Factor investing - ingredients for long-term excess returns

Research into factor strategies has generated a large number of additional factors in recent years. The most widely used factors today are value, quality, momentum, low volatility and size. The definition of factors varies significantly between different economists and providers. Many providers continue to orientate themselves on the classic factor definitions of Fama-French or other economists who mainly wrote their studies in the 1990s.

However, the following chart shows that the two classic factors of size (company size) and value (price-to-book ratio) according to the Fama-French definition have no longer been delivering excess returns for some years.

Ingredients for a sustainable alpha

The past few years have shown that many factors are struggling in the current market environment. Of course, no active strategy is immune to negative years. For an active multi-factor model to outperform the overall market over a longer period of time (e.g. five years), sophisticated definitions and consistent research are required today, which critically scrutinises the selected factors on the one hand and looks out for potential new factors on the other. Adjustments must always be carefully weighed up. The following questions provide orientation:

• When does a change have a fundamental background and when is it just a temporary deviation, such as during the dotcom bubble or the financial crisis?
• Does the customisation offer economic and significant added value or is it just data mining?
• Where is consistency needed, when is adaptation required?

Our approach

We rely on our many years of expertise and our systematic research process. We have continuously developed our model over the past 20 years.

Over the past five years, around 30 sub-factors have been analysed in detail. Of these, twelve were ultimately integrated into the model. Our multi-factor model currently consists of the three main factors of value, quality and momentum, which together are based on 26 sub-factors.

In addition to the significantly more extensive composition, our multifactor model differs from classic factor definitions in the following points:

• Our model is sector-neutral, which leads to significantly greater stability.
• Wherever possible, we use current market data or analyst estimates (forecasts) when calculating the factors. In this way, we avoid working with irrelevant data.
• Our portfolios are reallocated monthly up to a predefined maximum (usually 10 per cent). We also take transaction costs into account. This keeps our portfolios dynamic, but the transaction costs within reasonable limits.
• We can quickly integrate new sub-factors that are significant based on our research into the model. This does not require any prior adjustments to legal documents, which can quickly lengthen the customisation process.
• We are more flexible with the data and can make assumptions if data points are missing (e.g. using the median value). This allows us to tap into additional factors, as we do not have to adhere to a rigid catalogue.