Algorithmic financial trading with deep convolutional neural networks: Time series to image conversion approach

Quantitative assessment of machine learning reliability and resilience

Advances in machine learning (ML) have led to applications in safety-critical domains, including security, defense, and healthcare. These ML models are confronted with dynamically changing and actively hostile conditions characteristic of real-world applications, requiring systems incorporating ML to be reliable and resilient. Many studies propose techniques to improve the robustness of ML algorithms. However, fewer consider quantitative techniques to assess changes in the reliability and resilience of these systems over time. To address this gap, this study demonstrates how to collect relevant data during the training and testing of ML suitable for the application of software reliability, with and without covariates, and resilience models and the subsequent interpretation of these analyses. The proposed approach promotes quantitative risk assessment of ML technologies, providing the ability to track and predict degradation and improvement in the ML model performance and assisting ML and system engineers with an objective approach to compare the relative effectiveness of alternative training and testing methods. The approach is illustrated in the context of an image recognition model, which is subjected to two generative adversarial attacks and then iteratively retrained to improve the system's performance. Our results indicate that software reliability models incorporating covariates characterized the misclassification discovery process more accurately than models without covariates. Moreover, the resilience model based on multiple linear regression incorporating interactions between covariates tracks and predicts degradation and recovery of performance best. Thus, software reliability and resilience models offer rigorous quantitative assurance methods for ML-enabled systems and processes.

Multi-Sensor Temporal Fusion Transformer for Stock Performance Prediction: An Adaptive Sharpe Ratio Approach

Accurate prediction of the Sharpe ratio, a key metric for risk-adjusted returns in financial markets, remains a significant challenge due to the complex and stochastic nature of stock price movements. This paper introduces a novel deep learning model, the Temporal Fusion Transformer with Adaptive Sharpe Ratio Optimization (TFT-ASRO), designed to address this challenge. The model incorporates real-time market sensor data and financial indicators as input signals, leveraging multiple data streams including price sensors, volume sensors, and market sentiment sensors to capture the complete market state. Using a comprehensive dataset of US historical stock prices and earnings data, we demonstrate that TFT-ASRO outperforms traditional methods and existing deep learning models in predicting Sharpe ratios across various time horizons. The model's multi-task learning framework, which simultaneously predicts returns and volatility, provides a more nuanced understanding of risk-adjusted performance. Furthermore, our adaptive optimization approach effectively balances the trade-off between return maximization and risk minimization, leading to more robust predictions. Empirical results show that TFT-ASRO achieves a 18% improvement in Sharpe ratio prediction accuracy compared to state-of-the-art baselines, with particularly strong performance in volatile market conditions. The model also demonstrates superior uncertainty quantification, providing reliable confidence intervals for its predictions. These findings have significant implications for portfolio management and investment strategy optimization, offering a powerful tool for financial decision-makers in the era of data-driven investing.

A bibliometric literature review of stock price forecasting: From statistical model to deep learning approach

We introduce a comprehensive analysis of several approaches used in stock price forecasting, including statistical, machine learning, and deep learning models. The advantages and limitations of these models are discussed to provide an insight into stock price forecasting. Traditional statistical methods, such as the autoregressive integrated moving average and its variants, are recognized for their efficiency, but they also have some limitations in addressing non-linear problems and providing long-term forecasts. Machine learning approaches, including algorithms such as artificial neural networks and random forests, are praised for their ability to grasp non-linear information without depending on stochastic data or economic theory. Moreover, deep learning approaches, such as convolutional neural networks and recurrent neural networks, can deal with complex patterns in stock prices. Additionally, this study further investigates hybrid models, combining various approaches to explore their strengths and counterbalance individual weaknesses, thereby enhancing predictive accuracy. By presenting a detailed review of various studies and methods, this study illuminates the direction of stock price forecasting and highlights potential approaches for further studies refining the stock price forecasting models.

