Not logged in.
Quick Search - Contribution
|Title||Price Predictability in an Evolutionary Finance Model|
|Institution||University of Zurich|
|Faculty||Faculty of Economics, Business Administration and Information Technology|
|Zusammenfassung||Evolutionary finance is a relatively new field of research that applies the Darwinian concept of natural selection to financial markets. It allows to analyze price dynamics absent assumptions of rationality and utility maximization. So far, most research in evolutionary finance has been theoretical with the main finding being that prices in stable markets are driven by expected relative dividends. However, the price predictions of the theory have yet to be tested using real market data. This thesis aims to fill this gap and introduces a calibration method for an evolutionary finance model to analyze its price predictions. The first section of this thesis introduces the reader to the main components and results of the recent literature in evolutionary finance. The thesis focuses on the evolutionary finance model with a risk-free asset of Evstigneev, Hens, and Schenk-Hoppé (2011). The model and its assumptions are explained in detail in the second section. In addition, two model extensions that account for the flows of funds among strategies are discussed. In the third section, the calibration methodology for the model is derived. The approach minimizes the absolute average pricing error of the model and yields an optimal initial wealth distribution as well as a constant risk-free investment weight for a fixed set of trading strategies. The fitting procedure exploits a time-dependent reinvestment rate that is implicitly obtained from aggregate market data. In the empirical analysis, the model is calibrated to market data of 12 firms listed in the Dow Jones Industrial Average in the years 1982 to 2006. Using a small set of wellknown trading strategies, the calibration results find the Lambda Star rule which invests according to relative dividends as the main driver of long-term price dynamics. Fitting the model to the whole data set yields an average absolute pricing error of 34.41%. On an aggregate level, the wealth dynamics of the stock market are captured very well by the calibrated model until the advent of the dot-com bubble. The model implies that the bubble cannot be explained by fundamentals. The out-of-sample price predictions are tested using a rolling calibration window of 5 years. In these short-horizon windows prices are best explained by the market portfolio strategy. The price predictions are less accurate than the forecast that uses the last observed price. Overall, the thesis shows the applicability of the evolutionary finance model to explain long-term price dynamics as well as its limits to predict prices in the short run.|