یادگیری عمیق در بازارهای بورس
Abstract
1 Introduction
2- Case study
3- Applied deep learning architectures
4- Methodology
5- Results
6- Conclusions
Appendix
References
Abstract
We present the implementation of a short-term forecasting system of price movements in exchange markets using market depth data and a systematic procedure to enable a fully automated trading system. Three types of Deep Learning (DL) Neural Network (NN) methodologies are trained and tested: Deep NN Classifier (DNNC), Long Short-Term Memory (LSTM) and Convolutional NN (CNN). Although the LSTM is more suitable for multivariate time series analysis from a theoretical point of view, test results indicate that the CNN has on average the best predictive power in the case study under analysis, which is the UK to Win Horse Racing market during pre-live stage in the world’s most relevant betting exchange. Implications from the generalized use of automated trading systems in betting exchange markets are discussed.
Introduction
The increasing amount of data reveals that the Big Data era is here to stay and constitutes a new form of strategic behavior and business interaction. Data is currently considered one of the most valuable intangible assets in the world. The domain of data analytical techniques is a key step not only to facilitate the transformation and growth of firms but also to boost the level of digital literacy. Goodfellow et al. (2016) recognize that the use of Deep Learning (DL) constitutes an enabler of disruptive change for businesses due to its power of association, regression, classification and clustering. Machine learning incorporates a vast array of algorithmic implementations, which not all of them can be classified as DL. Indeed, the later only corresponds to a subset of the former field of research.
Historically emerging from cognitive and information theories, DL aims at imitating the learning process of human neurons and creates complex interconnected neuronal structures sim-ilar to human synapses. Hence, DL consists of the application of multi-neuron, multi-layer Neural Networks (NN) to perform learning tasks such as regression, classification, clustering or encoding/decoding. The ability for a NN to be used in a wide variety of data and learn indiscriminately implies that the DL approach can be applied to a considerable number of case studies rather than requiring the development of a structure for each new analysis. Varian (2014) recognizes the relevance of DL NN architectures for the economics field. Proficiency with data mining, data visualization tools and artificial intelligence rank as one of the most important skills in determining business success, thus, any effort to educate stakeholders is clearly advised.