Research articles for the 2021-04-11

CLVSA: A Convolutional LSTM Based Variational Sequence-to-Sequence Model with Attention for Predicting Trends of Financial Markets
Jia Wang,Tong Sun,Benyuan Liu,Yu Cao,Hongwei Zhu

Financial markets are a complex dynamical system. The complexity comes from the interaction between a market and its participants, in other words, the integrated outcome of activities of the entire participants determines the markets trend, while the markets trend affects activities of participants. These interwoven interactions make financial markets keep evolving. Inspired by stochastic recurrent models that successfully capture variability observed in natural sequential data such as speech and video, we propose CLVSA, a hybrid model that consists of stochastic recurrent networks, the sequence-to-sequence architecture, the self- and inter-attention mechanism, and convolutional LSTM units to capture variationally underlying features in raw financial trading data. Our model outperforms basic models, such as convolutional neural network, vanilla LSTM network, and sequence-to-sequence model with attention, based on backtesting results of six futures from January 2010 to December 2017. Our experimental results show that, by introducing an approximate posterior, CLVSA takes advantage of an extra regularizer based on the Kullback-Leibler divergence to prevent itself from overfitting traps.

Frequency-Dependent Higher Moment Risks
Jozef Barunik,Josef Kurka

Based on intraday data for a large cross-section of individual stocks and exchange traded funds, we show that short-term as well as long-term fluctuations of realized market and average idiosyncratic higher moments risks are priced in the cross-section of asset returns. Specifically, we find that market and average idiosyncratic volatility and kurtosis are significantly priced by investors mainly in the long-run even if controlled by market moments and other factors, while skewness is mostly short-run phenomenon. A conditional pricing model capturing the time-variation of moments confirms downward-sloping term structure of skewness risk and upward-sloping term structure of kurtosis risk, moreover the term structures connected to market skewness risk and average idiosyncratic skewness risk exhibit different dymanics.

Functional quantization of rough volatility and applications to the VIX
Ofelia Bonesini,Giorgia Callegaro,Antoine Jacquier

We develop a product functional quantization of rough volatility. Since the quantizers can be computed offline, this new technique, built on the insightful works by Luschgy and Pages, becomes a strong competitor in the new arena of numerical tools for rough volatility. We concentrate our numerical analysis to pricing VIX Futures in the rough Bergomi model and compare our results to other recently suggested benchmarks.

Is culture related to strong science? An empirical investigation
Mahmood Khosrowjerdi,Lutz Bornmann

National culture is among those societal factors which could influence research and innovation activities. In this study, we investigated the associations of two national culture models with citation impact of nations (measured by the proportion of papers belonging to the 10% and 1% most cited papers in the corresponding fields, PPtop 10% and PPtop 1%). Bivariate statistical analyses showed that of six Hofstede's national culture dimensions (HNCD), uncertainty avoidance and power distance had a statistically significant negative associa-tion, while individualism and indulgence had a statistically significant positive associationwith both citation impact indicators. The study also revealed that of two Inglehart-Welzel cultural values (IWCV), the value survival versus self-expression is statistically significantly related to citation impact indicators. We additionally calculated multiple regression analyses controlling for the possible effects of confounding factors including national self-citations, international co-authorships, invest-ments in research and development, international migrant stock, number of researchers ofeach nation, language, and productivity. The results revealed that the statistically significant associations of HNCD with citation impact indicators disappeared. But the statistically significant relationship between survivals versus self-expression values and citation impact indicators remained stable even after controlling for the confounding variables. Thus, the freedom of expression and trust in society might contribute to better scholarly communication systems, higher level of international collaborations, and further quality research.

On A Class Of Rank-Based Continuous Semimartingales
David Itkin,Martin Larsson

Using the theory of Dirichlet forms we construct a large class of continuous semimartingales on an open domain $E \subset \mathbb{R}^d$, which are governed by rank-based, in addition to name-based, characteristics. Using the results of Baur et al. [Potential Analysis, 38(4):1233-1258,2013] we obtain a strong Feller property for this class of diffusions. As a consequence we are able to establish the nonexistence of triple collisions and obtain a simplified formula for the dynamics of its rank process. We also establish conditions under which the process is ergodic. Our main motivation is Stochastic Portfolio Theory (SPT), where rank-based diffusions of this type are used to model financial markets. We show that three main classes of models studied in SPT -- Atlas models, generalized volatility-stabilized models and polynomial models -- are special cases of our framework.

