top of page

Insights into the World of Quants | Marko Momentum

From November 4 to 6, The Quant Conference took place in online format for the first time. This article summarises the topics and discussions of the three conference days.

Day 1

The event was kicked off by the founder of the Quant Conference, Nikita Fadeev, and the moderator Stuart MacDonald (Bride Valley Partners). The first keynote speaker was Peter Carr (NYU). In a very technical presentation he introduced his concept for a semi-static hedging strategy.

Robert Frey (FQS Capital) then spoke about challenges in dealing with fat tails. He explained which fundamentally different concepts describe arithmetic (i.e. additive) and geometric (multiplicative) processes. The point is to abandon naive empiricism and model actual processes instead of placing models above practical experience or blindly trusting them.

Coin Toss Experiment

Despite a clear statistical advantage, almost 30 percent of the participants lost (almost) all their stakes.

Source: Victor Haghani, The Quant Conference

The next presentation was given by Victor Haghani (Elm Partners), who presented a version of the coin toss experiment with amazing results: Despite a favourable 60-40 distribution of heads and tails, almost 30 percent of the participants lost (almost) all their stakes. According to Haghani, the reasons for this are erratic betting behaviour, too high or too low bets and the irrational tendency to bet on tails after a series of heads due to a „feeling“ despite the lower probability. He then went on to discuss position sizing via the Kelly criterion. In his opinion, too few investors deal with the crucial issue of position size and think too much about the selection of the best investments, which is actually even less important.

After a networking break it was the turn of Mark Yusko (Morgan Creek Capital Management). He said that many people overestimate the change that happens in two years, but at the same time underestimate what can change in ten years. The biggest danger, he said, was therefore to standstill. He said that the best companies understand how to profit from emerging changes. Accordingly, the best portfolios are those that overweight transforming technologies. As specific examples he cited e-sports, but above all crypto and blockchain investments: this technology has the potential to become the driving force of the future and generate more wealth than the Internet.


The longer the technology survives, the more robust it becomes.

Source: Mark Yusko, The Quant Conference

In the panel that followed, Boris Lerner (Morgan Stanley), Marty Lueck (Aspect Capital), Francesco Filia (Fasanara Capital) and Gary Bergstrom (Acadian Asset Management) exchanged views on whether quant strategies can deliver sustained returns or it is just an endless arms race. Lueck saw the development more as a research challenge to identify and exploit new strategies and to be able to abandon them in time. It is also important that the underlying ideas and concepts are understandable for investors. Lerner said that there have always been risks of external shocks for quants, but that competition is also intensifying. Bergstrom, who has been in business since the late 1960s, agreed with this and spoke of an evolutionary arms race. Filia added in the discussion that today's market micro-structure differs significantly from earlier times. In addition, he sees the technology sector in its importance no longer as such, but rather as a whole, own economy.

Next came keynote presentations from Jean-Philippe Bouchaud (Capital Fund Management) and Aaron Brown (Courant Institute). Bouchaud spoke about the economy as a complex system in which people interact. Therefore heterogeneous market participants should be assumed in models. Brown presented his scenario model and analysed its significance in relation to the corona crisis.

The second panel of the day was made up of Stan Beckers (AQR Asset Management), Michael Weinberg (APG) and a third participant (off record). The topic was sustainable investments. Beckers explained the theory according to which a limitation of the available universe should be a disadvantage for the portfolio, but also said that this could hardly be shown or disproved in practice, so it was ultimately a matter of opinion. It was added that the risks matter most and that these may ultimately be lower due to active engagement in ESG investments. Weinberg pointed out that in addition to the factors discussed, costs also play a role.

The final presentation of the day was given by Igor Halperin (Fidelity) on the topic of combined findings from models of physics and machine learning. He explained that equilibrium models such as CAPM are not well suited for non-stationary, dynamic markets. He then presented the formulas and relationships of his model.

