Model Based Reinforcement Learning with Non-Gaussian Environment Dynamics and its Application to Portfolio Optimization (preprint)

With the fast development of quantitative portfolio optimization in financial engineering, lots of AI-based algorithmic trading strategies have demonstrated promising results, among which reinforcement learning begins to manifest competitive advantages. However, the environment from real financial markets is complex and hard to be fully simulated, considering the observation of abrupt transitions, unpredictable hidden causal factors, heavy tail properties and so on. Thus, in this paper, first, we adopt a heavy-tailed preserving normalizing flows to simulate high-dimensional joint probability of the complex trading environment and develop a model-based reinforcement learning framework to better understand the intrinsic mechanisms of quantitative online trading. Second, we experiment with various stocks from three different financial markets (Dow, NASDAQ and S&P) and show that among these three financial markets, Dow gets the best performance based on various evaluation metrics under our back-testing system. Especially, our proposed method is able to mitigate the impact of unpredictable financial market crises during the COVID-19 pandemic period, resulting in a lower maximum drawdown. Third, we also explore the explanation of our RL algorithm. (1), we utilize the pattern causality method to study the interactive relation among different stocks in the environment. (2), We analyze the dynamic loss and actor loss to ensure the convergence of our strategies. (3), by visualizing high dimensional state transition data comparisons from real and virtual buffers with t-SNE, we uncover some effective patterns of better portfolio optimization strategies. (4), we also utilize eigenvalue analysis to study the convergence properties of the environmen's model.

How does psychological resilience influence subjective career success of Internet marketers in china? A moderated mediation model

In this study, the research objective of psychological resilience refers to the emerging professional group of Internet marketers under the background of the COVID-19 pandemic environment. This paper studies the effect of the psychological resilience of Internet marketers on their subjective career success. The result shows that Internet marketers’ psychological resilience has a positive impact on their subjective career success. The work engagement of Internet marketers plays a mediating role in the relationship between psychological resilience and subjective career success. Meanwhile, Internet marketers’ workload positively moderates the mediating effects of work engagement. This study starts from the perspective of positive psychology to investigate the psychological resilience of Internet marketers and broadens the scope of application of positive organizational behavior and psychology.

Stock Trading Optimization through Model-based Reinforcement Learning with Resistance Support Relative Strength (preprint)

Reinforcement learning (RL) is gaining attention by more and more researchers in quantitative finance as the agent-environment interaction framework is aligned with decision making process in many business problems. Most of the current financial applications using RL algorithms are based on model-free method, which still faces stability and adaptivity challenges. As lots of cutting-edge model-based reinforcement learning (MBRL) algorithms mature in applications such as video games or robotics, we design a new approach that leverages resistance and support (RS) level as regularization terms for action in MBRL, to improve the algorithm's efficiency and stability. From the experiment results, we can see RS level, as a market timing technique, enhances the performance of pure MBRL models in terms of various measurements and obtains better profit gain with less riskiness. Besides, our proposed method even resists big drop (less maximum drawdown) during COVID-19 pandemic period when the financial market got unpredictable crisis. Explanations on why control of resistance and support level can boost MBRL is also investigated through numerical experiments, such as loss of actor-critic network and prediction error of the transition dynamical model. It shows that RS indicators indeed help the MBRL algorithms to converge faster at early stage and obtain smaller critic loss as training episodes increase.

GlPS1 overexpression accumulates coumarin secondary metabolites in transgenic Arabidopsis (preprint)

The dried root of Glehnia littoralis is a traditional Chinese herbal medicine mainly used to treat lung diseases and plays an important role in fighting coronavirus disease 2019 pneumonia in China. This study focused on the key enzyme gene GlPS1 for furanocoumarin synthesis in G. littoralis. In the 35S:GlPS1 transgenic Arabidopsis study, the Arabidopsis thaliana-overexpressing GlPS1 gene was more salt-tolerant than Arabidopsis in the blank group. Metabolomics analysis showed 30 differential metabolites in Arabidopsis, which overexpressed the GlPS1 gene. Twelve coumarin compounds were significantly upregulated, and six of these coumarin compounds were not detected in the blank group. Among these differential coumarin metabolites, isopimpinellin and aesculetin have been annotated by the Kyoto Encyclopedia of Genes and Genomes and isopimpinellin was not detected in the blank group. Through structural comparison, imperatorin was formed by dehydration and condensation of zanthotoxol and a molecule of isoprenol, and the difference between them was only one isoprene. Results showed that the GlPS1 gene positively regulated the synthesis of coumarin metabolites in A. thaliana and at the same time improved the salt tolerance of A. thaliana.

