RAM Active’s European Equity Market Neutral

Capitalising on valuation normalisation

Hamlin Lovell
Originally published on 03 May 2022

RAM Active Investments’ RAM Lux Systematic Funds – Long/Short European Equities has received The Hedge Fund Journal’s UCITS Hedge award for best risk-adjusted returns in the Equity Market Neutral – Europe (Quantitative) category, in 2021, when the strategy made 18.9%, and over two, five, seven and ten years ending in December 2021. The strategy has integrated ESG and all RAM equity products make disclosures under SFDR category 8, bar one under category 9.

The resurgence of performance in 2021 came after the strategy’s longest and deepest drawdown, between 2018-2020, when equity market neutral strategies saw net outflows of capital that helped to amplify the extreme bifurcation in the market. “Low interest rates crowded investors into longer duration stocks, which culminated in USD 6 trillion of stocks valued at more than 20 times sales in early 2021. Meanwhile, short interest dropped to historic lows around 2% after the meme stock frenzy and squeezes on GameStop and AMC. Valuation dispersion on some measures peaked at levels even above the late 1990s as value stocks were derated. All of this was a double headwind for equity market neutral managers. Long and short books lost money with shorts detracting most,” explains Emmanuel Hauptmann, Partner and Senior Equity Fund Manager, who co-founded the Geneva-based firm in 2007.

The next few years could provide a strong multi-year opportunity set for market neutral like the 2009-2015 period.

Emmanuel Hauptmann, Partner and Senior Equity Fund Manager, RAM Active Investments

Now valuations have started to normalize and re-converge, and RAM’s strategy, which is not country nor sector neutral, has been trading these spreads within and between countries and sectors. Parts of the biotech, food delivery, renewable energy, hydrogen, electric vehicles and software sectors have posted some of the steepest drops and made the biggest contributions to RAM’s short book profits. Valuation dispersion as of March 2021, on measures such as earnings yield or operating cash flow, remains well above historical norms: it exceeds the average of the past 10 years, and is still not far below the level seen around the TMT bubble in 2000.

RAM’s sector positioning is dynamically shifting with relative valuations. Since late 2020, auto-makers have performed well on the long side, as have more traditional industrial firms with strong free cash flow yields. In early 2022, the strategy has developed a net long bias in technology as valuations of some names have come down to GARP (growth at a reasonable price) levels. Meanwhile, the strategy has also been net long of energy and materials since January 2022 and has thus exploited the inflation theme.

Factor weights, which were steady in the early years of the program, are now moving around more, rotating from a strong value and momentum bias in 2021, to a more defensive bent in 2022. “Healthcare has been increased and cyclical exposure in consumer discretionary, industrials, has been reduced. The quality bias has also become stronger,” says Hauptmann. This sort of rebalancing takes place on average monthly, with trades staggered every few days in the European strategy, which invests in a universe of 1,500 relatively liquid stocks and is the firm’s largest equity market neutral strategy, with AUM of EUR 216 million, while its smallest is global equity market neutral, running EUR 40 million in 3,500 stocks. Total firm assets are EUR 2.4 billion including long only equity, bonds and systematic macro.

Since the end of 2020, the strategy has been firing on all cylinders, with all long book engines – value, momentum, defensive and machine learning – performing. Shorts have done well since March 2021 after the short capitulation, with all four short engines – value, momentum, quality and machine learning – contributing positively. A shorter-term statistical arbitrage, mean reversion book introduced in late 2020 to diversify risk has also generated returns. It is rebalanced on a daily basis, and tends to profit from wide return dispersion, such as in the Covid recovery phase and during dislocated markets. 


The European strategy invests in a universe of 1,500 relatively liquid stocks and is the firm’s largest equity market neutral strategy

Valuation mean reversion has been an especially important driver of returns for RAM, but in late 2021 momentum became a more important explanation for performance and then in early 2022 defensive has come to the fore. The same stocks can make a journey through a shifting factor landscape: as value stocks recover, they can migrate into the momentum bucket. But value exposure remains clearly evident, with longs trading on lower valuations, higher dividend and free cash flow yields, while shorts in Europe have on average zero free cash flow yield, and those globally have a negative FCF yield of 5% as of March 2021. This could in itself be a source of valuation compression. If stock buybacks helped to drive valuation expansion, potential for issuance could work in the opposite direction. “Firms with negative cashflow will need refinancing, which will be harder to get with rising interest rates and credit spreads,” points out Hauptmann. 

Beyond traditional fundamental analysis

Value and other traditional factors should not be exaggerated however, because the models now use over 500 signals based on a wide variety of inputs, including fundamentals, sentiment, positioning and ESG data. New datasets have been added to monitor inputs including stock lending costs, insider trades, and some 50 of the 500 signals are ESG related. The RAM process involves multiple adjustments to clean data inputs. For instance, one financial statement variable adjusts earnings and cashflow valuations for R&D spending, which can be capitalized or expensed by different companies. Research into crowded shorts has found that those with more expensive borrow and higher short interest also tend to generate more alpha.

