— Back to Blog

Maximizing the Number of Independent Bets in Your Portfolio

Maximizing the Number of Independent Bets in Your Portfolio

This summary is written by Rich based on a conversation in our CTA series between Guillaume Jamet, Principal Manager and Co-CIO at Metori Capital Management, and the podcast hosts, Niels and Alan.

About Metori Capital Management

Niels and Alan were joined by Guillaume Jamet, Principal Manager and Co-CIO at Metori Capital Management. 

They discussed the powerful investment strategy of trend following and managed futures strategies. 

Based in Paris, Metori Capital was established in 2016 by a group of former employees from Société Générale and Lyxor Asset Management. This group previously managed the Lyxor Epsilon Global Trend Fund before starting Metori as an independent spin-off. 

Today, the company manages over $800 million in assets and remains dedicated solely to trend following strategies. 

Guillaume has been associated with the Epsilon program since 2011, and the company has a subsidiary in China. 

Alan posed a question to Guillaume about the origins of his focus on trend following strategies. Guillaume explained that his initial background was in pure mathematics, and he moved into finance in the early 2000s, primarily in the derivative business.

During this period, he gained valuable insights, including experiences from the Global Financial Crisis. Around 2010, he moved into fund management at Lyxor, and in 2011, he got involved with the Epsilon program as the head of research.

His co-founder, Nicolas, was the CIO of Lyxor at the time. They both shared a similar engineering and mathematical background, which helped them develop a unique perspective on markets. They weren't primarily interested in expected return but rather in understanding how risk changes, focusing on elements like correlations and forward volatilities.

Using Financial Mathematics to Understand Trends

With their scientific backgrounds, trend following emerged as a sensible strategy because they recognized the necessity of expected return against risk in the markets. 

This led to a natural interest in the scientific problem of stochastic filtering, which involves separating expected returns and trends from risk. They viewed this challenge from a unique angle, distinct from many market participants, and aimed to understand and address this filtering problem.

In financial mathematics, stochastic filtering is a complex process that involves estimating the value of an underlying variable or set of variables over time, given a series of observations. The observations usually contain noise or uncertainty, so the problem is essentially one of signal extraction - distinguishing the true signal (in this case, expected returns and trends) from the noise (risk).

In the context of the conversation, Guillaume Jamet and his team at Metori Capital Management are applying this concept to their investment strategy. They use stochastic filtering to separate out expected returns and trends (the factors they want to capitalize on) from market risk (the 'noise' they want to manage or mitigate).

Most market participants don't view investments through this lens. Traditional investors might focus more on qualitative aspects, like a company's management team or the overall economy's health, rather than employing complex mathematical models to drive their decisions.

By viewing the problem through the lens of stochastic filtering, Guillaume and his team believe they can better understand the market dynamics at play, allowing them to make more informed investment decisions based on the trends and expected returns they identify, while also being aware of the risks they need to manage. This unique approach is part of their edge as a quantitative fund management firm.

Alan asked Guillaume whether the mathematical complexity they apply to trend following is necessary for risk management and signal generation. Guillaume stated that while the intuition behind trend following is simple (i.e., one can easily observe when prices consistently move in one direction for a prolonged period), the challenge lies in determining when this constitutes a trend and how strong it is. Therefore, one doesn't necessarily require an engineering or physics background to be a trend follower.

Alan further asked if Metori focuses purely on trend following or if they incorporate other strategies. Guillaume clarified that they are committed to trend following without adding other types of strategies like carry. However, he disagreed with the notion that trend following is simple, highlighting the complexity of doing it right. He emphasized the importance of considering correlation as a key component of effective trend following.

The Relevance of Sharpe 

Niels raised a question about the relevance of the Sharpe ratio as a measure of a strategy's quality in the CTA world. He questioned whose Sharpe ratio matters more – the manager's or the client's portfolio overall when these strategies are included.

Guillaume responded by expressing that the Sharpe ratio is a valuable metric for evaluating a strategy's quality, especially when only the track record of a fund strategy is available. He noted that the Sharpe ratio is a good measure of return adjusted for risk. The crucial aspect, he stressed, is whether the risks are accurately reflected in the strategy's volatility.

