The DFS Portfolio Solutions Manager Selection Process

This document provides a step-by-step outline of the DFS Manager Selection Process. To assist with the explanation we have also included examples of our research at a number of stages of the research process. The examples provided are from various asset class reviews that have been conducted over the past 2 years.

The DFSPS Manager Selection methodology employs a rigorous process combining quantitative screening techniques with a qualitative assessment to establish a best of breed model portfolio. The quantitative approach employed by DFS efficiently filters out the vast majority of fund-managers to derive our “Reserve-List” of managers to which comprehensive qualitative overlays are applied. As stated in the review process, the objective of our quantitative-based model is to assess each manager’s ability to consistently convert the risk it has taken into returns in excess of multiple benchmarks and the qualitative inputs applied by DFS will generally include a broad assessment of each portfolio’s characteristics as well as an appraisal of the underlying asset composition.

STEP 1: CONSTRUCTING THE INVESTMENT UNIVERSE

Specialist managers (that solely invest in the one asset class) are selected to form our initial universe of funds under analysis. Specialist fund managers are preferred over generalists (those that invest across all asset classes) due to their greater stock selection insight and ability to provide superior risk adjusted returns.

DFS use a variety of sources to construct the initial investment universe of funds. These sources include (1) DFS Internal Database; (2) Lonsec; (3) APIR; (4) Morningstar.

For our Australian Equities Large Cap and International Equities Large Cap Reviews, each manager is grouped within the respective investment style (Value, Neutral, Growth, and Long-Short/Market Neutral/Absolute Return). Monthly returns over a 10-year period are then obtained from Lonsec (Financial Express) to run through our multi-factor Model. We do not conduct any further analysis in excess of 10 years as the vast majority of managers do not possess longer term track records. As such, a high degree of diligence has been incorporated to ensure that we adequately compare the consistency of each manager’s risk adjusted performance generation against other managers over a number of disparate periods.

STEP 2: QUANTITATIVE ANALYSIS STAGE 1 – RANKING DFS PERFORMANCE INDICATORS BASED ON MONTHLY RETURN SERIES

The objective of our investment process is to accurately rank the quality, consistency and efficiency of returns generated by each fund manager. This is executed through our multi-factor model that assesses each manager’s performance relative to its peers and the market index. The Model also runs a raft of analyses to ascertain whether managers are truly active rather than simply hugging the Index. The rankings generated from the “Quantitative Analysis Stage 1” are then used to construct a shortlist of managers. To be eligible for the DFS shortlist, managers have to either:

  • achieve a top quartile ranking as measured by the DFS Performance Indicators across all time periods;
  • have consistently improved their ranking across various time periods whilst ranking in the top quartile in recent times;
  • display superior risk-adjusted returns over rolling 12 months and rolling 3 years for a prolonged period as displayed by the DFS Performance Indicator heat charts.

Through our quantitative assessment, we generally observe that approximately 80% of funds analysed are effectively screened out.

STEP 3: QUANTITATIVE ANALYSIS STAGE 2 – RANKING DFS PERFORMANCE INDICATORS BASED ON DAILY RETURN SERIES

The use of daily data substantially improves the statistical efficacy of our analysis. DFS is particularly interested in the behaviour of tail events, which is certainly enriched through the application of daily data to accurately identify the source of risk and alpha generation amongst the short-listed managers. To maintain the integrity of the data used, DFS reconciles the daily data series and monthly data series sourced from each shortlisted manager. We also reconcile the monthly return series sourced from Lonsec (Financial Express) against the series produced by each Manager. This gives us the utmost confidence in the results of any analysis conducted.

In addition to assessing the daily returns of each short-listed manager against its peers and market benchmarks, we further refine our analysis to scrutinise the performance of each manager under different market conditions, namely:

  • Bull Market Conditions – 15% p.a. + returns
  • Bear Market Conditions – less than 5% p.a. returns
  • Normal Market Conditions – returns between 5-15%p.a.

This macroeconomic overlay is an important part of our manager selection process, as we actively tilt to managers that are expected to generate superior performance under market conditions that are anticipated by DFS over the next several years.

We use the “Excess Return Distribution” and “Observed Excess Return” bar chart to provide insights into each manager’s performance attributes under disparate market conditions. Daily return data provides an ample number of statistical observations in order to confidently assess each manager’s performance across all market conditions.

STEP 4: QUANTITATIVE ANALYSIS STAGE 3 – REGRESSION ANALYSIS

An integral step in our quantitative process is to regress each shortlisted managers’ excess returns against; (1) Investment Style Indices; (2) Cap-based (or market segmented) Indices; (3) the FUM for the strategy/fund and; (4) net cash flows for strategy/fund.

Style regression uses two internally constructed indices. The first index represents the difference between value and growth returns, and the second index represents the difference between small cap and large cap returns. The analysis identifies a fund’s sensitivity to growth & value factors as well as the portfolio’s dynamic exposures to small and large cap factors.

FUM and cash flow regression is also conducted on each manager’s excess return to assess the relationship between assets managed against portfolio performance. The FUM and cash flow regression is a multi-faceted approach that distinguishes between market timing and stock selection in determining a manager’s ability to generate alpha.