Decision support system for the differentiation of schizophrenia and mood disorders using multiple deep learning models on wearable devices data

In the modern world, with so much inherent stress, mental health disorders (MHDs) are becoming more common in every country around the globe, causing a significant burden on society and patients' families. MHDs come in many forms with various severities of symptoms and differing periods of suffering, and as a result it is difficult to differentiate between them and simple to confuse them with each other. Therefore, we propose a support system that employs deep learning (DL) with wearable device data to provide physicians with an objective reference resource by which to make differential diagnoses and plan treatment. We conducted experiments on open datasets containing activity motion signal data from wearable devices to identify schizophrenia and mood disorders (bipolar and unipolar), the datasets being named Psykose and Depresjon. The results showed that, in both workflow approaches, the proposed framework performed well in comparison with the traditional machine learning (ML) and DL methods. We concluded that applying DL models using activity motion signal data from wearable devices represents a prospective objective support system for MHD differentiation with a good performance.

Event prediction within directional change framework using a CNN-LSTM model

Financial forecasting has always been an intriguing research area in the field of finance. The widely accepted approach to forecast financial data is to perform predictions using time series data. In time series analysis, sampling the financial data with a predefined frequency (e.g. hourly, daily) leads to an uneven and discontinued data flow. Directional Change is a newly proposed approach that replaces physical time within the financial data and establishes an event-driven framework. With the emergence of the machine and deep learning-based methods, researchers have utilised them in financial time series. These techniques have shown to outperform conventional approaches. This paper aims to employ the CNN-LSTM model to investigate its predictive competence within the Directional Change (DC) framework to predict DC event prices. To obtain this objective, we first create the tick bars/candles of the GBPUSD, EURUSD, USDCHF, and USDCAD tick prices from January to August 2019. Then, the DC-based summaries of the selected tick bar/candle for each currency pair will be generated and fed to the CNN-LSTM model. The CNN-LSTM network architecture incorporates the robustness of Convolutional Neural Network (CNN) in feature extraction and Long Short-Term Memory (LSTM) in predicting sequential data. The results suggest that the performance of the CNN-LSTM model improves significantly within the DC framework.

Forecasting Crude Oil Risk Using a Multivariate Multiscale Convolutional Neural Network Model

In light of the increasing level of correlation and dependence between the crude oil markets and the external influencing factors in the related financial markets, we propose a new multivariate empirical decomposition convolutional neural network model to incorporate the external influence of financial markets such as stock market and exchange market in a multiscale setting into the modeling of crude oil market risk movement. We propose a multivariate empirical model decomposition to analyze the finer details of interdependence among risk movement of different markets across different time horizons or scales. We also introduce the convolutional neural network to construct a new nonlinear ensemble algorithm to reduce the estimation bias and improve the forecasting accuracy. We used the major crude oil price data, stock market index, and the euro/United States dollar exchange rate data to evaluate the performance of the multivariate empirical model decomposition convolutional neural network model. The combination of both the multivariate empirical model decomposition and the convolutional neural network model in this paper has produced the risk forecasts with significantly improved risk forecasting accuracy.

AEI-DNET: A Novel DenseNet Model with an Autoencoder for the Stock Market Predictions Using Stock Technical Indicators

Predicting stock market prices is an important and interesting task in academic and financial research. The volatile nature of the stock market means that predicting stock market prices is a challenging task. However, recent advancements in machine learning, especially in deep learning techniques, have made it possible for researchers to use such techniques to predict future stock trends based on historical financial data, social media news, financial news, and stock technical indicators (STIs). This work focused on the prediction of closing stock prices based on using ten years of Yahoo Finance data of ten renowned stocks and STIs by using 1D DenseNet and an autoencoder. The calculated STIs were first used as the input for the autoencoder for dimensionality reduction, resulting in less correlation between the STIs. These STIs, along with the Yahoo finance data, were then fed into the 1D DenseNet. The resultant features obtained from the 1D DenseNet were then used as input for the softmax layer residing inside the 1D DenseNet framework for the prediction of closing stock prices for short-, medium-, and long-term perspectives. Based on the predicted trends of the stock prices, our model presented the user with one of three suggested signals, i.e., buy, sell, or hold. The experimental results showed that the proposed approach outperformed the state-of-the-art techniques by obtaining a minimum MAPE value of 0.41.