Guillaume mentioned that for a directional strategy like trend following, using linear instruments, it's difficult to argue against looking at risk metrics. He emphasized that understanding where the risk comes from (correlation with the equity market or other structural correlation with other markets) is essential and leads to the question of beta and alpha. If a strategy doesn't have a structural correlation with the main market risks and maintains a good Sharpe ratio, it indicates that the strategy is well-executed.

Niels posed a follow-up question to Guillaume, noting that the Sharpe ratio penalizes beneficial volatility and doesn't consider the risk of significant drawdowns or a strategy's potential to fail, as evidenced by several high Sharpe ratio strategies in the past. Niels asked how Guillaume adjusts for these concerns while endorsing the Sharpe ratio.

Guillaume responded by reiterating that the Sharpe ratio is a measure of return adjusted to risk and it assumes that past realized volatility is a reliable risk indicator. He acknowledged that there are certain strategies, like carry strategy, where risk may not be accurately reflected in the volatility. These strategies collect insurance premiums consistently, but an unexpected event could lead to significant losses.

However, for a strategy like CTA, Guillaume argued that the Sharpe ratio still serves as a good measure. If past volatility hasn't revealed any significant risk, then it's reasonable to use the Sharpe ratio for risk assessment. He implied that despite the potential shortcomings of the Sharpe ratio, for a CTA strategy, there might not be a better alternative available.

Alan presented a hypothetical scenario wherein various trend following programs were ranked based on their Sharpe ratios from 2010 to 2019, a decade without major equity drawdowns. He noted that managers with the best Sharpe ratios might not perform well in a situation like that of 2020 or 2022. He referenced Cliff Asness' paper, suggesting that while the Sharpe ratio measures risk-adjusted return, it doesn't account for a potential second mandate: Crisis Alpha.

Guillaume agreed, stating that trend following strategies, such as theirs, exhibit crisis alpha protection due to their convex profile. He explained that trend followers implicitly replicate a basket of straddle strategies, which show increased returns as the market moves strongly in one direction. However, Guillaume cautioned that while CTAs are uncorrelated to equities, they are not negatively correlated, meaning they don't necessarily offer a hedge in a crisis situation.

Alan then suggested that while the Sharpe ratio captures risk-adjusted return, it doesn't account for higher moments of the distribution like skew and kurtosis, or for whether a strategy performs during an equity drawdown crisis.

Guillaume responded that their goal is to be the best possible manager in all market regimes, not just during equity drawdowns. He argued that there are many macro surprises and various sources of market risk that may trend in certain directions, and their strategy aims to catch and diversify across all these risks.

Alan acknowledged that different managers emphasize different aspects of Trend following: some focus on its defensive nature, others balance absolute return against Crisis Alpha characteristics, and others, like Guillaume, prioritize absolute return.

Niels brought up the Serenity Ratio, a less well-known metric that takes into account actual drawdowns and path dependency. However, he acknowledged that this ratio might be too complicated for most people to calculate, especially as it's not universally applicable. Despite its relevance to a CTA strategy, Niels decided not to delve further into it, suggesting that the conversation move on to other topics.

The Evolution of Metori’s Program Strategy 

Guillaume shared that since his involvement in the program around 2010 or 2011, there has been considerable evolution. A major shift occurred at the end of 2012, when they moved from a traditional approach that involved optimizing combinations of signals using technical analysis to a more statistical process for detecting trends.

This new approach took market correlations into account, both at the portfolio construction stage and in trend detection. Guillaume holds the belief that if markets are interconnected, these connections should be considered in their strategy.

Furthermore, Guillaume emphasized the ongoing nature of their research process. He believes in consistently revisiting and challenging their strategy design choices for potential improvements, leading to continual refinement of the program.

Alan asked Guillaume to explain his approach to signal filtering and correlation research in trend following strategies. Guillaume clarified that signals, for them, imply expected sharp ratios. For every signal in each market, they assess the expected risk-adjusted return for the individual trade.

Regarding correlation, Guillaume's philosophy is to prioritize market data over predefined sector groups or budgets to avoid strategy bias. Despite the challenge and noisiness involved in estimating dynamic correlations, they embrace it. Guillaume noted the irony in people's comfort in predicting expected returns but their hesitance to predict correlations.