These dimensions are blended in non-linear combinations, which could include traditional and more alternative factors and signals: “We look at how value inputs interact with other inefficiencies. We seek value stocks with other attractive qualities to back up valuations. Firms with high dividend yields might even be good shorts if the payout is not backed by operating cashflows. Style bias alone is not enough,” explains Hauptmann.

New analytical techniques are also being used to fine tune portfolio rebalancing. “Machine learning permeates all of the books and is needed to validate them. They are being used to make return predictions, which then scale every trade in the book, rather than equally weighting them. Positions that were previously liquidated are now more likely to just be downsized, reducing portfolio turnover. This is a more robust way to scale positions across strategies,” says Hauptmann. The machine learning and deep learning infrastructure also employs Bayesian methods and processes. “These techniques better industrialize and condition the process of training the models, pick the optimal learning rate and scale to choose the architecture of the models. They have streamlined the return process and strongly enhanced robustness and accuracy out of sample,” he adds. 

We look at how value inputs interact with other inefficiencies. We seek value stocks with other attractive qualities to back up valuations.

Emmanuel Hauptmann, Partner and Senior Equity Fund Manager, RAM Active Investments

All five of the team have backgrounds in data science. The latest hire in this area was Tian Guo in 2019, who also has expertise in NLP (natural language programming) and neural networks. One of RAM’s research papers, Unlocking the Secrets in Semantics, outlines how the firm uses NLP to analyse a larger volume and wider variety of higher frequency data. “We are capturing data from new datasets and news-flow, including unstructured data that is more often text than images, earnings announcements, and management Q&A. We have state of the art language models that turn the news into a structured format and plug it into the machine learning pipeline. ESG research reports react more rapidly to negative ESG news, capturing some tail risk,” says Hauptmann.

Expanding ESG research

RAM’s research using NLP found that firms with adverse ESG news-flow also proved to be riskier. A key priority is to exclude ESG laggards, though it is also possible to short firms with high ESG scores if their fundamentals are negative, such as some renewable energy names that became very glamorous and richly valued.

RAM are making a special effort to control for unintended biases in ESG data, which can arise from ESG scores overlapping with traditional factors such as quality, growth, low risk, and large cap. For instance, ESG data is biased towards large caps that make more extensive disclosures, which handicaps smaller companies. “Our research in emerging markets has nonetheless found it is possible to maintain value and size biases and integrate ESG properly. We can find value and small cap names. Our emerging markets book has a PE ratio of 9 versus 12 for MSCI Emerging Markets Index, and higher ESG ratings,” says Hauptmann. 

Given the overlaps with other factors, ESG performance attribution is not straightforward: “However, RAM has used statistical Shapley techniques to isolate the ESG impact and found that ESG was in fact less useful than sentiment, earnings, valuation, quality and carry,” says Hauptmann.

ESG could make a larger contribution in future. It feeds into the deep learning optimization process, which is constantly evolving as is data quality. “All ESG metrics are being enriched, with data from ESG ratings agencies, transparency, diversity and governance indicators, as well as new ESG risk inputs that would have flagged up the Wirecard risk,” points out Hauptmann.

Carbon data is also getting better thanks to more companies and third parties disclosing and estimating it. The data could come direct from companies, from CDP and from agencies such as MSCI ESG. Scope 3 carbon emission data is currently being used to reduce tail risk by eliminating some stocks with high scope 3 emissions, but not yet for positive selection. In future wider and more consistent disclosure might pave the way for it to be used as a stock-picking signal. 

RAM’s ESG engagement could drive forward this virtuous circle of more disclosures allowing for more sophisticated ESG signals. “Our ESG engagement is mainly collaborative, via groups such as CDP, IIGCC and Climate Action 100+, pushing for better data disclosure on ESG, carbon and water. In response to press alerts, we may engage directly with firms on controversies as well, says Cyrille Joye, Head of Client Services, who leads the engagement drive. RAM has been UNPRI signatories since 2014 and are A-rated on strategy and governance.

Russia has recently become an ESG issue. RAM largely eliminated Russian equity exposure at the start of 2022, but a few locally listed names that brokers declined to trade remained, adding up to 0.36% at the time Moscow share trading was suspended. The exposure was marked down to zero as of March 2022.


The team see plenty of potential for further valuation reconvergence, and there are multiple analytical projects underway to refine and enhance myriad other signals and develop new ones. “The next few years could provide a strong multi-year opportunity set for market neutral like the 2009-2015 period,” says Hauptmann.