| Abstract [eng] |
A simple Stock Exchange Game Model (SEGM) was introduced in 2009 by prof. Jonas Mockus to simulate the behavior of several stockholders using fixed buying-selling margins at fixed bank yield. The new model, Extended Stock Exchange model (USEGM), is under development, which advantage is application of the Nash Equilibrium (NE) to strategies that define buying-selling margins and bank haircuts (HC) dynamically. In addition, USEGM includes the transaction costs to reflect the reality better. This enables us to simulate market illiquidity in many different scenarios. This paper describes in deep the mathematical models of making the financial prediction . To represent virtual users that prefer linear utility functions, USEGM adds Wiener process-model RW and two autoregressive models: AR(p) which optimising least square deviations and AR-ABS(p), which optimising the absolute values. The AR-ABS(p) prediction model were not used previously in SEGM model and currently is included in USEGM as an improvement. In this work the prediction methods described above will be investigated. The objective of USEGM is not forecasting, but simulation of financial time series that are affected by predictions of the virtual participants. This work is based under SEGM model in which in the time of investigation were revealed inconsistencies in predictions. In the time of explicit investigation are prooved, that all of these prediction methods - RW, AR(p) and AR-ABS(p) are suitable for new USEGM model to make a well predictions. In the time of investigation the found inconsitencies in prediction algorithms were removed and finally the methods are working correctly. |