Guillaume shared that they've challenged conventional wisdom, specifically around the idea that Markowitz's portfolio theory doesn't work. Instead of discarding the model, they've attempted to identify its limitations and seek improvements, demonstrating their commitment to a data-driven, scientific approach.

Alan discussed correlation and asked about a more effective technique and the optimal timeframe for measuring it. Guillaume referred to the Heisenberg Uncertainty Principle in physics, illustrating the trade-off between accuracy and noise in estimates with shorter timeframes. Despite the noise, Guillaume believes that shorter correlation timeframes are better initially.

Alan agreed that correlation can fluctuate and shift quickly. Guillaume emphasized that the difficulty lies in estimating these shifts accurately and managing the potential noise. He mentioned that their process is not simple, considering these challenges.

Alan further asked about estimating volatility. Guillaume confirmed they use the same approach as with correlation. He explained that volatility and correlation estimates are challenging, especially in portfolio construction, and can lead to changing market regimes being overlooked if long term correlations are used.

The Importance of a Systematic Process

Niels asked whether their models were purely systematic or whether there was a human decision-making element in choosing and changing strategy parameters.

Guillaume clarified that their strategy is strictly systematic, meaning that it doesn't involve discretionary decision-making. The model they use is believed to be the best with the optimal parameterization and doesn't require day-to-day recalibration. 

They base the design of their model on certain 'universal truths' and apply a rigorous scientific process to validate their hypotheses. 

If parameters needed to change regularly, it should be part of a systematic process that can be validated through backtesting. 

Thoughts About Replication and Manager Selection

In response to Niels's question about replication of Commodity Trading Advisor (CTA) strategies, Guillaume pointed out two different approaches: oversimplification (open-source) and reverse-engineering. He expressed concern about fiduciary obligations and transparency, suggesting that managers should be responsible for their portfolio decisions rather than just filtering returns of others.

When asked how he would advise investors to evaluate different CTA strategies, Guillaume admitted it's a difficult task and that there isn't a "silver bullet." He suggested preference for quantitative and systematic managers, with caution towards those claiming to be systematic but with high discretionary input. 

Lastly, Guillaume emphasized the importance of transparency, advising investors to understand where the returns and risks are coming from in their investments.

Active Investment in Chinese Markets and Levels of Market Diversification

Guillaume discussed their presence in China through a subsidiary, trading in the Chinese market specifically for onshore Chinese investors. He mentions that the returns have been incredible and complement Western trends, suggesting that a strategy that includes both Western and Chinese markets in its investment universe would be beneficial.

He emphasized that having a larger investment universe is generally better, but the challenge arises when markets are correlated. He noted that the popular rule of thumb, that the Sharpe ratio improves with the square root of the number of markets, applies only if markets are uncorrelated.

Guillaume cautioned that simply adding more markets to an investment universe might not bring additional risk premia if these markets are highly correlated. He gave the example of adding more European equity indices, which would have a minimal effect due to their high correlation. He also mentioned the operational costs and specific risks of adding new markets, implying a careful, considered approach to expanding their investment universe.

Alan and Guillaume discussed the potential of expanding into more markets, such as China, to capitalize on new risk factors. Guillaume acknowledged that while having more markets can be beneficial, issues such as scalability and liquidity can be potential challenges. He also mentioned a new flagship project for 2023 that incorporates Chinese market exposures.

Alan raised concerns about risks associated with trading in markets that are subject to geopolitical issues, potential market shutdowns, or the implementation of capital controls. Guillaume acknowledged these risks, mentioning that they are present not only in the Chinese markets but also in Western ones. He highlighted that specific risks, such as the non-deliverable currency in the Chinese markets, need to be recognized and priced into their models. He emphasizes that counterparty risk must be considered as part of their investment strategy.

Guillaume also indicated that the decision to expand into more markets should be made on a case-by-case basis. He mentioned that many of the largest commodity markets are in China, making it a natural place to look for opportunities, but they are also being cautious about their approach to these markets.