Day 2

The second day began with a keynote presentation by Vishwanath Tirupattur (Morgan Stanley) on the past, present and future of quant finance research. He spoke about how, despite uniform terms such as Value Investing, there is no clear definition of what this means exactly, but that it is the details that have a major impact on the outcome. At the same time, he said, simple rules and common sense rather than complexity are important. All in all, Tirupattur sees no clear distinction between quant and non-quant anymore, as the corresponding tools are used in many sub-areas today.

In the first panel Ashley Lester (Schroders) and three other participants who were “off record” discussed the topic of Man versus Machine. Lester said that discretionary traders today act very differently than in the past and are supported by machines. When people win, it's only with machines – the natural evolution of markets is from discretionary to quantitative trading. However, in extreme phases such as the US elections in 2016 or the Corona crash in 2020, human macro traders delivered the better results. It was mentioned that this is an N-problem of insufficient data from past scenarios, which causes high risk. One solution is to intervene in these situations by reducing overall risk exposure while leaving the algorithm untouched.

When people win, it's only with machines.

Next came a presentation by Sushil Wadhwani (QMA Wadhwani) on the topic of diversification and risk management before the networking break.

The second panel was composed of Yuval Reisman (YRD Capital), Dmitry Tokarev (Copper), Peter Habermacher (Aaro Capital) and Amos Nadler (Fabriik) and focused on developments in the crypto sector. Reisman said that an increasing activity of institutional investors was observed, which was accompanied by a shift from spot to futures markets. Tokarev agreed and added that the perception is increasingly positive; in addition, better security solutions exist today than they did a few years ago. However, counterparty risk remains problematic. Nadler agreed with this. One should only work with trustworthy partners. The big advantage of crypto for investors is the fundamentally uncorrelated behaviour towards other assets. However, one must also be invested early in order not to miss the boat. Institutional investors, who are limited by regulations, can position themselves via corresponding funds. Habermacher added that there are still many information asymmetries that open up opportunities for active managers. As far as classic portfolios are concerned, even a minimal allocation to crypto had an enormous yield effect without significantly higher risks.

The Crypto Panel

Source: The Quant Conference

In the keynote presentation that followed, Alex Lipton (SilaMoney) said that the old banking and financial sector could be about to come to an end. Traditional transactions are too complicated; a distributed, efficient, trusted system is an attractive alternative. In this way, the biggest problem from the Internet age, the trust gap, could be closed. However, Bitcoin and Ethereum also have inherent difficulties and it could turn out to be both an advantage and a disadvantage that anyone can participate and transactions are public.

The third panel of the day also revolved around the topic of crypto. Max Boonen (B2C2) spoke of liquidity being much less fragmented here compared to classic foreign exchange. Kevin Zhou (Galois Capital) agreed and added that one could distinguish between US and non-US liquidity, but that ultimately more or less the same people would make the market. Daniel Matuszewski (CMS Holdings) added that structurally a lot has improved; the processes run through stable coins and even larger investors are looking at crypto. Kyle Davis (Three Arrows Capital) explained that the prime broker model of bundled services is not sustainable in the crypto sector. The professionals have learned over time to handle everything themselves and can therefore work without intermediaries. Boonen agreed: There is no classic one-stop effect in crypto because the processes work well on a standalone basis.

„There is no classic one-stop effect in crypto.“ (Max Boonen, B2C2)

Next Campbell Harvey (Duke University) spoke about the negative effects of data mining. He showed examples of methodologically clearly flawed research in recognized studies. As a cause he denounced the toxic (academic) culture of only rewarding success: This is a strong incentive for data mining, especially against the background that a publication in one of the top journal resembles a job guarantee. Harvey speaks of a new kind of risk that published results can no longer be trusted because data has been strategically selected or adjusted or cleaned in some way. As a solution to that, incentives should be based on a professional research process itself.

