Sarwa’s ETF portfolios are constructed using a three-stage process:

High Quality ETF Selection

Estimation of each ETF’s historical performance

Portfolio Optimization

The aim of this process is to help investors build portfolios that provide them with maximum possible benefits of diversification. The approach is based off Nobel Prize winning research by Harry Markowitz -Modern Portfolio Theory- with adjustments that reflect our modern understanding of the strengths and weaknesses of this approach. The ultimate goal is an investment strategy that is primarily passive (except periodic rebalancing) and that keeps various fees (management and transaction fees) low.

Stage 1 (High Quality ETF Selection):

We start from a large database of ETFs (1000s) and only consider ETFs from most reputable ETF managers (e.g.: Vanguard, BlackRock). Afterwards, we look for funds in following asset classes:

- Bonds (US and international)

- Real Estate

- Equities (US, International-Developed, International-Emerging)

From those, we pick ETFs in each class that have the best mix of high liquidity (transaction volume), low fees and historical performance. This stage helps us establish the six to nine ETFs used to build our portfolios.

Stage 2 (Estimation of each ETF’s Historical Performance):

We estimate the risk of those investments by looking at their 5-year window of historical ETF monthly returns. If a particular ETF does not have a 5-year window, we look at the entire window available. However, we do not use the historical monthly returns to estimate future returns. This is because estimates are known to be inaccurate and such inaccuracies may have huge impacts on the portfolio. For example, if an asset’s future returns are underestimated by 1%, asset’s representation in the portfolio may fall by 50 or 100%.

To account for these problems, we assume expected returns are in line with a mix of two considerations:

Longer term historical averages (30 years)

Fair “risk-reward” going forward

Stage 3 (Portfolio Optimization):

During this stage, we undertake the standard mean-variance optimization, but with an extra criterion. We ensure that the picked portfolios do not worsen too much under different risk-reward situations. Simply put, we attempt to build “robust” portfolios that does well in a variety of plausible scenarios regarding the joint distribution of returns across asset classes.
Was this article helpful?
Cancel
Thank you!