arXiv subjects

Mikhail Semenov

Publications and source records attributed to Mikhail Semenov.

13 recordsLinked to original sources

arXiv2026

Finance-Grounded Optimization For Algorithmic Trading

Deep Learning is evolving fast and integrates into various domains. Finance is a challenging field for deep learning, especially in the case of interpretable artificial intelligence (AI). Although classical approaches perform very well with natural language processing, computer vision, and forecasting, they are not perfect for the financial world, in which specialists use different metrics to evaluate model performance. We first introduce financially grounded loss functions derived from key quantitative finance metrics, including the Sharpe ratio, Profit-and-Loss (PnL), and Maximum Draw down. Additionally, we propose turnover regularization, a method that inherently constrains the turnover of generated positions within predefined limits. Our findings demonstrate that the proposed loss functions, in conjunction with turnover regularization, outperform the traditional mean squared error loss for return prediction tasks when evaluated using algorithmic trading metrics. The study shows that financially grounded metrics enhance predictive performance in trading strategies and portfolio optimization.

cs.LG↗

arXiv2025

Deep Learning Models Meet Financial Data Modalities

Algorithmic trading relies on extracting meaningful signals from diverse financial data sources, including candlestick charts, order statistics on put and canceled orders, traded volume data, limit order books, and news flow. While deep learning has demonstrated remarkable success in processing unstructured data and has significantly advanced natural language processing, its application to structured financial data remains an ongoing challenge. This study investigates the integration of deep learning models with financial data modalities, aiming to enhance predictive performance in trading strategies and portfolio optimization. We present a novel approach to incorporating limit order book analysis into algorithmic trading by developing embedding techniques and treating sequential limit order book snapshots as distinct input channels in an image-based representation. Our methodology for processing limit order book data achieves state-of-the-art performance in high-frequency trading algorithms, underscoring the effectiveness of deep learning in financial applications.

cs.LG↗

arXiv2025

Multimodal Stock Price Prediction: A Case Study of the Russian Securities Market

Classical asset price forecasting methods primarily rely on numerical data, such as price time series, trading volumes, limit order book data, and technical analysis indicators. However, the news flow plays a significant role in price formation, making the development of multimodal approaches that combine textual and numerical data for improved prediction accuracy highly relevant. This paper addresses the problem of forecasting financial asset prices using the multimodal approach that combines candlestick time series and textual news flow data. A unique dataset was collected for the study, which includes time series for 176 Russian stocks traded on the Moscow Exchange and 79,555 financial news articles in Russian. For processing textual data, pre-trained models RuBERT and Vikhr-Qwen2.5-0.5b-Instruct (a large language model) were used, while time series and vectorized text data were processed using an LSTM recurrent neural network. The experiments compared models based on a single modality (time series only) and two modalities, as well as various methods for aggregating text vector representations. Prediction quality was estimated using two key metrics: Accuracy (direction of price movement prediction: up or down) and Mean Absolute Percentage Error (MAPE), which measures the deviation of the predicted price from the true price. The experiments showed that incorporating textual modality reduced the MAPE value by 55%. The resulting multimodal dataset holds value for the further adaptation of language models in the financial sector. Future research directions include optimizing textual modality parameters, such as the time window, sentiment, and chronological order of news messages.

q-fin.ST↗

arXiv2024

ExoMol line lists -- LXIV: Empirical rovibronic spectra of phosphorous mononitride (PN) covering the IR and UV regions

A new phosphorous mononitride (${}^{31}$P${}^{14}$N, ${}^{31}$P${}^{15}$N) line list PaiN covering infrared, visible and ultraviolet regions is presented. The PaiN line list extending to the $A\,{}^{1}Π$ -- $X\,{}^{1}Σ^{+}$ vibronic band system, replaces the previous YYLT ExoMol line list for PN. A thorough analysis of high resolution experimental spectra from the literature involving the $X\,{}^{1}Σ^{+}$ and $A\,{}^{1}Π$ states is conducted, and many perturbations to the $A\,{}^{1}Π$ energies are considered as part of a comprehensive MARVEL study. Ab initio potential energy and coupling curves from the previous work [Semenov et al., Phys. Chem. Chem. Phys., 23, 22057 (2021)] are refined by fitting their analytical representations to 1224 empirical energy levels determined using the MARVEL procedure. The PaiN line list is compared to previously observed spectra, recorded and calculated lifetimes, and previously calculated partition functions. The ab initio transition dipole moment curve for the $A$--$X$ band is scaled to match experimentally measured lifetimes. The line list is suitable for temperatures up to 5000 K and wavelengths longer than 121 nm. PaiN is available from www.exomol.com.

astro-ph.EP↗

arXiv2024

The 2024 release of the ExoMol database: molecular line lists for exoplanet and other hot atmospheres

