FX Trading Using Gaussian Processes
Abstract: Machine learning and its application within finance have gained popularity the last decade. The traditional trading roles are changing rapidly and are being increasingly automated with algorithmic trading strategies, by proprietary trading firms, market makers, and other financial institutions. FX trading often involves strategies in the form of technical analysis – suggesting that the efficient market hypothesis might not always hold. Different machine learning techniques are often used in trading activities by Quant fund and other algorithmic and high-frequency trading firms. In this thesis, I investigate if the Gaussian Process Regression (GPR) can predict prices on a EUR/USD FX Future from CME Globex. The GPR approach has its advantages, being a non-parametric and probabilistic method, and often being much simpler to implement, in contrast to other machine learning techniques like neural networks, which might not always be easy to apply in practice. The last decades of developments within GPR has made it a solid competitor for real supervised learning applications. In this thesis the ARIMA model is used as a benchmark model for prediction.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)