Drawdown Killer

Translation from article about OpenMetrics Solutions LLC in www.institutional-money.com (as of March 2018)

A risk analysis and management system originally developed by scientists from the Institute for Theoretical Physics at the Swiss Federal Institute of Technology (ETH Zurich) is now being commercially marketed by the university spin-off OpenMetrics – and the data is very promising.

It all began with ETH professor Diethelm Würtz’s unquenchable thirst for research; sadly he died in July 2016 after a car accident. He focused on the application of advanced statistical and physical methods in the area of finance, and founded the Econophysics Group within the Department of Physics at the internationally renowned technical university of applied sciences, thus establishing an interdisciplinary branch of science dealing with the theoretical modelling of complex economic systems. His work led, among other things, to the introduction of new stability indicators for financial markets and economic regions. Würtz however was also active on an entrepreneurial level: he (co-)founded the Rmetrics Association, the IT company Finance Online GmbH, and Sidenis AG.

Company foundation 2016

The practical results of his work are also visible in OpenMetrics Solutions GmbH, founded in Zurich in November 2016. The company focuses on complex modelling of regime changes on different investment markets, and the provision of high-performance risk management solutions for the financial industry. The Zurich-based technical university participates in the company, as it did with earlier companies which were founded as a result of research. Since 1996, more than 350 companies were formed this way. Felix Fernandez is the CEO of OpenMetrics Solutions GmbH, with CTO Tobias Setz as the technical lead.

Fernandez has an Electronics & Information Technology degree from Frankfurt University of Applied Sciences, with a specialisation in software simulation environments. In addition, he has many years of practical experience as Senior Advisor Cash & Derivatives Markets at Eurex and Deutsche Börse AG. He has been a research partner of the Rmetrics Association Zurich since 2016. In this role, he was responsible for the transfer of academic research results into real-world applications within the financial industry. Today, Fernandez is in charge of developing the OpenMetrics Solutions business and implementing a value-added product structure for institutional investors.

Tobias Setz studied Computational Science and Engineering, specialising in theoretical physics, at ETH Zurich with professor Würtz. He has meanwhile received his doctorate from ETH's Theoretical Physics Department. Since the establishment of the company, he has been the mastermind behind the OpenMetrics Solutions Technology frameworks; he also implements client solutions.

Partners and clients

Fernandez has already had his first sales successes. Two Swiss pension funds are already using his risk management overlay solution. One of them is the Swiss Federal Pension Fund PUBLICA: with assets under management of almost 40 billion Swiss francs, it is the largest pension fund in Switzerland.

As OpenMetrics has its roots within ETH, it is the consulting company's target to build bridges between academic research and the financial sector. More specifically,

OpenMetrics develops tailor-made risk reports by way of financial engineering. Risk and investment committees then use these reports as a basis for their decision-making processes. Furthermore, the company also offers support in selecting or hedging investment vehicles. Fernandez explains: "For example, if a client pursuing a multi-asset approach in his asset allocation provides us with information on his strategic asset allocation and his benchmarks, our signals can help him decide in which asset classes to increase or decrease hedging positions, and with which instruments."

The Swiss firm is flexible when it comes to shaping its services. In their simplest form, the client can subscribe to risk reports and signals, implementing them in-house himself; however, he can also request tailor-made quantitative research solutions if he so wishes. The latter ranges from the development of specific risk-adjusted indices to the implementation of risk management and portfolio optimisation systems.

Fernandez and Setz not only want to sell their services to institutional investors, but also to data and index providers and exchanges, e.g. in order to develop a new generation of risk management products such as risk warnings for sectors and risk signals as a basis for new risk indices. These might take the form, for example, of a risk-protected EURO STOXX 50, or risk indices as underlyings for new ETFs or new derivatives contracts such as futures and options.

The risk signals OpenMetrics generates could however also be of interest for clearing houses: such signals can be used as a basis for optimising margin requirements. Central bankers could also be potentially interested in the risk signals, as they can be useful when preparing for unexpected movements, for example, on the foreign exchange markets. Thanks to the universal nature of the analyses, the experts from Switzerland can also use their evaluations outside the financial world, for instance when discerning structural changes in consumer data.