Fusion Models for Generalized Classification of Multi-Axial Human Movement: Validation in Sport Performance

We introduce a set of input models for fusing information from ensembles of wearable sensors supporting human performance and telemedicine. Veracity is demonstrated in action classification related to sport, specifically strikes in boxing and taekwondo. Four input models, formulated to be compatible with a broad range of classifiers, are introduced and two diverse classifiers, dynamic time warping (DTW) and convolutional neural networks (CNNs) are implemented in conjunction with the input models. Seven classification models fusing information at the input-level, output-level, and a combination of both are formulated. Action classification for 18 boxing punches and 24 taekwondo kicks demonstrate our fusion classifiers outperform the best DTW and CNN uni-axial classifiers. Furthermore, although DTW is ostensibly an ideal choice for human movements experiencing non-linear variations, our results demonstrate deep learning fusion classifiers outperform DTW. This is a novel finding given that CNNs are normally designed for multi-dimensional data and do not specifically compensate for non-linear variations within signal classes. The generalized formulation enables subject-specific movement classification in a feature-blind fashion with trivial computational expense for trained CNNs. A commercial boxing system, 'Corner', has been produced for real-world mass-market use based on this investigation providing a basis for future telemedicine translation.

Convolutional neural network analysis of recurrence plots for high resolution melting classification

BACKGROUND AND OBJECTIVE

High resolution melting (HRM) analysis is a rapid and correct method for identification of species, such as, microorganism, bacteria, yeast, virus, etc. HRM data are produced using real-time polymerase chain reaction (PCR) and unique for each species. Analysis of the HRM data is important for several applications, such as, for detection of diseases (e.g., influenza, zika virus, SARS-Cov-2 and Covid-19 diseases) in health, for identification of spoiled foods in food industry, for analysis of crime scene evidence in forensic investigation, etc. However, the characteristics of the HRM data can change due to the experimental conditions or instrumental settings. In addition, it becomes laborious and time-consuming process as the number of samples increases. Because of these reasons, the analysis and classification of the HRM data become challenging for species which have similar characteristics.

METHODS

To improve the classification accuracy of HRM data, we propose to use image (visual) representation of HRM data, which we call HRM images, that are generated using recurrence plots, and propose convolutional neural network (CNN) based models for classifying HRM images. In this study, two different types of recurrence plots are generated, which are black-white recurrence plots (BW-RP) and gray scale recurrence plots (GS-RP) and four different CNN models are proposed for classifying HRM data.

RESULTS

The classification performance of the proposed methods are evaluated based on average classification accuracy and F1 score, specificity, recall, and precision values for each yeast species. When BW-RP representation of HRM data is used as input to the CNN models, the best classification accuracy of 95.2% is obtained. The classification accuracies of CNN models for melting curve and GS-RP data representations of HRM data are 90.13% and 86.13%, respectively. The classification accuracy of support vector machines (SVM) model that take melting curve representation of HRM data is 86.53%. Moreover, when BW-RP representation of HRM data is used as input to the CNN models, the F1 score, specificity, recall and precision values are the highest for almost all of species.

CONCLUSIONS

Experimental results show that using BW-RP representation of HRM data improved the classification accuracy of HRM data and CNN models that take these images as input outperformed CNN models that take melting curve and GS-RP representations of HRM data as inputs and SVM model that take melting curve representation of HRM data as input.

Using Convolutional Neural Network and Candlestick Representation to Predict Sports Match Outcomes

The interdisciplinary nature of sports and the presence of various systemic and non-systemic factors introduce challenges in predicting sports match outcomes using a single disciplinary approach. In contrast to previous studies that use sports performance metrics and statistical models, this study is the first to apply a deep learning approach in financial time series modeling to predict sports match outcomes. The proposed approach has two main components: a convolutional neural network (CNN) classifier for implicit pattern recognition and a logistic regression model for match outcome judgment. First, the raw data used in the prediction are derived from the betting market odds and actual scores of each game, which are transformed into sports candlesticks. Second, CNN is used to classify the candlesticks time series on a graphical basis. To this end, the original 1D time series are encoded into 2D matrix images using Gramian angular field and are then fed into the CNN classifier. In this way, the winning probability of each matchup team can be derived based on historically implied behavioral patterns. Third, to further consider the differences between strong and weak teams, the CNN classifier adjusts the probability of winning the match by using the logistic regression model and then makes a final judgment regarding the match outcome. We empirically test this approach using 18,944 National Football League game data spanning 32 years and find that using the individual historical data of each team in the CNN classifier for pattern recognition is better than using the data of all teams. The CNN in conjunction with the logistic regression judgment model outperforms the CNN in conjunction with SVM, Naïve Bayes, Adaboost, J48, and random forest, and its accuracy surpasses that of betting market prediction.

