The advance of technology and big data as tools in portfolio management has been dramatic in recent times, bringing with it concepts that blur the distinction between active and passive investing; smart beta being the most obvious example.

For active portfolio managers, there are quant tools and algorithms available that might best be described as smart alpha. They can give the portfolio manager a clearer picture of market behaviour, backward and forward looking, and used alongside other data can help a PM create a more robust approach to security selection.

The StarMine Combined Alpha Model (CAM) from Thomson Reuters is a leading-edge example of how quant models can help generate alpha by harnessing all available data on specific securities. The models assess stocks, individual markets, style biases, regional variations and other factors to identify the stocks that are likely to over/underperform relative to a global or regional universe.

Factors make a potent mix

Factor-based investing is catching on with institutional asset owners as it becomes clear that much outperformance can be explained by factors such as value, momentum and earnings quality. Now it has been shown that different factor-based strategies used in combination can help generate significant extra alpha. By embedding these models into their investment processes, portfolio managers can gain a real edge over their competitors.

These types of alpha factors are an important component in many of our customers’ investment approaches, and for those with a fundamental research overlay to their research, this is a frequent and popular screening tool that helps ensure their analysis is conducted on a subset of their investable universe with attractive quantitative characteristics.

They are equally valuable for long-only institutional asset management, for the equity (and credit) hedge fund and for the pure quant. These approaches certainly show strong performance, in part reflecting the lower amount of money (relative to US/Europe) tied to quantitative factors in the region. The factors are a powerful and innovative content set that show our continued thought leadership in the investment management space.

Taken individually, the factor strategies have proved effective in generating outperformance. Thomson Reuters has already established its StarMine Value-Momentum model (Val-Mo) as adding significant value over a basic value-momentum combination. But it is the combination of several factors within a model that has proved most potent in generating consistent alpha.

Although factor investing is a broadly accepted concept, what is less well understood is that the different factors deliver quite different performance across market cycles and geographies. And since it is hard to predict which factor will outperform, the StarMine CAM utilises a diversified approach optimising the different performance trends of factors globally.

The StarMine CAM has been tested through various market cycles going back to 1998 and shows significantly better performance than even the best individual or dual-factor strategies.

The models used in StarMine CAM are Analyst Revisions (ARM), Relative Valuation (RV), Intrinsic Valuation (IV), Price Momentum, Earnings Quality (EQ), Smart Holdings, Insider Filings (US only), and Short Interest (US only).

Every StarMine factor model is based on sound economic reasoning supported by a wealth of academic research, as well as rigorous research and backtesting carried out by Thomson Reuters. It is the correct combination of factors, with subtle variations by geographic region, and recognising the many forces driving performance, that provide a key edge.

The use of these enhanced and combined factors can outperform both passive and many ‘smart beta’ strategies.

The weights employed in StarMine CAM reflect general patterns of factor efficacy in different regions. The quality of data allows Thomson Reuters to develop predictive scenarios, but the model does not attempt to determine which factors will outperform. It uses static weightings that account for the different performance of factors globally.

Tried and tested

To examine the performance of CAM, we split the full 1998-2015 market testing period splits into three distinct phases: early (May 1999-2007), crisis (January 2008-December 2009), and recent (January 2010-May 2016) (click on graph, left). This enables us to highlight general patterns in factor performance by region, as well as consistency or variability over time.

From our analysis of the different cycles it is apparent that during a crisis period, momentum (ARM and Price Mo) fare poorly in every region. In contrast, value (RV) fares relatively well. Also notable is the consistency in factor performance in Developed Asia ex-Japan, Developed Europe, and Emerging Markets between the early and late periods. All three regions tend to favour momentum signals, and the weights in StarMine CAM reflect this (click on table, left).

Correlation dynamics

In addition to performing well on its own, in order for a factor to have significant value in a multi-factor model it should be uncorrelated to other factors in the model. Value and momentum are negatively correlated to one another, while there is positive correlation of Short Interest and Insider Filings with value, as short sellers and company insiders tend to be value investors.

We continue to see robust performance in all the StarMine alpha models and expect this performance to persist for many years to come.

How you can use StarMine CAM

The StarMine Combined Alpha Model is available in Thomson Reuters Eikon™. A daily data feed will soon be available through DataScope Select. Historical testing files are also available for those who wish to backtest their strategies. 

Contact:
For more information on the StarMine Combined Alpha Model or other Thomson Reuters solutions please e-mail us at asia.salesenquiries@thomsonreuters.com