolio.
- Automate data-driven decision making.
- Create a documentation repository, typically something like Atlassian's Confluence, and document each step of the process. This ensures that transition risk remains low, especially important from a consulting perspective.
- Reach out to all internal teams and data and software vendors to catalog all sources of disparate data.
- Architect a data warehouse solution that accounts for all disparate data.
- Implement the data warehouse schema from the previous step.
- Architect and implement ETL processes for all disparate data.
- Create rigorous QA routines around the dataset to ensure data integrity.
- Source and import comprehensive market datasets.
- Work with analysts to define and implement methods for importing financial models and forecast data.
- Work with managers to prioritize the goals.
- Implement best practices software using an agile approach to achieve the goals from point 1.