Journal of Investment Strategies

This issue marks the start of the fourth year for The Journal of Investment Strategies and I would like to sincerely thank everyone who has made this milestone possible. I am grateful to our stellar editorial board for guiding the journal through its first three years, and to our exceptional publishing editors and the whole team at Risk Journals for producing a top-notch publication and getting us in front of many subscribers worldwide. My sincere thanks go to our authors and referees for contributing highquality papers and making sure we select the best and most interesting ones for the benefit of our readers.

This issue continues the tradition, with four papers devoted to portfolio construction topics, including statistical properties of bonds and stocks, the risk and diversification aspects of portfolio construction and the methodologies for aggregating alpha signals. In the first paper of the issue, "The stock-bond correlation" by Nic Johnson,Vasant Naik, Sebastien Page, Niels Pedersen and Steve Sapra, the research team from PIMCO presents the results of an investigation into an extremely topical issue, which is at the forefront of many investors' thoughts right now: the correlation between stocks and bonds. It dovetails nicely with a 2013 paper from this journal, "Credit portfolio management in a turning rates environment" by A. M. Berd, E. Ranguelova and A.Baldaque da Silva (Volume 3, Issue 1, pp. 123-142), which considered a similar topic of correlation between interest rates and credit spreads. While we all know that, in recent history, bonds and stocks have been negatively correlated, the degree of this correlation, and occasionally even its sign, can depend on market conditions. This was amply displayed during the May-June 2013 "taper tantrum" and many worry that it might still come to pass when the Fed actually starts raising rates, possibly as early as mid-2015. Without giving away the paper's punchline, I will just mention that to properly answer this question one has to view things in a wider perspective: one must not just investigate the historical patterns of correlations, one should really attempt to understand what drives them, and then estimate the impact of such driving variables. This is what Johnson et al. have done and this is why all portfolio managers should read this paper and try to draw their own lessons from its findings.

In the issue's second paper, "Intertemporal risk parity: a constant volatility framework for factor investing", Romain Perchet, Raul Leote de Carvalho and Pierre Moulin present the results of their investigation into the benefits of volatility targeting, ie, building portfolios that achieve constant volatility over time. They call this "intertemporal risk parity" to contrast it with cross-sectional risk parity, which attempts to equalize the risk contributions of different portfolio components during a single investment period. I amnot so sure about the utility of such an analogy, though, as volatility targeting is indeed a well-established practice and the newmoniker seems
more confusing than illuminating. The authors find that the benefits of volatility targeting depend strongly on the asset class and the targeted risk factor. In particular, they find that the greatest improvement in risk-adjusted returns is achieved for equity and foreign exchange, while it is muted for government bonds. They conjecture that this is related to the volatility clustering and fat tails properties of the corresponding returns.

In our third paper, "The number of stocks in your portfolio should be larger than you think: diversification evidence from five developed markets", Vitali Alexeev and Francis Tapon investigate a very practical and important question about the level of sufficient diversification that is recommended for investors' portfolios. While professional and institutional investors usually hold diverse portfolios, many individual investors fail to do so - to the detriment of their long-term investment results. By studying the question in greater detail and by focusing not only on the average behavior of the portfolio but also its behavior during times of market distress, the authors find that a substantial increase in the number of stocks is required to attain the diversification benefit at least 90% of the time. They confirm their findings by examining five different developed markets. I believe this paper and its results could be particularly useful for investment advisors who help their clients to build and maintain custom portfolios of stocks.

The issue's fourth paper, "Factor models for alpha streams" by Zura Kakushadze, introduces factor models for what the author calls "alpha streams"; in other words, the signals developed for various investment strategies targeting the underlying investable stocks. This problem is interesting, especially for systematic/algorithmic investment managers. While the factor models for the stock returns are well-known and widely used, dealing with potentially large numbers of alpha signals based on the same universe of stocks introduces its own challenges and requires a careful treatment for optimal portfolio construction and execution.With the advent of new algorithmic techniques for discovering newalpha signals - including methods from artificial intelligence,data mining and other methods bound to produce an overwhelming number of choices - it becomes even more important to get a handle on the robust statistical properties of these signals. Kakushadze's paper makes important progress toward this goal.

I would like to again thank our readers for their continued support and interest,and hope that with each new issue of The Journal of Investment Strategies they find something interesting and useful for their own work. We are confident that future issues will provide you with yet more compelling reasons to keep reading the journal.

Arthur M. Berd
General Quantitative LLC