The ExoMol database (www.exomol.com) provides molecular data for spectroscopic studies of hot atmospheres. These data are widely used to model atmospheres of exoplanets, cool stars and other astronomical objects, as well as a variety of terrestrial applications. The 2024 data release reports the current status of the database which contains recommended line lists for 91 molecules and 224 isotopologues giving a total of almost 10$^{12}$ individual transitions. New features of the database include extensive "MARVELization" of line lists to allow them to be used for high resolutions studies, extension of several line lists to ultraviolet wavelengths, provision of photodissociation cross sections and extended provision of broadening parameters. Some of the in-house data specifications have been rewritten in JSON and moved to conformity with other international standards. Data products, including specific heats, a database of lifetimes for plasma studies, and the ExoMolHR web app which allows exclusively high resolution data to be extracted, are discussed.

astro-ph.GA↗

arXiv2023

ExoMol line lists -- LIV: Empirical line lists for AlH and AlD and experimental emission spectroscopy of AlD in $A$ $^1Π$ ($v=0, 1, 2$)

New ExoMol line lists AloHa for AlH and AlD are presented improving the previous line lists WYLLoT (Yurchenko et al., MNRAS 479, 1401 (2018)). The revision is motivated by the recent experimental measurements and astrophysical findings involving the highly excited rotational states of AlH in its $A\,^{1}Π-{X}\,^{1}Σ^{+}$ system. A new high-resolution emission spectrum of ten bands from the ${A}\,^{1}Π-{X}\,^{1}Σ^{+}$ system of AlD, in the region $17300 - 32000$ cm$^{-1}$ was recorded with a Fourier transform spectrometer, which probes the predissociative $A\,^1Π$ $v=2$ state. The AlD new line positions are combined with all available experimental data on AlH and AlD to construct a comprehensive set of empirical rovibronic energies of AlH and AlD covering the $X\,^1Σ^+$ and $A\,^1Π$ electronic states using the MARVEL approach. We then refine the spectroscopic model WYLLoT to our experimentally derived energies using the nuclear-motion code Duo and use this fit to produce improved line lists for $^{27}$AlH, $^{27}$AlD and $^{26}$AlH with a better coverage of the rotationally excited states of $A\,^1Π$ in the predissociative energy region. The lifetimes of the predissociative states are estimated and are included in the line list using the new ExoMol data structure, alongside the temperature-dependent continuum contribution to the photo-absorption spectra of AlH. The new line lists are shown to reproduce the experimental spectra of both AlH and AlD well, and to describe the AlH absorption in the recently reported Proxima Cen spectrum, including the strong predissociative line broadening. The line lists are included into the ExoMol database www.exomol.com.

astro-ph.SR↗

arXiv2022

ExoMol line lists -- {XLVI}: Empirical rovibronic spectra of silicon mononitrate (SiN) covering the 6 lowest electronic states and 4 isotopologues

Silicon mononitride ($^{28}$Si$^{14}$N, $^{29}$Si$^{14}$N, $^{30}$Si$^{14}$N, $^{28}$Si$^{15}$N) line lists covering infrared, visible and ultraviolet regions are presented. The \name\ line lists produced by ExoMol include rovibronic transitions between six electronic states: \XS, \AS, \BS, \DS, \asi, \bsi. The \ai\ potential energy and coupling curves, computed at the multireference configuration interaction (MRCI/aug-cc-pVQZ) level of theory, are refined for the observed states by fitting their analytical representations to 1052 experimentally derived SiN energy levels determined from rovibronic bands belonging to the $X$--$X$, $A$--$X$ and $B$--$X$ electronic systems through the MARVEL procedure. The SiNful line lists are compared to previously observed spectra, recorded and calculated lifetimes, and previously calculated partition functions. SiNful is available via the \url{www.exomol.com} database.

astro-ph.EP↗

arXiv2021

ExoMol line lists -- XLIV. IR and UV line list for silicon monoxide (SiO)

A new silicon monoxide ($^{28}$Si$^{16}$O) line list covering infrared, visible and ultraviolet regions called SiOUVenIR is presented. This line list extends the infrared EBJT ExoMol line list by including vibronic transitions to the $A\,{}^{1}Π$ and $E\,{}^{1}Σ^{+}$ electronic states. Strong perturbations to the $A\,{}^{1}Π$ band system are accurately modelled through the treatment of 6 dark electronic states: $C\,{}^{1}Σ^{-}$, $D\,{}^{1}Δ$, $a\,{}^{3}Σ^{+}$, $b\,{}^{3}Π$, $e\,{}^{3}Σ^{-}$ and $d\,{}^{3}Δ$. Along with the $X\,{}^{1}Σ^{+}$ ground state, these 9 electronic states were used to build a comprehensive spectroscopic model of SiO using a combination of empirical and ab initio curves, including the potential energy (PE), spin-orbit (SO), electronic angular momentum (EAM) and (transition) dipole moment curves. The ab initio PE and coupling curves, computed at the multireference configuration interaction (MRCI) level of theory, were refined by fitting their analytical representations to 2617 experimentally derived SiO energy levels determined from 97 vibronic bands belonging to the $X$-$X$, $E$-$X$ and $A$-$X$ electronic systems through the MARVEL procedure. 112 observed forbidden transitions from the $C$-$X$, $D$-$X$, $e$-$X$, and $d$-$X$ bands were assigned using our predictions, and these could be fed back into the MARVEL procedure. The SiOUVenIR line list was computed using published ab initio transition dipole moments for the $E$-$X$ and $A$-$X$ bands; the line list is suitable for temperatures up to 10,000 K and for wavelengths longer than 140 nm. SiOUVenIR is available from www.exomol.com and the CDS database.

