Essays on Market Microstructure and Financial Markets Stability

Doctoral thesis (2022)

The present doctoral thesis consists of three main chapters. The chapters of the thesis can be considered independently. Each of the three chapters raises a research question, reviews the related ... [more ▼]

The present doctoral thesis consists of three main chapters. The chapters of the thesis can be considered independently. Each of the three chapters raises a research question, reviews the related literature, proposes a method for the analysis, and, finally, reports results and conclusions. Chapter 1 is entitled Dark Trading and Financial Markets Stability and it is based on a working paper co-authored with Prof. Dr. Jorge Goncalves and Prof. Dr. Roman Kraussl. This paper examines how the implementation of a new dark order -- Midpoint Extended Life Order (M-ELO) on Nasdaq -- impacts financial markets stability in terms of occurrences of mini-flash crashes in individual securities. We use high-frequency order book data and apply panel regression analysis to estimate the effect of dark order trading activity on market stability and liquidity provision. The results suggest a predominance of a speed bump effect of M-ELO rather than a darkness effect. We find that the introduction of M-ELO increases market stability by reducing the average number of mini-flash crashes, but its impact on market quality is mixed. Chapter 2 is entitled Dark Pools and Price Discovery in Limit Order Markets and it is a single-authored work. This paper examines how the introduction of a dark pool impacts price discovery, market quality, and aggregate welfare of traders. I use a four-period model where rational and risk-neutral agents choose the order type and the venue and obtain the equilibrium numerically. The comparative statics on the order submission probability suggests a U-shaped order migration to the dark pool. The overall effect of dark trading on market quality and aggregate welfare was found to be positive but limited in size and depended on market conditions. I find mixed results for the process of price discovery. Depending on the immediacy need of traders, price discovery may change due to the presence of the dark venue. Chapter 3 is entitled Machine Learning and Market Microstructure Predictability and it is another single-authored piece of work. This paper illustrates the application of machine learning to market microstructure research. I outline the most insightful microstructure measures, that possess the highest predictive power and are useful for the out-of-sample predictions of such features of the market as liquidity volatility and general market stability. By comparing the models' performance during the normal time versus the crisis time, I come to the conclusion that financial markets remain efficient during both periods. Additionally, I find that high-frequency traders activity is not able to forecast accurately neither of the market features. [less ▲]

MULTI-STAGE PROCESS FOR A HIGHER FLEXIBILITY OF BIOGAS PLANTS WITH (CO-) FERMENTATION OF WASTE – OPTIMISATION AND MODELLING

Doctoral thesis (2022)

The European Union has been striving to become the first climate-neutral continent by 2050. This implicates an intensified transition towards sustainability. The most applied renewable energy sources are ... [more ▼]

