The team here at Libra Investment Services have been analysing global economies and markets for the best part of 40 years and, over that time have seen how the interaction of macro and micro trends, stock level financials and thematic and sector dynamics influence and, ultimately, drive the complex adaptive system that is the financial market. Our key resource is an analytical framework that uses Machine Learning and neural network techniques to systematically analyse and forecast individual companies; their expected returns and associated probabilities in a way that allows us to aggregate and build our world view from the bottom up.
Standing on the shoulders of Apollo
We call this system Apollo and, building on this proprietary dataset that goes back to 2004, we are able to run and operate the strategic advisory product, known as Smart Alpha. By modelling stock level dynamics, we can construct a systematic understanding of what is happening at sector, market, factor and thematic level in real time. By combining this top-down insight with a factor-based risk and return analysis (see below) and the individual stock level forecasts we generate daily, a fully integrated picture of the financial market is developed.
It is by using this framework that we construct the range of Smart Alpha portfolios featured here https://libra-apollo.com/strategies. Because we see stock investment factors such as value or quality as not only stock specific but as evolving characteristics of the market environment itself, we categorise every stock in our universe into one of several dynamic factor baskets that reflect the combination of not only the company’s fundamentals but of how the market environment prices and risk assesses these fundamentals. This means that a value stock may evolve into a quality stock, or a growth stock may become a junk stock through time not only because of changes in the absolute fundamentals of the company (its earnings outlook or balance sheet strength for example) but on how the market as a whole evaluates the risks associated with them.
For strategic investment purposes, we limit our selections to the four “long only” categories of:
- Deep Value – stocks where growth outlooks are very uncertain but value discounts (the implied cost of capital), although historically high, are potentially dropping as stocks rerate.
- Value – stocks where discount rates remain high but are clearly declining (and hence “re-rating”) as growth prospects become clearer and suggest rising future payoffs for investors.
- Quality – stocks where valuations are more normally aligned with mostly visible (e.g. tangible) return outlooks that are positive and rising.
- Growth – stocks where risk premiums in the form of discount rates/implied cost-of-capital are very low compared to either their peers or their own history as (both tangible and intangible) growth prospects underpin confidence in positive returns.
The remaining categories of high-risk Growth (where risk discounts are too low and growth prospects are becoming uncertain) high-risk shorts (where discount rates are still rising but news shocks to negative growth prospects can surprise on the upside) and Junk (rising discount rates and negative and deteriorating growth outlooks) are preselected out of the strategies by design.
This is an applied but differentiated way of looking at Investment Factors. Whilst the traditional approach broadly captures the same concept of how fundamentals are being “risk priced” by the market at a point in time (discounting value with the price gives an implied risk premium), we are not relying on relative ranking or score-based categorisations determined by historical averages or ranked and sorted multiples (book-to -price for example). The implicit assumptions underpinning this – of a fixed, sortable universe (dividing stocks by an arbitrary B/P ratio of 1 for instance) and a return-to- the-historical-mean assumption of equilibrium related to a baseline market measure of the equity risk premium are rarely observed in the real world.
Instead, we are using the Apollo models to generate absolute, forward-looking views on the implied cost of capital and future expected payoffs from owing individual stocks and using these as a framework for categorisation. In other words, it is these, constantly-updating, dynamic factors that provide a means of identifying clusters of forecast returns and their probabilities from within the universe of possible returns – there are no externalities imposed upon the definition of a factor group as they “self describe” into the categories listed above.
A word on momentum
At this point it might help to explain our approach to momentum. Most traditional categorisations of momentum reflect a relatively crude determination of periodic price action – the “strength and direction” of a share price across a period of time with the assumption that strong current trends are likely to persist. Signal analytics based upon technical observations – the moving average convergence and divergence of different periods of price action (MACD) for example or a Relative Strength Index (RSI). Apollo defines Investment momentum not as the momentum of the share price per se but the momentum of the Expected Return. By taking this approach, momentum becomes a complementary factor to Value, Growth, Quality etc. We calculate this measure at the stock level daily and capture both the direction and amplitude of Expected Return momentum as part of the input data set for our Smart Alpha models
So, where shorter-term Investment Momentum trends are clearly negative, we simply filter out these stocks from the first four factor groupings, prior to stock selection. In addition, stocks where uncertainty is significantly higher than normal (we also have a proprietary indicator for this) are also filtered out.
Themes and Factors
One of the questions that we are often “asked to ask” of our data is whether a particular factor is working in our favour (as a long only investor) at a particular point in time. We can obviously form a view in regard to this from the bottom-up, stock level work that we do as we are able to aggregate measures of value, momentum, and uncertainty across any range of or collection of individual companies. However, the reality is it is often about identifying which factors are NOT working well and why. This may come down to there being a lack of available factor opportunities within a selected universe (no sensible value stocks appearing in a sector or market universe at a particular point in time for example) or just a lack of confidence in returns leading to very wide probability distributions for expected returns.
This helps to underpin the portfolio construction approach outlined above where we treat factors as part of a portfolio risk management process – not the portfolio-return selection process. Value often “performs well” when Growth is facing rising uncertainty for example even though its own fundamentals are not driving direct interest. Factor based rotations are often part of the quarterly rebalancing approach for funds and this is particularly the case where factor-based ETFs are used by asset managers as part of their portfolio management process. The clear outperformance of Value in Q1 2025 for example, reflects growth NOT working well as opposed to value suddenly being attractive in its own right. For Libra, it is a balance of all (positively supportive factors – our multifactor approach – that outperforms both subgroups (see Chart 1) and makes the case against trying to factor bet over time.
This is in stark contrast to theme investing. The overwhelming majority of themes are plays on future growth – particularly in relation to Technology and Innovation and future trend themes such as robotics. We can also view this as a play on ‘intangibles’ where potential growth from Space Technology, AI, Robotics, Medical tech etc. is associated with a shift in expectations as new information emerges that inspires positive expectations at a rapid pace. By extension, these themes tend not to be diversified from growth (or from each other) and, as such, make the case for themes being satellite components of portfolios – not core elements of them. They are about stock level alpha as much as about thematic timing and so need to be analysed accordingly.
We set out this backdrop as an aide-memoire for those occasional readers of our commentaries and postings on markets and strategies and as a source of reference for those interested in a little more detail as to how our proprietary data and models allow us to make the systematic calls and recommendations that are associated with the Apollo system.