Short-term electricity price forecasting is now crucial practice in deregulated electricity markets, as it forms the basis for maximizing the profits of the market participants. In this thesis, short-term electricity prices are forecast using three different predictor schemes, Artificial Neural Networks (ANNs), Support Vector Machine (SVM) and a hybrid scheme, respectively. ANNs are the very popular and successful tools for practical forecasting. In this thesis, a hidden-layered feed-forward neural network with back-propagation has been adopted for detailed comparison with other forecasting models. SVM is a newly developed technique that has many attractive features and good performance in terms of prediction. In order to overcome the limitations of individual forecasting models, a hybrid technique that combines Fuzzy-C-Means (FCM) clustering and SVM regression algorithms is proposed to forecast the half-hour electricity prices in the UK electricity markets. According to the value of their power prices, thousands of the training data are classified by the unsupervised learning method of FCM clustering. SVM regression model is then applied to each cluster by taking advantage of the aggregated data information, which reduces the noise for each training program. In order to demonstrate the predictive capability of the proposed model, ANNs and SVM models are presented and compared with the hybrid technique based on the same training and testing data sets in the case studies by using real electricity market data. The data was obtained upon request from APX Power UK for the year 2007. Mean Absolute Percentage Error (MAPE) is used to analyze the forecasting errors of different models and the results presented clearly show that the proposed hybrid technique considerably improves the electricity price forecasting.

Uma condição fundamental para que um mercado seja competitivo é que existam muitos compradores e, em especial, compradores que possam responder aos preços. Os consumidores reagem para se ajustarem aos preços de acordo com sua disposição em consumir um determinado bem. À medida que o preço se eleva, os consumidores tendem a reduzir a quantidade demandada e, quando o preço cai, os consumidores tendem a aumentar o volume demandado. A sensibilidade dos consumidores às mudanças de preços é caracterizada pela elasticidade-preço da demanda. Contudo, nos desenhos de mercados de energia elétrica, é comum a concentração de atenção no lado do suprimento, assumindo-se, implicitamente, que toda a demanda é inelástica. O presente trabalho contempla uma análise dos mecanismos de formação de preços de curto prazo adotados em mercados de energia elétrica (i.e. formação baseada em custos e formação baseada em ofertas) e, a partir desse contexto, avalia os benefícios da introdução de mecanismos de incentivo à participação da demanda na determinação dos preços do mercado de curto prazo como forma de elevar sua eficiência econômica. / Given an economic environmental, a fundamental condition for a market be suitable to competition is that must has a plenty of buyers and, in special, those who can react to price signals. The consumers reaction aims at to adjust their energy requirements to the prices according to their disposal to access a certain product or service. As the price increases, the consumers tend to reduce the demanded volume and, on the other hand, when the prices decreases, the consumers increase the demanded volume. The consumers´ reaction to the price changes is characterized by the price elasticity of demand. However, in the electric energy market design, it is common to pay attention to the supply side, taking into account, implicitly, that all demand is inelastic. This work performs an analysis of mechanisms of spot price formation adopted by electric energy markets (i.e. cost based and bid based prices) and, from this context, evaluates the benefits of incentive mechanisms to the demand participation in determining short-term market price as an option to improve the economic efficiency.