Data Mining

Source: Campbell Harvey, The Quant Conference

The last panel of the day was about alternative data. Jasmine Burgess (Coremont) explained that a diversified pool of such data was needed, as well as a macro-overlay to assess when which of it were relevant. The others agreed. There are countless alternative data sets available today, but about 95 percent of them do not contribute to better forecasting corporate earnings. That's why it is important to use effective, pragmatic tests based on a clear idea, and to discard inappropriate data quickly enough given the many possibilities. When asked whether alternative data would provide an unfair advantage, Burgess replied that it could be seen as such. However, specialized investors could in turn try to discover the dearly bought advantages in price data through momentum screenings and thus take advantage of it. Jared Broad (QuantConnect) said that alternative data may require regulation. Margaret Holen (Princeton University) added that the incremental value that alternative data can add is critical. In addition, it would allow not only investments but also the companies themselves to be more successful, thus reducing the gap between Wall Street and Main Street.

Day 3

The last day started with a panel on the topic of technology in the quant area. The panel was chaired by Christina Qi (Databento) and the participants Chao Yan (SIGTech) and Palak Patel (S3 Partners). Patel said that the cloud was a game changer in the industry, enabling scalability without associated infrastructure issues. But there is so much choice in data now that it is difficult to manage it efficiently and integrate it into the investment process; the last mile on the way to the portfolio managers' decision takes too long. Yan agreed with this and said that the data processing effort required to arrive at the final findings was considerable. Nevertheless, Patel said that alternative data would bring clear benefits. He illustrated this with a case study of real-time short interest before the negative report on Nikola Motors was announced.

In the keynote presentation that followed, Darko Matovski (causaLens) spoke about the exciting future topic of causal artificial intelligence. While classical machine learning based on correlation works excellently for static applications, it is unsuitable for markets because correlation does not imply causality. Therefore, a new approach is needed with causal artificial intelligence. Its development is still in its infancy, but a first use case is to identify the most meaningful of many time series for a specific application via filtering out spurious correlations.

Traditional Machine Learning

Source: Darko Matovski, The Quant Conference

The second panel, moderated by Terri Duhon (Morgan Stanley), focused on developing expertise and building efficient quant teams. Participants in the discussion were Charles-Albert Lehalle (Capital Fund Management), Gordon Ritter (Ritter Alpha) as well as a third, anonymous expert. According to Lehalle, mathematics is particularly in demand in the quant field, as it represents a common language across different areas. Ritter said that genuine interest in the markets is crucial. At the same time, he said, it requires the attitude of being able to work on individual problems for long periods of time, as well as a healthy dose of skepticism and a commitment to precise, scientific processes. The others agreed with this and it was added that ultimately it is about the ability to solve really difficult problems. In terms of building good teams, everyone agreed that the most important thing was to work together and to appreciate each other in order to bring out the best in everyone. Alpha is a zero-sum game, and only truly creative, cooperative teams can maintain an advantage in this highly competitive environment.

„Mathematics is particularly in demand in the quant field, as it represents a common language across different areas.“ (Charles-Albert Lehalle, Capital Fund Management)

The last panel focused on diversity and talent acquisition in the quant area. The panelists were Ganesh Mani (Carnegie Mellon), Silvia Stanescu (GAM Systematic), Kathryn M. Kaminski (Alpha Simplex) and Mehrzad Madhavi (FDP Institute).

The final keynote presentation of the conference was given by Peter Niculescu (Capital Market Risk Advisors). He spoke about the lessons quants learned in the turbulent year 2020. In the first place he mentioned not to rely too much on historical data. Some managers had not taken any action in the corona crash because they found it difficult to quantify the situation at all. One solution could be to include future scenarios. According to a survey, other lessons learned were the need for additional, hypothetical stress scenarios and regular review of the risk and investment models used.


Conferences on Quantitative Finance are by nature quite sophisticated; and some topics certainly not easy to understand even for professionals. In view of this, the number of participants was astonishingly large, at around 700. There were only a few theoretical, formula-based presentations – most keynotes and expert panels were very interesting. All in all, the conference was sufficiently entertaining and yet offered a steep learning curve. The participants were able to learn and understand a lot of new things about the industry or refresh and deepen their existing knowledge.

The article was written by Dr Marko Graenitz.


bottom of page