The European Union has been striving to become the first climate-neutral continent by 2050. This implicates an intensified transition towards sustainability. The most applied renewable energy sources are the sun and wind, which are intermittent. Thus, great fluctuating shares in the energy network are expected within the next years. Consequently, there might occur periods of no congruence between energy demand and energy supply leading to destabilization of the electricity grid. Therefore, an urgency to overcome the intermittency arises. One feasible option is to use a third renewable energy source, biomass, which can be produced demand-oriented. Hence, a flexible biogas plant running on a two-stage mode, where the first stage would serve as a storage for liquid intermediates, could be a viable option to create demand-driven and need-orientated electricity. Since vast amounts of food waste are thrown away each year (in 2015 they amounted 88 mio. tones within the EU-28, accounting for ca. 93 TWh of energy), one could energetically recover this substrate in the above-described process. This is a promising concept, however, not widely applicable as it faces many challenges, such as technical and economical. Additionally, food waste is inhomogeneous, and its composition is country- and collecting season dependent. The motivation of this work was to contribute to a vaster understanding of the two-stage anaerobic digestion process by using food waste as a major substrate. At first, an innovative substitute for a heterogeny food waste was introduced and examined at two different loadings and temperature modes. It was proven that the Model Kitchen Waste (MKW) was comparable to the real Kitchen Waste (KW), at mesophilic and thermophilic mode for an organic loading in accordance with the guideline VDI 4630 (2016). For an “extreme” loading, and mesophilic mode, the MKW generated similar biogas, methane, and volatile fatty acid (VFA) patterns as well. Furthermore, another two MKW versions were developed, allowing covering a variety of different organic wastes and analyzing the impact of fat content on the biogas production. Afterwards, a semi-continuous one-stage experiment of 122 days was conducted. It was followed by an extensive semi-continuous two-stage study of almost 1.5-year runtime. Different loadings and hydraulic retention times were investigated in order to optimize this challenging process. Additionally, the impact of co-digestion of lignocellulose substrate was analyzed. It was concluded that two-stage mode led to a higher biogas and methane yield than the one-stage. However, the former posed challenges related to the stability and the process maintenance. Additionally, it was found that co-digestion of food waste and maize silage results in methane yield, atypical for the acidic stage. Apart from the experiments, the Anaerobic Digestion Model No. 1 (ADM1), originally developed for wastewater, was modified so that it would suit the anaerobic digestion of food waste of different fat contents, at batch and semi-continuous mode consisting of one- and two-stages. The goodness of fit was assessed by the Normalized Root Mean Square Error (NRMSE) and coefficient of efficiency (CE). For the batch mode, two temperature modes could be properly simulated at loadings conform and nonconform to the VDI 4630 (2016). For each mode, two different sets of parameters were introduced, namely for substrates of low-fat content and for substrates of middle/high fat content (ArSo LF and ArSo MF, with LF standing for low fat and MF for middle fat). The models could be further validated in another experiment, also using a co-digestion of lignocellulose substances. Further, parameters estimated for the batch mode, were applied for the semi-continuous experiment. It proved successful, however, due to a high amount of butyrate (HBu) and valerate (HVa), the model underwent calibration so that it could better predict the acids (model developed for one-stage semi-continuous experiment was called: ArSo M LF*). This could be validated on another semi-continuous reactor running on one-stage. Finally, the acidic-stage of the two-stage mode was analyzed. The model applied for one-stage fitted the data of the two-stage mode as far as the VFA are concerned. Nevertheless, due to a vast amount of acids, it was adjusted and called ArSo M LF**. [less ▲]

A Quick and Dirty Guide to Digital Storytelling

Speeches/Talks (2022)

Machine Learning in Space Weather and how to handle it

Presentation (2022, July 06)

Machine leaning (ML), an imposing but not necessarily new method, is living today its golden age by achieving unforeseen results in numerous industrial applications. On the other side, Space Weather (SW ... [more ▼]

Machine leaning (ML), an imposing but not necessarily new method, is living today its golden age by achieving unforeseen results in numerous industrial applications. On the other side, Space Weather (SW), which describes changing environmental conditions in near-Earth space, is becoming more and more important to our society. But how can SW benefit from the ongoing ML revolution? In the last decade, many researchers, like E. Camporeale, showed that SW possesses all the ingredients often required for a successful ML application. Using this large and freely data set of in situ and remote observations collected over several decades of space missions, it is possible to forecast and nowcast solar activity and thus protect our increasing satellites constellations and us! In this talk, we will give a warm introduction to this field and point out a number of open challenges that we believe is worth to discuss and to undertake. [less ▲]

Mean-field theory for the structure of strongly interacting active liquids

in Journal of Chemical Physics (2022)

Active systems, which are driven out of equilibrium by local non-conservative forces, exhibit unique behaviors and structures with potential utility for the design of novel materials. An important and ... [more ▼]

Active systems, which are driven out of equilibrium by local non-conservative forces, exhibit unique behaviors and structures with potential utility for the design of novel materials. An important and difficult challenge along the path toward this goal is to precisely predict how the structure of active systems is modified as their driving forces push them out of equilibrium. Here, we use tools from liquid-state theories to approach this challenge for a classic minimal active matter model. First, we construct a nonequilibrium mean-field framework that can predict the structure of systems of weakly interacting particles. Second, motivated by equilibrium solvation theories, we modify this theory to extend it with surprisingly high accuracy to systems of strongly interacting particles, distinguishing it from most existing similarly tractable approaches. Our results provide insight into spatial organization in strongly interacting out-of-equilibrium systems. [less ▲]