Making out regime changes

The main objective of statistical procedures based on the newest research findings is, on the one hand, to recognise regime changes faster and more exactly than competitors and, on the other hand, to convert these changes into a reliable indicator of market stability.

The last step of the analysis is creating a so-called heat map which visualises the different risk regimes and their changes. The 'risk signals' chart illustrates the various risk phases for each sector of the STOXX® EUROPE 600 in different colours. Green indicates no danger, yellow is a neutral stability indicator and orange and red point to high-risk situations. The sectoral breakdown makes sense, as the individual sectors do not develop in a parallel manner.

The STOXX 600 Construction Index, for example, showed no overheating tendencies between January 2016 and January 2018, whilst high and very high risk could be observed in the telecommunications sector. Fernandez explains that the values of the stabilisation indicator can be directly used for the exposure management applications.

Another option is using the technology for portfolio optimisation, when going beyond the optimisation of individual parameters. Taking the generated risk signals as the basis, managing a dynamic hedging process is also possible – the objective is not only to improve risk-adjusted returns, but also to minimise hedging costs.

Dynamic hedging

The dynamic hedge, based on OpenMetrics' risk signals (please refer to the identically named chart in the one-asset-case based on the STOXX 600 index), is suitable for futures contracts, for example. The simulation calculation starting from December 2005 gives a rough idea of the algorithm's timing qualities, and to what extent drawdowns can be averted.

This calculation also takes the major financial and economic crisis from 2007 to 2009 into account. The green line in the upper chart shows the result of the consistent signal implementation for the period from December 2005 to January 2018 in comparison to the STOXX 600 (Net Return) Index (unhedged; dark blue line) and to the STOXX 600 Index (passively hedged by 50%; black). The annualised return of the traditional index thus amounts to an annual 4.98 per cent within the period, with an annual volatility of 14.47 per cent and a Sharpe ratio of 0.34, and including a maximum drawdown of 54.34 per cent and an average drawdown of 14.57 per cent.

The portfolio tracking the dynamically hedged STOXX® EUROPE 600 Index – has a fluctuating level of investment exposure – which can be determined based on the monthly risk signals of OpenMetrics and implemented either by changing the portfolio allocation or using derivatives (futures contracts), offers a significantly better risk/return profile: an annualised return of 5.64 per cent, volatility of 9.25 per cent, and improved risk indicators (e.g. a Sharpe ratio of 0.61, maximum drawdown of only 16.06 per cent and also a significantly improved average drawdown of 6.81 per cent). The average exposure is at approximately 73 per cent.

Hedges: dynamic vs static

The success of the dynamic hedge component is also visible when taking a closer look at the STOXX Europe 600 Index portfolio, half of which is statically hedged. Its return and risk characteristics are materially worse than the dynamically hedged portfolio. A return of 2.46 per cent and volatility of 7.26 per cent lead to a Sharpe ratio of 0.34, a maximum drawdown of 32.43 per cent and an average drawdown of 7.93 per cent. The drawdown comparison illustrates very clearly that the dynamically hedged option can reduce its under-water period. The strategy could have generated a substantial outperformance during the financial and economic crisis.

The fact that the portfolio would have had near to no risk between January 2008 and mid-2009 shows how consistent the concept is in a 'case of emergency'. However, the first three to four months after the market turned on 09 March 2009 would have been missed. The signals observed during the periods from August 2011 to May 2012 (EUR sovereign debt crisis) and from January to October 2016 provoked a mostly complete market withdrawal to cash. It should be noted that this 'money market' phase in 2016 was very protracted, as the market had already reached its turning point in mid-February after six nasty weeks at the beginning of the year.

In this third critical phase everyone was simply running after the market, and nobody was generating an out-performance; thus, the leading edge established in 2008/09 slightly dwindled. In addition, following the six weak weeks at the beginning of 2016, it took at least half a year until the algorithm was 'fully invested' again in autumn.

However, the signal-based regime change approach was definitely in line with the target of reducing risk whilst enhancing return and thus improving the Sharpe ratio. Under real conditions, one would allocate the released cash to non-equity portfolios, thus allowing the generation of additional profits.