The rankings of the shortlisted managers as per the multi-factor model are assessed in conjunction with the regression analysis to construct a reserve-list to which further qualitative research is conducted

STEP 5: QUALITATIVE ANALYSIS STAGE 1 – ATTRIBUTION & RISK ANALYSIS

While our quantitative assessment identifies the quality, consistency and efficiency of returns generated by each fund manager, our qualitative assessment identifies the underlying source and drivers behind such returns. DFS assesses the sector composition and regional allocation of the portfolio over a protracted period of time to identify the extent to which the portfolio’s performance is attributed to a variety of sources.

We further assess the manager’s performance against their own return and risk objectives to identify whether there is a relationship between a manager’s ability to meet internal objectives and the consistency we observe through our multi-factor model.

Finally we request a copy of the manager’s portfolio risk report to develop a detailed understanding of how each manager thinks about the portfolio’s risk exposure.

STEP 6: QUALITATIVE ANALYSIS STAGE 2 – DUE DILIGENCE QUESTIONNAIRE & MANAGER MEETING

The selected managers who have been successful in making the Reserve-List have proven their ability to generate consistent and meaningful performance. The second stage of our qualitative process is designed to identify the managers who are likely to maintain superior performance exhibited during the quantitative process. The approach to the second qualitative stage is two-fold: firstly we ask each investment manager to submit the completed FSC Questionnaire on the strategy. DFS also has an internal document that acts as an overlay to the FSC-based questionnaire that is completed by each manager. The questionnaires allow us to assess the qualitative aspects of each reserve-listed manager. This covers investment related information (investment philosophy & process, research efforts, portfolio construction techniques, risk management process, investment team) as well as the non-investment related details (organization and staff, business profitability, compliance, risk management, custody, etc.).

The primary objective of the manager meetings is to allow DFS to better qualify (1) those managers who have a sustainable and competitive investment insight; (2) the source of such sustainable insights and; (3) the barriers to entry (the extent to which their competitive edge can be replicated by peers).

The manager meeting consists of two areas, namely portfolio specific and team-centric considerations. Firstly, DFS clarifies any outstanding queries that have arisen throughout the previous stages of the review process – be that at our attribution analysis, regression or performance analysis during specific periods. We also assess how (or whether) the manager views the current and projected climate and how these themes may be incorporated in the portfolio construction process. We also further consider and discuss each managers’ portfolio metrics and qualify how they relate to the managers’ investment philosophy and/or macro views. The remainder of the meeting is then focused on the firm and investment team. We confirm the information obtained from the two questionnaires via a step-by-step breakdown of the research and portfolio construction process. DFS is particularly interested in assessing the belief system, mindset; and interaction of team members. Consequently, questions are focused around core beliefs; team dynamics; staff remuneration and turnover; internal relationships; break-up of day-to-day responsibilities at varying levels within the investment team are discussed in detail. DFS is further interested in the business strategy of the investment firm and its current business development pipeline.

STEP 7: QUALITATIVE ANALYSIS STAGE 3 – CORRELATION ANALYSIS AND MODEL COMPOSITION

The final stage in the DFS investment process is to assess the relationship of each of the reserve-listed managers. This assessment is based on the return correlation of each fund and the stability of their correlations under time varying volatility. This analysis is considered alongside the results of each preceding stage (for example, considerations are given towards style biases of the portfolio) to construct optimised model portfolios. This is ultimately achieved by anchoring our forward looking assessment on market conditions to our risk management framework to determine the appropriate weighting of the appointed manager within the portfolio.

STEP 8: ONGOING MONITORING OF RESERVE-LISTED MANAGERS

DFS conducts semi-annual “monitoring” reviews on each of the reserve-listed managers. This assessment is conducted on the daily returns of the reserve-listed managers and is performed to qualify the results of the formal review process. Specifically, DFS monitors the reserve-list to ascertain whether the perceived performance sustainability of each manager is maintained against their peers. This is a proactive measure that allows DFS to efficiently replace any appointed manager, if required. DFS holds all the reserve-listed managers in high regard, whether or not appointed. The preference to appoint one reserve-listed manager over another will often depend on our forward looking (active) views, rather any notion of absolute superiority.

The monitoring process begins by assessing the daily returns of the reserve-listed managers as per the Quantitative Analysis Stage 2 from above. In addition to running the DFSPI on the reserve-list, the current DFS portfolio will also be included in the dataset so we can assess the risk metrics of the aggregate portfolio.

We also monitor if each manager is meeting the internal return and risk objectives as per above. This allows us to determine whether there is a relationship between a manager’s ability to meet internal objectives and the consistency we observe through our multi-factor model.

The next step in the monitoring process we construct a Bayesian framework to monitor manager performance. This is a simple but effective way to update our expectations of a manager’s performance, and gives us a framework to decide whether we remain confident in their ability to outperform.

We also have regular meetings/teleconferences with appointed managers to discuss current portfolio positioning and performance attribution is also conducted. We will also meet with any of the reserve-listed managers where appropriate.