Economic Sanctions and International Investment Law: Investor and State Defences

Scientific Conference (2022, July 04)

This workshop examines the law and practice of economic sanctions in light of current developments in the Ukraine-Russia conflict. It explores the legality and implementation of, and possible challenges ... [more ▼]

This workshop examines the law and practice of economic sanctions in light of current developments in the Ukraine-Russia conflict. It explores the legality and implementation of, and possible challenges to, economic sanctions imposed by a number of international actors (including the US, the EU and the UK), as well as their potential impact on businesses and dispute resolution. [less ▲]

Non-Orthogonal Multiple Access for Next-Generation Satellite Systems: Flexibility Exploitation and Resource Optimization

Doctoral thesis (2022)

In conventional satellite communication systems, onboard resource management follows pre-design approaches with limited flexibility. On the one hand, this can simplify the satellite payload design. On the ... [more ▼]

In conventional satellite communication systems, onboard resource management follows pre-design approaches with limited flexibility. On the one hand, this can simplify the satellite payload design. On the other hand, such limited flexibility hardly fits the scenario of irregular traffic and dynamic demands in practice. As a consequence, the efficiency of resource utilization could be deteriorated, evidenced by mismatches between offered capacity and requested traffic in practical operations. To overcome this common issue, exploiting multi-dimension flexibilities and developing advanced resource management approaches are of importance for next-generation high-throughput satellites (HTS). Non-orthogonal multiple access (NOMA), as one of the promising new radio techniques for future mobile communication systems, has proved its advantages in terrestrial communication systems. Towards future satellite systems, NOMA has received considerable attention because it can enhance power-domain flexibility in resource management and achieve higher spectral efficiency than orthogonal multiple access (OMA). From ground to space, terrestrial-based NOMA schemes may not be directly applied due to distinctive features of satellite systems, e.g., channel characteristics and limited onboard capabilities, etc. To investigate the potential synergies of NOMA in satellite systems, we are motivated to enrich this line of studies in this dissertation. We aim at resolving the following questions: 1) How to optimize resource management in NOMA-enabled satellite systems and how much performance gain can NOMA bring compared to conventional schemes? 2) For complicated resource management, how to accelerate the decision-making procedure and achieve a good tradeoff between complexity reduction and performance improvement? 3) What are the mutual impacts among multiple domains of resource optimization, and how to boost the underlying synergies of NOMA and exploit flexibilities in other domains? The main contributions of the dissertation are organized in the following four chapters: First, we design an optimization framework to enable efficient resource allocation in general NOMA-enabled multi-beam satellite systems. We investigate joint optimization of power allocation, decoding orders, and terminal-timeslot assignment to improve the max-min fairness of the offered-capacity-to-requested-traffic ratio (OCTR). To solve the mixed-integer non-convex programming (MINCP) problem, we develop an optimal fast-convergence algorithmic framework and a heuristic scheme, which outperform conventional OMA in matching capacity to demand. Second, to accelerate the decision-making procedure in resource optimization, we attempt to solve optimization problems for satellite-NOMA from a machine-learning perspective and reveal the pros and cons of learning and optimization techniques. For complicated resource optimization problems in satellite-NOMA, we introduce deep neural networks (DNN) to accelerate decision making and design learning-assisted optimization schemes to jointly optimize power allocation and terminal-timeslot assignment. The proposed learning-optimization schemes achieve a good trade-off between complexity and performance. Third, from a time-domain perspective, beam hopping (BH) is promising to mitigate the capacity-demand mismatches and inter-beam interference by selectively and sequentially illuminating suited beams over timeslots. Motivated by this, we investigate the synergy and mutual influence of NOMA and BH for satellite systems to jointly exploit power- and time-domain flexibilities. We jointly optimize power allocation, beam scheduling, and terminal-timeslot assignment to minimize the capacity-demand gap. The global optimal solution may not be achieved due to the NP-hardness of the problem. We develop a bounding scheme to tightly gauge the global optimum and propose a suboptimal algorithm to enable efficient resource assignment. Numerical results demonstrate the synthetic synergy of combining NOMA and BH, and their individual performance gains compared to the benchmarks. Fourth, from the spatial domain, adaptive beam patterns can adjust the beam coverage to serve irregular traffic demand and alleviate co-channel interference, motivating us to investigate joint resource optimization for satellite systems with flexibilities in power and spatial domains. We formulate a joint optimization problem of power allocation, beam pattern selection, and terminal association, which is in the format of MINCP. To tackle the integer variables and non-convexity, we design an algorithmic framework and a low-complexity scheme based on the framework. Numerical results show the advantages of jointly optimizing NOMA and beam pattern selection compared to conventional schemes. In the end, the dissertation is concluded with the main findings and insights on future works. [less ▲]

