Research Article Open Access

The Tadpole Bayesian Model for Detecting Trend Changes in Financial Quotations

Abstract

The Tadpole Model basing on robust Bayesian regression method is introduced. The paper describes the numerical algorithm for detecting trend changes in the financial quotation or generally - in time-dependent functions. The application of Bayesian fitting algorithm makes the model insensitive to local fluctuations and finally is noise-free. The presented algorithm detects trend changes in Stock Exchange quotations, in the currency exchange rate, etc. The model can work on-line, which means it systematically receives the current value of the analyzed quotation and finds the potential critical and inflection points of the function. The model was tested on the real historical data concerned with several dozens of hourly currency exchange rate and the Warsaw Stock Exchange quotations. About 60% of the model’s trend change detections were correct.

Krzysztof Wojciech Fornalski*

To read the full article Download Full Article | Visit Full Article