astro-ph.EP↗

arXiv2021

Rovibronic spectroscopy of PN from first principles

We report an ab initio study on the rovibronic spectroscopy of the closed-shell diatomic molecule phosphorous mononitride, PN. The study considers the nine lowest electronic states, $X\,{}^{1}Σ^{+}$, $A\,{}^{1}Π$, $C\,{}^{1}Σ^{-}$, $D\,{}^{1}Δ$, $E\,{}^{1}Σ^{-}$, $a\,{}^{3}Σ^{+}$, $b\,{}^{3}Π$, $d\,{}^{3}Δ$ and $e\,{}^{3}Σ^{-}$ using high level electronic structure theory and accurate nuclear motion calculations. The ab initio data cover 9 potential energy, 14 spin-orbit coupling, 7 electronic angular momentum coupling, 9 electric dipole moment and 8 transition dipole moment curves. The Duo nuclear motion program is used to solve the coupled nuclear motion Schrödinger equations for these nine electronic states and to simulate rovibronic absorption spectra of $^{31}$P$^{14}$N for different temperatures, which are compared to available spectroscopic studies. Lifetimes for all states are calculated and compared to previous results from the literature. The calculated lifetime of the $A\,{}^{1}Π$ state shows good agreement with an experimental value from the literature, which is an important quality indicator for the ab initio $A$-$X$ transition dipole moment.

physics.atom-ph↗

arXiv2020

The 2020 release of the ExoMol database: molecular line lists for exoplanet and other hot atmospheres

The ExoMol database (www.exomol.com) provides molecular data for spectroscopic studies of hot atmospheres. While the data is intended for studies of exoplanets and other astronomical bodies, the dataset is widely applicable. The basic form of the database is extensive line lists; these are supplemented with partition functions, state lifetimes, cooling functions, Landé g-factors, temperature-dependent cross sections, opacities, pressure broadening parameters, $k$-coefficients and dipoles. This paper presents the latest release of the database which has been expanded to consider 80 molecules and 190 isotopologues totaling over 700 billion transitions. While the spectroscopic data is concentrated at infrared and visible wavelengths, ultraviolet transitions are being increasingly considered in response to requests from observers. The core of the database comes from the ExoMol project which primarily uses theoretical methods, albeit usually fine-tuned to reproduce laboratory spectra, to generate very extensive line lists for studies of hot bodies. The data has recently been supplemented by line lists deriving from direct laboratory observations, albeit usually with the use of ab initio transition intensities. A major push in the new release is towards accurate characterisation of transition frequencies for use in high resolution studies of exoplanets and other bodies.

astro-ph.SR↗

arXiv2017

Portfolio Risk Assessment using Copula Models

In the paper, we use and investigate copulas models to represent multivariate dependence in financial time series. We propose the algorithm of risk measure computation using copula models. Using the optimal mean-$CVaR$ portfolio we compute portfolio's Profit and Loss series and corresponded risk measures curves. Value-at-risk and Conditional-Value-at-risk curves were simulated by three copula models: full Gaussian, Student's $t$ and regular vine copula. These risk curves are lower than historical values of the risk measures curve. All three models have superior prediction ability than a usual empirical method. Further directions of research are described.

q-fin.RM↗

arXiv2017

Predicted Landé $g$-factors for open shell diatomic molecules

The program {\sc Duo} (Yurchenko {\it et al.}, Computer Phys. Comms., 202 (2016) 262) provides direct solutions of the nuclear motion Schrödinger equation for the (coupled) potential energy curves of open shell diatomic molecules. Wavefunctions from {\sc Duo} are used to compute Landé $g$-factors valid for weak magnetic fields, the results are compared with the idealized predictions of both Hund's case (a) and Hund's case (b) coupling schemes. Test calculations are performed for AlO, NO, CrH and C$_2$. The computed $g_J$'s both provide a sensitive test of the underlying spectroscopic model used to represent the system and an indication of whether states of the molecule are well-represented by the either of the Hund's cases considered. The computation of Landé $g$-factors is implemented as a standard option in the latest release of {\sc Duo}.

physics.chem-ph↗

arXiv2013

Computational investigation of plastic deformation in face-centered cubic materials

A mathematical model of plastic deformation in face-centered cubic (FCC) materials based on a balance model taking into account fundamental properties of deformation defects of a crystal lattice was developed. This model is based on a system of ordinary differential equations (ODE) accounting for various mechanisms of generation and annihilation of deformation defects for different external conditions. In-house developed software, SPFCC (Slip Plasticity of Face-Centered Cubic), was employed to solve the system of ordinary differential equations. The implemented code solves efficiently the stiff ODE system and provides a user-friendly interface for investigation of various features of plastic deformation in FCC materials. Simulation of plastic deformation in the FCC metals was performed for the case of constant strain rate. The modelling results were validated by comparing experimental data and simulation results (stress-strain curves) and good agreement was obtained.

cond-mat.mtrl-sci↗