LSTM-Based Distributed Conditional Generative Adversarial Network for Data-Driven 5G-Enabled Maritime UAV Communications

in IEEE Transactions on Intelligent Transportation Systems (2022)

5G enabled maritime unmanned aerial vehicle (UAV) communication is one of the important applications of 5G wireless network which requires minimum latency and higher reliability to support mission ... [more ▼]

5G enabled maritime unmanned aerial vehicle (UAV) communication is one of the important applications of 5G wireless network which requires minimum latency and higher reliability to support mission-critical applications. Therefore, lossless reliable communication with a high data rate is the key requirement in modern wireless communication systems. These all factors highly depend upon channel conditions. In this work, a channel model is proposed for air-to-surface link exploiting millimeter wave (mmWave) for 5G enabled maritime unmanned aerial vehicle (UAV) communication. Firstly, we will present the formulated channel estimation method which directly aims to adopt channel state information (CSI) of mmWave from the channel model inculcated by UAV operating within the Long Short Term Memory (LSTM)-Distributed Conditional generative adversarial network (DCGAN) i.e. (LSTM-DCGAN) for each beamforming direction. Secondly, to enhance the applications for the proposed trained channel model for the spatial domain, we have designed an LSTM-DCGAN based UAV network, where each one will learn mmWave CSI for all the distributions. Lastly, we have categorized the most favorable LSTM-DCGAN training method and emanated certain conditions for our UAV network to increase the channel model learning rate. Simulation results have shown that the proposed LSTM-DCGAN based network is vigorous to the error generated through local training. A detailed comparison has been done with the other available state-of-the-art CGAN network architectures i.e. stand-alone CGAN (without CSI sharing), Simple CGAN (with CSI sharing), multi-discriminator CGAN, federated learning CGAN and DCGAN. Simulation results have shown that the proposed LSTM-DCGAN structure demonstrates higher accuracy during the learning process and attained more data rate for downlink transmission as compared to the previous state of artworks. [less ▲]

Hybrid Active-Passive Reconfigurable Intelligent Surface-Assisted Multi-User MISO Systems

Poster (2022, July 04)

We consider a multi-user multiple-input single- output (MISO) communications system which is assisted by a hybrid active-passive reconfigurable intelligent surface (RIS). Unlike conventional passive RISs ... [more ▼]

We consider a multi-user multiple-input single- output (MISO) communications system which is assisted by a hybrid active-passive reconfigurable intelligent surface (RIS). Unlike conventional passive RISs, hybrid RIS is equipped with a few active elements with the ability to reflect and amplify incident signals to significantly improve the system performance. Towards a fairness design, we maximize the minimum rate among all users through jointly optimizing the transmit beamforming vectors and RIS reflecting/amplifying coefficients. Combining tools from block coordinate ascend and successive convex approximation, the challenging nonconvex problem is efficiently solved by a low- complexity iterative algorithm. The numerical results show that a hybrid RIS with 4 active elements out of a total of 50 elements with a power budget of −1 dBm offers an improvement of up to 80% to the considered system, while that achieved by a fully passive RIS is only 27% [less ▲]

Investigating the Industry 4.0 paradigm change in brownfield manufacturing facilities: from data-driven approaches to blockchain-based traceability

Doctoral thesis (2022)