Perspectives on ESG

Niels and Guillaume discussed the complexities of integrating ESG (Environmental, Social, and Governance) considerations into their investment strategies, particularly with the consideration of expanding into the Chinese market. 

Guillaume acknowledged the difficulties inherent in building a measure of ESG risk as a Commodity Trading Advisor (CTA), considering the need to aggregate various types of equities, bonds, foreign exchange, and commodities.

The existing ESG scoring systems for different instruments don't clearly translate to situations where investments are short. In response to these challenges, Guillaume's team has developed a methodology to aggregate scores from different instruments and account for both long and short positions. This methodology has been shared with investors and academics and has received positive feedback.

However, Guillaume also noted that ESG considerations ultimately come down to investor choice. For example, some investors in their UCITS fund prefer to stay away from certain commodities and Chinese markets due to ESG concerns. Niels appreciated Guillaume's explanation and acknowledged the difficulty of the ESG question for everyone in the field.

Maintaining Investor Confidence in Trend Following and Problems of Timing Investment

Niels and Guillaume discussed the challenges of maintaining investor confidence in trend-following strategies, particularly during periods of less remarkable performance. Guillaume noted the importance of setting correct expectations with investors, warning against expecting consistently outstanding performance.

He emphasized the inherent volatility of these investments and the need to accept that some years will not yield significant positive returns. He cited behavioural finance, noting that humans are risk averse and often focus on short-term losses rather than long-term trends. 

Guillaume also warned against trying to time investments in Commodity Trading Advisors (CTAs), stating that he would incorporate such a timing strategy into his models if it were possible.

Niels agreed with Guillaume's sentiments, noting that they regularly advise clients against trying to time their investments. Guillaume pointed out that the nature of CTAs, with their random noise and convexity, makes timing investments especially tricky. He contrasted this with carry strategies, where timing might be more feasible due to steady returns interrupted by sudden bursts of negative performance. Niels wraped up the conversation by acknowledging the pervasive pressure to time investments, especially due to the influence of financial media and experts.

Allocating to Trend Following as Part of a Broader Portfolio

Alan inquired about the role of trend-following in a broader multi-asset portfolio and asks Guillaume's perspective on optimal allocation to this strategy. Guillaume responded by emphasizing the long-term positive impact of integrating trend-following into a bond-equity portfolio. However, he also noted that it's a complex investment decision that requires a sophisticated understanding of the strategy's benefits.

When asked if any adjustments should be made to the trend-following program to optimize its role in a portfolio, such as capping the equity beta or altering its speed, Guillaume stated that these decisions should be left to the investor. He emphasized that their focus is on refining their trend-following program to be the best it can be, rather than managing overlays or trying to influence investor decisions.

Alan queried why trend-following strategies haven't seen as much adoption in Europe compared to the US. Guillaume suggested that this might be due to misunderstandings about trend-following, which might be viewed as too exotic or complex for some portfolios. Additionally, the fact that these strategies don't predict outcomes might also deter some investors.

Niels then asked if there's such a thing as too much money in trend-following from a mathematical or research perspective. Guillaume responded that the size of their industry is minuscule compared to the markets they trade in, making him sceptical of the idea that CTAs could significantly move the market.

Niels remarked on the robustness of trend-following, noting that there have been very few instances of well-known CTAs experiencing major financial setbacks.

Misconceptions of Trend Following 

Niels asked Guillaume about misconceptions he frequently hears about trend-following that he disagrees with. Guillaume pointed out two primary misconceptions: first, the belief that CTAs are long volatility, which is not accurate as they are long on large moves, not volatility; second, the idea that it's possible to time investment in CTAs, which he also disagrees with.

Outlook for 2023

Looking ahead at the rest of 2023, Guillaume expressed that the launch of their new fund is an exciting prospect. He also mentioned his interest in the ongoing changes in the financial landscape since 2020, including shifts in interest rates and inflation. He highlighted the advantage of trend-following strategies, which are flexible and can adapt to such changes.

This is based on an episode of Top Traders Unplugged, a bi-weekly podcast with the most interesting and experienced investors, economists, traders and thought leaders in the world. Sign up to our Newsletter or Subscribe on your preferred podcast platform so that you don’t miss out on future episodes.