Autonomous inference of complex network dynamics from incomplete and noisy data (preprint)

The availability of empirical data that capture the structure and behavior of complex networked systems has been greatly increased in recent years, however a versatile computational toolbox for unveiling a complex system's nodal and interaction dynamics from data remains elusive. Here we develop a two-phase approach for autonomous inference of complex network dynamics, and its effectiveness is demonstrated by the tests of inferring neuronal, genetic, social, and coupled oscillators dynamics on various synthetic and real networks. Importantly, the approach is robust to incompleteness and noises, including low resolution, observational and dynamical noises, missing and spurious links, and dynamical heterogeneity. We apply the two-phase approach to inferring the early spreading dynamics of H1N1 flu upon the worldwide airline network, and the inferred dynamical equation can also capture the spread of SARS and COVID-19 diseases. These findings together offer an avenue to discover the hidden microscopic mechanisms of a broad array of real networked systems.

DNA damage contributes to age-associated differences in SARS-CoV-2 infection (preprint)

Coronavirus disease 2019 (COVID-19), caused by coronavirus SARS-CoV-2, is known to disproportionately affect older individuals1,2. How aging processes affect the disease progression remains largely unknown. Here we found that DNA damage, one of the major causes of aging3, promoted susceptibility to SARS-CoV-2 infection in cells and intestinal organoids. SARS-CoV-2 entry was facilitated by DNA damage caused by telomere attrition or extrinsic genotoxic stress and hampered by inhibition of DNA damage response (DDR). Mechanistic analysis revealed that DDR increased expression of ACE2, the receptor of SARS-CoV-2, by activation of transcription factor c-Jun in vitro and in vivo. Expression of ACE2 was elevated in the older tissues and positively correlated with γH2Ax and phosphorylated c-Jun (p-c-Jun). Finally, targeting DNA damage by increasing the DNA repair capacity, alleviated cell susceptibility to SARS-CoV-2. Our data provide insights into the age-associated differences in SARS-CoV-2 infection and a novel target for anti-viral intervention.

Heparin-induced thrombocytopenia is associated with a high risk of mortality in critical COVID-19 patients receiving heparin-involved treatment (preprint)

Background Coronavirus infectious disease 2019 (COVID-19) has developed into a global pandemic. It is essential to investigate the clinical characteristics of COVID-19 and uncover potential risk factors for severe disease to reduce the overall mortality rate of COVID-19. Methods Sixty-one critical COVID-19 patients admitted to the intensive care unit (ICU) and 93 severe non-ICU patients at Huoshenshan Hospital (Wuhan, China) were included in this study. Medical records, including demographic, platelet counts, heparin-involved treatments, heparin-induced thrombocytopenia-(HIT) related laboratory tests, and fatal outcomes of COVID-19 patients were analyzed and compared between survivors and nonsurvivors. Findings Sixty-one critical COVID-19 patients treated in ICU included 15 survivors and 46 nonsurvivors. Forty-one percent of them (25/61) had severe thrombocytopenia, with a platelet count (PLT) less than 50x109/L, of whom 76% (19/25) had a platelet decrease of >50% compared to baseline; 96% of these patients (24/25) had a fatal outcome. Among the 46 nonsurvivors, 52.2% (24/46) had severe thrombocytopenia, compared to 6.7% (1/15) among survivors. Moreover, continuous renal replacement therapy (CRRT) could induce a significant decrease in PLT in 81.3% of critical CRRT patients (13/16), resulting in a fatal outcome. In addition, a high level of anti-heparin-PF4 antibodies, a marker of HIT, was observed in most ICU patients. Surprisingly, HIT occurred not only in patients with heparin exposure, such as CRRT, but also in heparin-naive patients, suggesting that spontaneous HIT may occur in COVID-19. Interpretation Anti-heparin-PF4 antibodies are induced in critical COVID-19 patients, resulting in a progressive platelet decrease. Exposure to a high dose of heparin may trigger further severe thrombocytopenia with a fatal outcome. An alternative anticoagulant other than heparin should be used to treat COVID-19 patients in critical condition.

Highly pathogenic coronavirus N protein aggravates lung injury by MASP-2-mediated complement over-activation (preprint)

An excessive immune response contributes to SARS-CoV, MERS-CoV and SARS-CoV-2 pathogenesis and lethality, but the mechanism remains unclear. In this study, the N proteins of SARS-CoV, MERS-CoV and SARS-CoV-2 were found to bind to MASP-2, the key serine protease in the lectin pathway of complement activation, resulting in aberrant complement activation and aggravated inflammatory lung injury. Either blocking the N protein:MASP-2 interaction or suppressing complement activation can significantly alleviate N protein-induced complement hyper-activation and lung injury in vitro and in vivo. Complement hyper-activation was also observed in COVID-19 patients, and a promising suppressive effect was observed when the deteriorating patients were treated with anti-C5a monoclonal antibody. Complement suppression may represent a common therapeutic approach for pneumonia induced by these highly pathogenic coronaviruses.