Automated Inorganic Pigment Classification in Plastic Material Using Terahertz Spectroscopy

This paper presents an automatic classification of plastic material's inorganic pigment using terahertz spectroscopy and convolutional neural networks (CNN). The plastic materials were placed between the THz transmitter and receiver, and the acquired THz signals were classified using a supervised learning approach. A THz frequency band between 0.1-1.2 THz produced a one-dimensional (1D) vector that is almost impossible to classify directly using supervised learning. This paper proposes a novel pre-processing of 1D THz data that transforms 1D data into 2D data, which are processed efficiently using a convolutional neural network. The proposed pre-processing algorithm consists of four steps: peak detection, envelope extraction, and a down-sampling procedure. The last main step introduces the windowing with spectrum dilatation that reorders 1D data into 2D data that can be considered as an image. The spectrum dilation techniques ensure the classifier's robustness by suppressing measurement bias, reducing the complexity of the THz dataset with negligible loss of accuracy, and speeding up the network classification. The experimental results showed that the proposed approach achieved high accuracy using a CNN classifier, and outperforms 1D classification of THz data using support vector machine, naive Bayes, and other popular classification algorithms.

Construction and Simulation of Financial Audit Model Based on Convolutional Neural Network

Big data has brought a new round of information revolution. Faced with the goal of full coverage of audit and supervision, making full use of big data is the main method to promote the realization of the goal of full coverage of audit and supervision. Data analysis and utilization is an indispensable task of auditing. Actively exploring multidimensional and intelligent data analysis methods and developing big data audit cases are the new development direction of auditing. The convolutional neural network's excellent ability to extract data features well meets the relevant requirements of financial auditing. However, in practical applications, convolutional neural networks often encounter various problems such as disappearance of gradients and difficulty in convergence, which reduces its expected performance in financial audit applications. In order to make the performance of the financial audit model based on convolutional neural network more excellent, after summarizing the characteristics of genetic algorithm, this article applies genetic algorithm to the optimization of the convolutional neural network model. We applied genetic algorithm to optimize the initial weights of the convolutional neural network. The error sensitivity and learning rate changes of different hidden layers are discussed, the influence of different learning rates on the convergence speed of convolutional neural networks is analyzed, and the recognition performance of other algorithms on financial audit data sets is simulated and compared. We conducted experiments on the network structure and parameter optimization on the financial audit database. The results show that the recognition error rate of the convolutional neural network model with improved learning rate algorithm in the financial audit data set is lower than that of the multilayer feedforward network, so it has better performance.

Stock Trend Prediction Using Deep Learning Approach on Technical Indicator and Industrial Specific Information

A stock trend prediction has been in the spotlight from the past to the present. Fortunately, there is an enormous amount of information available nowadays. There were prior attempts that have tried to forecast the trend using textual information; however, it can be further improved since they relied on fixed word embedding, and it depends on the sentiment of the whole market. In this paper, we propose a deep learning model to predict the Thailand Futures Exchange (TFEX) with the ability to analyze both numerical and textual information. We have used Thai economic news headlines from various online sources. To obtain better news sentiment, we have divided the headlines into industry-specific indexes (also called “sectors”) to reflect the movement of securities of the same fundamental. The proposed method consists of Long Short-Term Memory Network (LSTM) and Bidirectional Encoder Representations from Transformers (BERT) architectures to predict daily stock market activity. We have evaluated model performance by considering predictive accuracy and the returns obtained from the simulation of buying and selling. The experimental results demonstrate that enhancing both numerical and textual information of each sector can improve prediction performance and outperform all baselines.

Machine Learning-Based Algorithms to Knowledge Extraction from Time Series Data: A Review

To predict the future behavior of a system, we can exploit the information collected in the past, trying to identify recurring structures in what happened to predict what could happen, if the same structures repeat themselves in the future as well. A time series represents a time sequence of numerical values observed in the past at a measurable variable. The values are sampled at equidistant time intervals, according to an appropriate granular frequency, such as the day, week, or month, and measured according to physical units of measurement. In machine learning-based algorithms, the information underlying the knowledge is extracted from the data themselves, which are explored and analyzed in search of recurring patterns or to discover hidden causal associations or relationships. The prediction model extracts knowledge through an inductive process: the input is the data and, possibly, a first example of the expected output, the machine will then learn the algorithm to follow to obtain the same result. This paper reviews the most recent work that has used machine learning-based techniques to extract knowledge from time series data.

