As long as markets players operate within a liberal and capitalistic framework where risk is to be remunerated, a long allocation to Beta strategies is a sustainable source of long-term returns. The most evident challenge, when incorporating Betas into a portfolio, is to balance individual asset classes (for instance equities and bonds) to allow for a static portfolio to become stable and reliable.
To solve the allocation problem, the core assumption is that Betas react in understandable ways based on the relationship between their expected premia relative to the economic environment. This investment program offers automated systematic execution to a multiple set of strategies based on well comprehended fundamental logics.
As a first step, the allocation algorithms analyses macroeconomic biases: each asset class performs well in certain environments and poorly in others (for instance, bonds perform well during times of deflationary recessions, while stocks perform best during periods of growth). Macroeconomic biases need to be understood and in a solid portfolio they must balance-off so that different risks will offset each other.
Principal Component Analysis (PCA), the analysis of the components driving asset price changes, can be used to map asset classes in different environments; the most critical aspect is to frame global growth and global inflation as the two main key drivers of asset performance (as together they explain over 75% of the asset classes price variance).
Stocks and bonds can offset each other during growth shocks, but there are other environments that can hurt both stocks and bonds, such as rising inflation, where FX and interest rates are expected to provide a hedge. The scope is to put equal risk on each possible macro environmental scenario to achieve balance.
Portfolio rebalancing is not based on correlations, as they are unstable, nor on timing, instead it operates around the specific themes of each class. It is based on understanding the ways that discounted economic conditions are reflected in asset pricing, and most importantly, that the asset allocation is balanced with respect to the rising and falling growth and inflation rates, considering the phase of the strategy cycle.
The algorithms exploit Beta and Alpha sources, displaying positive expectations, in order to achieve higher degree of diversification than traditional portfolios with a more attractive risk-return profile. The asset allocation approach is based on the idea of combining several sources of Beta in order to minimize cross-correlation among components and limit subjective overlays at operational level. Allocations to single Beta components are normalized in terms of risk factors and volatilities.
The allocation process separates beta sources from alpha sources and then sizes them, according to the antagonism between the strategies, to obtain maximum portfolio diversification:
I. Beta sources are the key drivers of medium to long term performance as individual Beta components have positive expected returns and are meant to outperform cash over time. Most of the Beta sources exhibit high cross-correlations and short-term volatility, but tend to be very reliable over long time horizons.
II. Alpha sources are driven by market inefficiencies and typically exhibit lower cross-correlations and more interesting risk-reward profiles. However Alpha sources require a high degree of active management and significant analytical, technological and computational skills. These are:
a. short-term trading algorithms, which are completely trend-agnostic. Models exploit price, time and volatility relationships allowing for fast risk-rebalancing. Characterized by a very robust risk/return profile, its (synthetic long-vega) positioning allows this component to particularly benefit in case of abrupt and large market moves, providing both a significant contribution and a meaningful risk protection.
b. value algorithms capture relative value returns across foreign exchange, fixed income and equity indices. The value models are often uncorrelated or even anti-correlated to each other. Value portfolios are constructed as equal-volatility-weighted allocations to individual contracts or asset classes.
The employment of state-of-the-art proprietary technology is a key tool in the advanced statistical analysis of massive data allowing to isolate and highlight patterns underpinning the cause-effect of events and also to implement the disciplined automation of empirically tested business decisions, without resorting to dangerous predictions and forecasting exercises.
Drawdowns cannot be eliminated, however if their causes are well comprehended, they can be controlled. An investment program based on non-convergent, but on antagonist/complementary strategies shows an improved control of risk, with the aim of reducing the uncompensated risk associated with individual securities and/or markets, and trying to smoothen the negative side, enhancing the consistency of the portfolio returns.
Contributo a cura di
Enrico Laddaga, Business Developer e BGB Weston, Sub-Investment Manager del fondo Zest Pilot