Exchange rate forecasting is an important subject in financial market. This article applies both parametric (group method of data handling, GMDH) and nonparametric (analog complexing, AC) self-organising modelling methods for exchange rate forecasting. The AC method used the data themselves to identify patterns with similar characteristics. The GMDH algorithm is used to combine the analog patterns and identify an optimum ensemble which has similar characteristics with the modelling object. The empirical results show that the combined method can well forecast exchange rate.% 汇率波动预测是金融市场的一个重要课题,本文结合GMDH算法(分组数据处理算法)和AC算法(相似体合成算法)建立模型用于预测汇率市场的波动。首先用相似体合成算法选择与当前时期有相同特征的相似体,再用分组数据处理算法将相似体进行加权组合,选择最优模式,用于预测当前时期的发展趋势。实证结果表明,此组合模型的预测效果较好。
展开▼