Figure 1 - uploaded by Crescenzio Gallo
Content may be subject to copyright.
Source publication
The study of Artificial Neural Networks derives from first trials to translate in mathematical models the principles of biological processing. An Artificial Neural Network deals with generating, in the fastest times, an implicit and predictive model of the evolution of a system. In particular, it derives from experience its ability to be able to re...
Similar publications
The study of Artificial Neural Networks derives from first trials to translate in mathematical models the principles of biological “processing”. An Artificial Neural Network deals with generating, in the fastest times, an implicit and predictive model of the evolution of a system. In particular, it derives from experience its ability to be able to...
Citations
... Recently, artificial neural networks have been widely and successfully employed in solving classification and regression problems in finance and many other fields. One of their main advantages is that they are able to capture complex, non-linear interactions (see, e.g., [1]). This is what other traditional financial economic tools often fail to handle (see, e.g., [2,3]). ...
... With our results, we contribute to the literature on early warning systems for the stock market and to the literature on the application of neural networks in financial time series forecasts. Neural networks and deep learning models were successfully used in financial modeling for various tasks (see, e.g., [1] or [6] for an overview). The applications include the prediction of financial market movement directions (see [7]), the construction of optimal portfolios (see [8]) and trading strategies (see [9]), predicting techniques for the equity premium (see [2]) or exchange rates (see [10,11], as well as the quantification of enterprise risk (see [12]) and the examination of risk management tools (see [13]). ...
... where CCD i [1] refers to the first day in the i-th set of core crisis dates CCD i . For the end day of a crisis, we first denote t Low as the day of the lowest index value before the next 26 week high is reached, i.e., ...
Artificial neural networks have gained increasing importance in many fields, including quantitative finance, due to their ability to identify, learn and regenerate non-linear relationships between targets of investigation. We explore the potential of artificial neural networks in forecasting financial crises with micro-, macroeconomic and financial factors. In this application of neural networks, a huge amount of available input factors, but limited historical data, often leads to over-parameterized and unstable models. Therefore, we develop an input variable reduction method for model selection. With an iterative walk-forward forecasting and testing procedure, we create out-of-sample predictions for crisis periods of the S&P 500 and demonstrate that the model selected with our method outperforms a model with a set of input factors taken from the literature.
... Research during the last 20 years shows that ML approaches are more effective than classical statistical techniques. The reason is that ML deals much better with complexity and non-linearity of FTS [38], [53], [79], [96], [108], [111]. ML offers a wide variety of modelling techniques, including clustering, Support Vector Machines (SVM), Decision Trees and Artificial Neural Networks (ANN) among others. ...
... ML offers a wide variety of modelling techniques, including clustering, Support Vector Machines (SVM), Decision Trees and Artificial Neural Networks (ANN) among others. Within this set of elements, ANN are the most popular [38], [53], [79], [96], [108], [111]. ...
... Literature is vast when referring to applications of ML to FTS in general and Stock Prices in particular. It includes a variety of techniques such as Clustering, Decision Trees, Support Vector Machines (SVM) and Artificial Neural Networks (ANN) [9], [10], [29], [53], [38], [39], [42], [50], [59], [78], [79], [87], [96], [100], [105], [106], [108], [111], [115]. However, most of the literature focuses on two techniques: SVM and ANN. ...
https://repositorio.unal.edu.co/handle/unal/69861
This work presents an innovative and highly competitive Algorithmic Trading (AT) Strategy, based on a Convolutional Neural Network price direction predictor that uses High Frequency (HF) transactions and Limit Order Book (LOB) data. Information used includes data from US and Colombian market. Data processing include more than 5 million raw data files of 21 stocks from different industries (Energy, Finance, Technology, Construction, among others).
Since data include two different sources (Transaction and LOB), applying feature engineering is necessary to homogenize inputs. For transaction data, an image-like representation (Grammian Angular Field GAF) is used. It converts Financial Time Series (FTS) to polar coordinates and creates a kernel based on cosine differences. Additionally, this work proposes a transformation for LOB data. This representation includes all available information deviated from LOB raw data and it will create an image-like representation of LOB. These two sources will feed up into a proposed 3D-Convolutional Neural Network (3D-CNN) architecture that generates price direction predictions.
These predictions will serve as a trading signal generator for two Algorithmic Trading Strategies. Both of them take real market constrains into consideration, such as liquidity provision, transaction costs, among others. The two proposed strategies works under di
erent risk aversion constraints.
Results from the proposed 3D-CNN predictor present a strong performance, ranging between 70% and 74% in Directional Accuracy (DA), while reducing model parameters as well as making inputs time invariant. Moreover, trading strategies results illustrate that the proposed CNN predictor can lead to profitable trades and liquidity improvement in the Colombian Market.
