Podcast: Lipton and De Prado on Covid-19 and optimal trading strategies

This podcast features two veteran quants, Alex Lipton, chief technical officer at SilaMoney and connection science fellow at MIT, and Marcos Lopez de Prado, co-founder and chief investment officer of True Positive Technologies and professor at Cornell University in New York.

The pair began collaborating recently. While reading de Prado’s book on quantitative investments, Lipton saw the possibility of solving an open question in that volume using a technique called heat potentials. “I got in touch with Marcos and we decided to work on that,” recalls Lipton. When Covid struck, they decided to work together to model the pandemic too.

The results of their first collaboration are presented in a paper published by Risk.net this month. In it, they provide a closed-form solution to finding the optimal thresholds for profit taking and setting stop losses in a mean-reverting market – a problem familiar to market-makers and execution traders. “Whether you are a liquidity provider or a liquidity consumer, the question of when to exit a position is a critical one,” explains de Prado.

Lipton and de Prado were dissatisfied with the way the problem was addressed in the existing literature and sought to create a solution that was both realistic and practical. They did so using heat potentials, a concept borrowed from physics, which allows them to calculate the boundaries of an optimal trading strategy.

Their epidemiological model on Covid-19, published in April, sprang more from a sense of “civic duty”, as they put it. Lipton was already familiar with standard epidemiological models, such as SEIR (susceptible, exposed, infectious and recovered). He previously studied them because he thought they could be adapted to solve a completely different problem – explaining variables in the market capitalisation of cryptocurrencies.

This effort was also motivated partly by dissatisfaction with existing models and a desire to develop a more realistic framework. Their model, called K-SEIR, assumes the distribution of the population is multi-modal, with distinct groups that are very differently affected by the virus.

Furthermore, in a rare case of quantitative finance influencing other fields, “we consider the rate of infection, R0, as an implicit variable, just like implied volatility in Black-Scholes”, says Lipton.

The pair also explain their gloomy outlook on the economic recovery from the pandemic and the value of nowcasting – which they touched on in a recent Risk.net article – in the current environment.  

“There’s uncertainty coming back, as observed in the volatility level and in bids and offers levels,” says de Prado, adding that nowcasting is well suited to situations such as the current one, where there is a wealth of information.

“The economy right now is in a rather dire situation,” says Lipton. “Pandemics unfortunately come with periodicity.” The next time this happens, “we need to be much better prepared”.

INDEX

00:00 Intro

03:45 Optimal trading strategies in mean-reverting markets

07:10 Heat potentials

11:05 Model assumptions

14:25 Results and comparison to existing methods

18:25 The Covid-19 model and the flaws of the standard approaches

28:30 The multi-modal distribution of the population and the unreliability of available data

33:55 What the model tells you

36:35 Application to policy and investment decisions

39:50 Outlook on the economic and financial recovery