Process

Step 1: Investable Universe

The Quantitative Investments team starts with an investable universe of approximately the top 200 JSE-listed shares, having filtered all listed counters for minimum levels of liquidity and market capitalisation.

Step 2: Forecast Alpha process – A systematic multi-factor return forecast

• The proprietary investment process begins with an analysis of factors driving share prices. It is based on how each share is influenced by factors (behavioural proxies) like risk, liquidity, valuation, profitability, momentum and analyst estimates,
• A share’s degree of “sensitivity” to these factors – called factor payoff – is key in share selection.
• The team then uses robust quantitative models to forecast expected returns for each share.
  • A multi-factor regression analysis is applied to groups of factors. This mathematical process determines to what extent these factors are important in explaining share price movements over time.
  • As each share’s return for the previous month is known, the boutique is able to simultaneously calculate their estimated monthly weights (sensitivity) to the various groups of factors.
  • In turn, they use this monthly history to calculate a rolling average weight (sensitivity) to each factor, and together with the factors, generate an expected twelve-month return.
• Shares are then ranked based on the highest expected alpha, with the top 40% representing our high-conviction alpha shares.

Step 3: Portfolio Construction process - Disciplined and objective

• Portfolio inputs – Inputs used by the team at the start of the portfolio construction process include the mandate constraints imposed by the client’s risk and return requirements, client risk aversion (guiding the portfolio’s tracking error parameters), and the returns per share forecast by the model, adjusted for the estimated implementation costs per share as well.
• Optimisation - The ranked shares are then optimized according to their risk and return characteristics, to obtain a portfolio with the highest expected return for a given risk parameter.
  • The constraints used in the optimization may include: tracking error, active sector exposures; active share exposures; turnover; liquidity adjustment; and other constraints.
• The portfolio is stress-tested to identify and eliminate inadvertent macroeconomic risks, quantifying the potential gains or losses due to exposure to macroeconomic factors such as currencies, commodities, interest rates, etc…
• Final Portfolio - The result is an optimized portfolio consistent with our quantitative philosophy and process. This final portfolio is tilted towards the factors currently dominating the market and therefore best positioned to generate alpha.
  • At all times we strive to have the portfolio overweight the 40% most attractive shares and underweight the 60% least attractive shares, resulting in a portfolio that represents our best investment view.
• As part of the boutique’s monthly investment process, this optimized portfolio is compared to the current portfolio, while also managing portfolio efficiency using measures including:
  • Transfer coefficient – A measure of how effectively the portfolio has captured the model’s forecast of alpha per share, and
  • Active share - An equivalent of tracking error, measuring how closely the fund moves versus its benchmark. Changes to the portfolio are implemented where the greatest alpha-generating and risk-reducing opportunities are identified.
  • To conclude the monthly investment process, a detailed attribution analysis is conducted, determining which factors contributed to and/ or detracted from performance. This enables us to understand and evaluate the sources of our returns.

Step 4: Cost-effective implementation

Quantitative Investments takes advantage of the large scale and experience of the Old Mutual Investment Group’s trading desk – the largest in South Africa - to ensure the lowest possible trading costs when buying and selling shares for its portfolios. The boutique always aims to minimize the costs of implementing our best views.

Unique to our process is the explicit inclusion of estimated transactions costs (including market impact costs) directly into our alpha forecast process. Forecast alphas are adjusted for estimated transaction costs using OMIGSA’s proprietary data. This ensures that an optimal trade-off between forecast returns and expected implementation costs is achieved.