Two robust long short-term memory frameworks for trading stocks

This paper aims to find a superior strategy for the daily trading on a portfolio of stocks for which traditional trading strategies perform poorly due to the low frequency of new information. The experimental work is divided into a set of traditional trading strategies and a set of long short-term memory networks. The networks incorporate general and specific trading patterns, where the former takes into account the universal decision factors for trading across many stocks, while the latter takes into account stock-specific decision factors. Our research shows that both long short-term memory networks, regardless of whether they are based on universal or stock-specific decision factors, significantly outperform traditional trading strategies. Interestingly, however, on average neither has the edge compared to the other, thus remaining ambivalent as to whether universality or specificality is to be preferred when it comes to designing long short-term memory networks for optimal trading.

Toward Smart Lockdown: A Novel Approach for COVID-19 Hotspots Prediction Using a Deep Hybrid Neural Network

COVID-19 caused the largest economic recession in the history by placing more than one third of world’s population in lockdown. The prolonged restrictions on economic and business activities caused huge economic turmoil that significantly affected the financial markets. To ease the growing pressure on the economy, scientists proposed intermittent lockdowns commonly known as “smart lockdowns”. Under smart lockdown, areas that contain infected clusters of population, namely hotspots, are placed on lockdown, while economic activities are allowed to operate in un-infected areas. In this study, we proposed a novel deep learning prediction framework for the accurate prediction of hotpots. We exploit the benefits of two deep learning models, i.e., Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) and propose a hybrid framework that has the ability to extract multi time-scale features from convolutional layers of CNN. The multi time-scale features are then concatenated and provide as input to 2-layers LSTM model. The LSTM model identifies short, medium and long-term dependencies by learning the representation of time-series data. We perform a series of experiments and compare the proposed framework with other state-of-the-art statistical and machine learning based prediction models. From the experimental results, we demonstrate that the proposed framework beats other existing methods with a clear margin.

Prediction of Stock Performance Using Deep Neural Networks

Stock performance prediction is one of the most challenging issues in time series data analysis. Machine learning models have been widely used to predict financial time series during the past decades. Even though automatic trading systems that use Artificial Intelligence (AI) have become a commonplace topic, there are few examples that successfully leverage the proven method invented by human stock traders to build automatic trading systems. This study proposes to build an automatic trading system by integrating AI and the proven method invented by human stock traders. In this study, firstly, the knowledge and experience of the successful stock traders are extracted from their related publications. After that, a Long Short-Term Memory-based deep neural network is developed to use the human stock traders’ knowledge in the automatic trading system. In this study, four different strategies are developed for the stock performance prediction and feature selection is performed to achieve the best performance in the classification of good performance stocks. Finally, the proposed deep neural network is trained and evaluated based on the historic data of the Japanese stock market. Experimental results indicate that the proposed ranking-based stock classification considering historical volatility strategy has the best performance in the developed four strategies. This method can achieve about a 20% earning rate per year over the basis of all stocks and has a lower risk than the basis. Comparison experiments also show that the proposed method outperforms conventional methods.

Optimization Parameters of Trading System with Constant Modulus of Unit Return

The unit return is determined as the return in the quotation currency (QCR) per the unit of base exchange medium (BEM). The main purpose is to examine the applicability of a trading system with a constant modulus of unit return (CMUR). The CMUR system supports speculative operations related to the exchange rate, given as the BEM quotation per the QCR. Premises for investment decisions are based on knowledge about the quotation dynamics described by its binary representation. This knowledge is described by a prediction table containing the conditional probability distributions of exchange rate increments. Any prediction table depends on observation range. Financial effectiveness of any CMUR system is assessed in the usual way by interest rate and risk index based on Shannon entropy. The main aim of our paper is to present algorithms which may be used for selecting effective CMUR systems. Required unit return modulus and observation range are control parameters applied for management of CMUR systems. Optimal values of these parameters are obtained by implementation of the proposed algorithm. All formal considerations are illustrated by an extensive case study linked to gold trading.