Testing results for both AT strategies on Colombian Market Data lead to interesting findings. Under different constrains of take profit, stop loss and transaction cost, both strategies aggressive and conservative lead to positive returns over the same period of time. Moreover, results of number of trades performed by the aggressive AT helps to understand how AT may impact positively liquidity provision in developing financial markets.
... A growing field of research in artificial neural networks studies the interactions between economics and computer science, studying their potential applications to economics (Boguslauskas and Mileris, 2009). Artificial neural networks represent an easily customizable tool for modelling the learning behaviour of agents and for studying many problems that are very difficult to analyze with standard economic models (Gallo, 2006). ...
Enterprise Resource Planning systems have proven to be efficient and have become a de facto standard for coordinating vital business components. However, the obvious question has arisen: if each company uses the same ERP system, what happens to the competitive aspect of the business after the implementation of the IT systems?
While for some organizations, ERPs have only become a necessity for running and organizing business, others want to exploit it to exceed the performance of competitors.
Consequently, ERP systems are often a combined solution between the legacies of the systems they have replaced and the model proposed by the ERP provider, resulting in systems with unique, customized features. Keeping this idea, we aim to add to the present paper elements of artificial intelligence within a module for managing car sales within an ERP.
Hybrid classification approaches on credit domain are widely used to obtain valuable information about customer behaviours. Single classification algorithms such as neural networks, support vector machines and regression analysis have been used since years on related area. In this paper, we propose hybrid classification approaches, which try to combine several classifiers and ensemble learners to boost accuracy on classification results. We worked with two credit datasets, German dataset which is a public dataset and a Turkish Corporate Bank dataset. The goal of using such diverse datasets is to search for generalization ability of proposed model. Results show that feature selection plays a vital role on classification accuracy, hybrid approaches which shaped with ensemble learners outperform single classification techniques and hybrid approaches which consists SVM has better accuracy performance than other hybrid approaches.
This paper presents improvements in financial time series prediction using a Deep Neural Network (DNN) in conjunction with a Discrete Wavelet Transform (DWT). When comparing our model to other three alternatives, including ARIMA and other deep learning topologies, ours has a better performance. All of the experiments were conducted on High-Frequency Data (HFD). Given the fact that DWT decomposes signals in terms of frequency and time, we expect this transformation will make a better representation of the sequential behavior of high-frequency data. The input data for every experiment consists of 27 variables: The last 3 one-minute pseudo-log-returns and last 3 one-minute compressed tick-by-tick wavelet vectors, each vector is a product of compressing the tick-by-tick transactions inside a particular minute using a DWT with length 8. Furthermore, the DNN predicts the next one-minute pseudo-log-return that can be transformed into the next predicted one-minute average price. For testing purposes, we use tick-by-tick data of 19 companies in the Dow Jones Industrial Average Index (DJIA), from January 2015 to July 2017. The proposed DNN’s Directional Accuracy (DA) presents a remarkable forecasting performance ranging from 64% to 72%.
L’obiettivo di questo breve tutorial è di introdurre una potente classe di modelli matematici: le reti neurali artificiali. In realtà, questo è un termine molto generico che comprende molti diversi modelli di tipo matematico e varie tipologie di approcci. Scopo di questo tutorial non è di esaminarli tutti, ma di far comprendere le funzionalità e le possibili implementazioni di questo potente strumento. Inizialmente vengono introdotte le reti, per analogia, con il cervello umano. Viene analizzato in dettaglio un tipo di rete neurale: la rete feed-forward con algoritmo di calcolo dell’errore backpropagation. Vengono quindi presentate le architetture dei modelli, i metodi dell’apprendimento e della rappresentazione dei dati. Nella sezione finale è illustrata una serie di applicazioni ed estensioni al modello di base.
Since the times of 1800s, which the economic literature started to appear, financial crises have been a subject of research. And economists have tried to understand and explain of these crises. As a result of theoretical studies made up to now, a wide literature of theories of crises has been formed. The literature has been growing continuously with the help of new studies and with the objects of changes in world economy. As a result, it has been update and fresh.
The objective of the thesis is to develop on early warning system to predict the financial crisis in Turkish economy likely to happen by using both Kaminsky, Lizondo and Reinhart (KLR) Model and relatively new Artificial Neural Networks (ANN) Model together. In the study, a data set covering the period of January 1992 - March 2011 of Turkish economy being used, an early warning system has been developed with KLR and YSA models. Financial Pressure Index (FPI), which is a result of calculation of percentage changes in dollar exchange rate, overnight interest rate and gross foreign exchange reserves of the Central Bank is used as the dependent variable, while thirty-two macro-economic indicators are used as the independent variables. Two models which are tested in Turkish case have given clear signals anticipating 1994 and 2001 crises 12 months before.