Optimal Market Making by Reinforcement Learning
Matias Selser,Javier Kreiner,Manuel Maurette

We apply Reinforcement Learning algorithms to solve the classic quantitative finance Market Making problem, in which an agent provides liquidity to the market by placing buy and sell orders while maximizing a utility function. The optimal agent has to find a delicate balance between the price risk of her inventory and the profits obtained by capturing the bid-ask spread. We design an environment with a reward function that determines an order relation between policies equivalent to the original utility function. When comparing our agents with the optimal solution and a benchmark symmetric agent, we find that the Deep Q-Learning algorithm manages to recover the optimal agent.

Simulation-based optimisation of the timing of loan recovery across different portfolios
Arno Botha,Conrad Beyers,Pieter de Villiers

A novel procedure is presented for the objective comparison and evaluation of a bank's decision rules in optimising the timing of loan recovery. This procedure is based on finding a delinquency threshold at which the financial loss of a loan portfolio (or segment therein) is minimised. Our procedure is an expert system that incorporates the time value of money, costs, and the fundamental trade-off between accumulating arrears versus forsaking future interest revenue. Moreover, the procedure can be used with different delinquency measures (other than payments in arrears), thereby allowing an indirect comparison of these measures. We demonstrate the system across a range of credit risk scenarios and portfolio compositions. The computational results show that threshold optima can exist across all reasonable values of both the payment probability (default risk) and the loss rate (loan collateral). In addition, the procedure reacts positively to portfolios afflicted by either systematic defaults (such as during an economic downturn) or episodic delinquency (i.e., cycles of curing and re-defaulting). In optimising a portfolio's recovery decision, our procedure can better inform the quantitative aspects of a bank's collection policy than relying on arbitrary discretion alone.

Uncovering the mesoscale structure of the credit default swap market to improve portfolio risk modelling
Ioannis Anagnostou,Tiziano Squartini,Drona Kandhai,Diego Garlaschelli

One of the most challenging aspects in the analysis and modelling of financial markets, including Credit Default Swap (CDS) markets, is the presence of an emergent, intermediate level of structure standing in between the microscopic dynamics of individual financial entities and the macroscopic dynamics of the market as a whole. This elusive, mesoscopic level of organisation is often sought for via factor models that ultimately decompose the market according to geographic regions and economic industries. However, at a more general level the presence of mesoscopic structure might be revealed in an entirely data-driven approach, looking for a modular and possibly hierarchical organisation of the empirical correlation matrix between financial time series. The crucial ingredient in such an approach is the definition of an appropriate null model for the correlation matrix. Recent research showed that community detection techniques developed for networks become intrinsically biased when applied to correlation matrices. For this reason, a method based on Random Matrix Theory has been developed, which identifies the optimal hierarchical decomposition of the system into internally correlated and mutually anti-correlated communities. Building upon this technique, here we resolve the mesoscopic structure of the CDS market and identify groups of issuers that cannot be traced back to standard industry/region taxonomies, thereby being inaccessible to standard factor models. We use this decomposition to introduce a novel default risk model that is shown to outperform more traditional alternatives.

Wealth distribution in modern societies: collected data and a master equation approach
Istvan Gere,Szabolcs Kelemen,Geza Toth,Tamas Biro,Zoltan Neda

A mean-field like stochastic evolution equation with growth and reset terms (LGGR model) is used to model wealth distribution in modern societies. The stationary solution of the model leads to an analytical form for the density function that is successful in describing the observed data for all wealth categories. In the limit of high wealth values the proposed density function has the accepted Tsallis-Pareto shape. Our results are in agreement with the predictions of an earlier approach based on a mean-field like wealth exchange process.