Uma condição fundamental para que um mercado seja competitivo é que existam muitos compradores e, em especial, compradores que possam responder aos preços. Os consumidores reagem para se ajustarem aos preços de acordo com sua disposição em consumir um determinado bem. À medida que o preço se eleva, os consumidores tendem a reduzir a quantidade demandada e, quando o preço cai, os consumidores tendem a aumentar o volume demandado. A sensibilidade dos consumidores às mudanças de preços é caracterizada pela elasticidade-preço da demanda. Contudo, nos desenhos de mercados de energia elétrica, é comum a concentração de atenção no lado do suprimento, assumindo-se, implicitamente, que toda a demanda é inelástica. O presente trabalho contempla uma análise dos mecanismos de formação de preços de curto prazo adotados em mercados de energia elétrica (i.e. formação baseada em custos e formação baseada em ofertas) e, a partir desse contexto, avalia os benefícios da introdução de mecanismos de incentivo à participação da demanda na determinação dos preços do mercado de curto prazo como forma de elevar sua eficiência econômica. / Given an economic environmental, a fundamental condition for a market be suitable to competition is that must has a plenty of buyers and, in special, those who can react to price signals. The consumers reaction aims at to adjust their energy requirements to the prices according to their disposal to access a certain product or service. As the price increases, the consumers tend to reduce the demanded volume and, on the other hand, when the prices decreases, the consumers increase the demanded volume. The consumers´ reaction to the price changes is characterized by the price elasticity of demand. However, in the electric energy market design, it is common to pay attention to the supply side, taking into account, implicitly, that all demand is inelastic. This work performs an analysis of mechanisms of spot price formation adopted by electric energy markets (i.e. cost based and bid based prices) and, from this context, evaluates the benefits of incentive mechanisms to the demand participation in determining short-term market price as an option to improve the economic efficiency.

В данной выпускной квалификационной работе исследованы и применены модели глубокого машинного обучения для анализа и прогнозирования краткосрочных ценовых движений на финансовых рынках. Основной целью работы является изучение эффективности использования глубоких нейронных сетей, таких как сверточные нейронные сети (CNN) и рекуррентные нейронные сети (RNN), для прогнозирования ценовых движений. Исследование основано на данных, полученных с API Tinkoff, включающих 7 269 изображений временных рядов и файлов CSV, разделенных на три класса: buy, sell и neutral. В работе были рассмотрены три модели: CNN с механизмом внимания, CNN с двумя путями и RNN с использованием GRU. Новизна исследования заключается в использовании моделей, обрабатывающих временные ряды как изображения, что является новаторским подходом и открывает новые перспективы для повышения точности и скорости прогнозов. Результаты показали, что модели глубокого машинного обучения могут эффективно анализировать и прогнозировать краткосрочные ценовые движения. Модель RNN с использованием GRU продемонстрировала наилучшую точность (94.67%) и низкие потери (0.13). Модель CNN с двумя путями также показала хорошие результаты с точностью 82.67% и потерями 0.72. Модель CNN с механизмом внимания, несмотря на более умеренные результаты, обладает потенциалом для дальнейшего улучшения благодаря способности фокусироваться на наиболее значимых частях данных. Применение глубоких нейронных сетей может значительно улучшить точность и оперативность торговых стратегий. Разработанные модели могут быть использованы трейдерами и финансовыми аналитиками для повышения эффективности принятия решений на высоковолатильных рынках. / This master's thesis investigates and applies deep machine learning models for analyzing and predicting short-term price movements in financial markets. The main goal of the work is to study the effectiveness of using deep neural networks, such as convolutional neural networks (CNN) and recurrent neural networks (RNN), for predicting price movements. The study is based on data obtained from the Tinkoff API, including 7,269 time series images and CSV files divided into three classes: buy, sell, and neutral. The work considered three models: CNN with attention mechanism, CNN with dual paths, and RNN with GRU. The novelty of the research lies in the use of models that process time series as images, which is an innovative approach and opens new prospects for improving the accuracy and speed of forecasts. The results showed that deep machine learning models can effectively analyze and predict short-term price movements. The RNN model with GRU demonstrated the highest accuracy (94.67%) and low losses (0.13). The CNN model with dual paths also showed good results with an accuracy of 82.67% and losses of 0.72. The CNN model with attention mechanism, despite more moderate results, has the potential for further improvement due to its ability to focus on the most significant parts of the data. The application of deep neural networks can significantly improve the accuracy and timeliness of trading strategies. The developed models can be used by traders and financial analysts to enhance decision-making efficiency in highly volatile markets.