networks and genetic algorithm to perform pattern recognition of the bull flag pattern and to learn the trading rules from price andvolumeoftheNYSECompositeIndex.Resultsshowedthat the forecasting method yielded statistically significant returns that are better than the overall average 20-day horizon price increase. Moody et al. [23], [25] used recurrent reinforcement learning without forecasting to train a trading system to trade using past prices of S&P500 stock index while accounting for the effects of transaction costs (please refer to [26] for details onreinforcementlearning).Chenetal.[24]usedaprobabilistic neuralnetwork(PNN)toforecastthedirectionofpricemoment of the Taiwan Stock Index and presented two PNN-guided investment strategies to translate the predicted direction to trading signals. Field and Singh [20] used Pareto evolutionary neural network (Pareto-ENN) to forecast 37 different interna- tional stock indexes. Althoughneuralnetworks possessthepropertiesrequiredfor technical financial forecasting, they cannot be used to explain the causal relationship between input and output variables be- cause of theirblack box nature.Neuro-fuzzy hybridization syn- ergizes neural networks and fuzzy systems by combining the human-like reasoning style of fuzzy systems with the learning andconnectioniststructureofneuralnetworks.Neuro-fuzzyhy- bridizationiswidelytermedasfuzzyneuralnetworks(FNNs)or neuro-fuzzy systems (NFSs) in the literature [27]. NFSs incor- poratesthehuman-likereasoningstyleoffuzzysystemsthrough theuseoffuzzysetsandalinguisticmodelconsistingofasetof IF–THEN fuzzy rules. Thus the main strength of NFSs is that they are universal approximators [28]–[30] with the ability to solicit interpretable IF–THEN rules [31]. In recent years, in- creasing number of research applied NFSs in financial engi- neering[32].SomeworksthatappliedNFSsinforecastingstock price are [8], [21], [33]–[35]. This paper proposes a novel rough set-based neuro-fuzzy stock trading decision model called stock trading using rough set-based pseudo outer-product (RSPOP). Section II reviews the two main NFSs and outlines the proposed rough set-based neuro-fuzzy approach. Section III reviews the commonly used time-delayed price forecast approach and the time-delayed price difference forecast approach in forecasting stock prices. Section IV presents experimental results of forecasting stock price difference using various neuro-fuzzy systems and neural networks on artificially generated price series data. Section V reviews existing trading models with and without forecast and presents the proposed forecast bottleneck free stock trading with RSPOP forecast model. Section VI presents extensive experimental results using the proposed stock trading with RSPOP forecast model on real world stock market data. The trading profits in terms of portfolio end values are presented and compared against the stock trading with dynamic evolving neural-fuzzy inference system (DENFIS) [36] forecast model, the stock trading without forecast model and the stock trading with ideal forecast model. Finally, Section VII concludes this paper.
networks and genetic algorithm to perform pattern recognition of the bull flag pattern and to learn the trading rules from price andvolumeoftheNYSECompositeIndex.Resultsshowedthat the forecasting method yielded statistically significant returns that are better than the overall average 20-day horizon price increase. Moody et al. [23], [25] used recurrent reinforcement learning without forecasting to train a trading system to trade using past prices of S&P500 stock index while accounting for the effects of transaction costs (please refer to [26] for details onreinforcementlearning).Chenetal.[24]usedaprobabilistic neuralnetwork(PNN)toforecastthedirectionofpricemoment of the Taiwan Stock Index and presented two PNN-guided investment strategies to translate the predicted direction to trading signals. Field and Singh [20] used Pareto evolutionary neural network (Pareto-ENN) to forecast 37 different interna- tional stock indexes. Althoughneuralnetworks possessthepropertiesrequiredfor technical financial forecasting, they cannot be used to explain the causal relationship between input and output variables be- cause of theirblack box nature.Neuro-fuzzy hybridization syn- ergizes neural networks and fuzzy systems by combining the human-like reasoning style of fuzzy systems with the learning andconnectioniststructureofneuralnetworks.Neuro-fuzzyhy- bridizationiswidelytermedasfuzzyneuralnetworks(FNNs)or neuro-fuzzy systems (NFSs) in the literature [27]. NFSs incor- poratesthehuman-likereasoningstyleoffuzzysystemsthrough theuseoffuzzysetsandalinguisticmodelconsistingofasetof IF–THEN fuzzy rules. Thus the main strength of NFSs is that they are universal approximators [28]–[30] with the ability to solicit interpretable IF–THEN rules [31]. In recent years, in- creasing number of research applied NFSs in financial engi- neering[32].SomeworksthatappliedNFSsinforecastingstock price are [8], [21], [33]–[35]. This paper proposes a novel rough set-based neuro-fuzzy stock trading decision model called stock trading using rough set-based pseudo outer-product (RSPOP). Section II reviews the two main NFSs and outlines the proposed rough set-based neuro-fuzzy approach. Section III reviews the commonly used time-delayed price forecast approach and the time-delayed price difference forecast approach in forecasting stock prices. Section IV presents experimental results of forecasting stock price difference using various neuro-fuzzy systems and neural networks on artificially generated price series data. Section V reviews existing trading models with and without forecast and presents the proposed forecast bottleneck free stock trading with RSPOP forecast model. Section VI presents extensive experimental results using the proposed stock trading with RSPOP forecast model on real world stock market data. The trading profits in terms of portfolio end values are presented and compared against the stock trading with dynamic evolving neural-fuzzy inference system (DENFIS) [36] forecast model, the stock trading without forecast model and the stock trading with ideal forecast model. Finally, Section VII concludes this paper.
การแปล กรุณารอสักครู่..
