Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611700
S. A. Pasha, P. Leong
Recently the availability of tick data is driving renewed interest in statistical tools for the analysis of high-dimensional irregularly spaced time series. Since the standard tools require that the data are evenly spaced, the traditional multivariate time series analysis techniques are inadequate for the analysis of tick data. We develop for perhaps the first time a proper procedure that performs cluster analysis of tick data using the joint information of the temporal process and the continuous-valued data at the actual sampling times. A simulation example studies the problem with the standard approach and demonstrates the reliability of our proposed method. Data analyses of major stock market indices and currencies are provided.
{"title":"Cluster analysis of high-dimensional high-frequency financial time series","authors":"S. A. Pasha, P. Leong","doi":"10.1109/CIFEr.2013.6611700","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611700","url":null,"abstract":"Recently the availability of tick data is driving renewed interest in statistical tools for the analysis of high-dimensional irregularly spaced time series. Since the standard tools require that the data are evenly spaced, the traditional multivariate time series analysis techniques are inadequate for the analysis of tick data. We develop for perhaps the first time a proper procedure that performs cluster analysis of tick data using the joint information of the temporal process and the continuous-valued data at the actual sampling times. A simulation example studies the problem with the standard approach and demonstrates the reliability of our proposed method. Data analyses of major stock market indices and currencies are provided.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132336899","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611694
F. Bahramy, S. Crone
Support Vector Regression (SVR) algorithms have received increasing interest in forecasting, promising nonlinear, non-parametric and data driven regression capabilities for time series prediction. But despite evidence on the nonlinear properties of foreign exchange markets, applications of SVR in price or return forecasting have demonstrated only mixed results. However, prior studies were limited to using only autoregressive time series inputs to SVR. This paper evaluates the efficacy of SVR to predict the Euro-US Dollar exchange rate using input vectors enhanced with explanatory variables on mean-reversion movements derived from Bollinger Bands technical indicators. Using a rigorous empirical out-of-sample evaluation of multiple rolling forecast origins, we assess the accuracy of different SVR input vectors, including upper and lower BB, binary trading signals of BB, and combinations of the above. As a result, a local SVR model using autoregressive lags in conjunction with BB bands and BB indicators, and recalibrated yearly, outperforms the random walk on directional and all other error metrics, showing some promise for an SVR application.
{"title":"Forecasting foreign exchange rates using Support Vector Regression","authors":"F. Bahramy, S. Crone","doi":"10.1109/CIFEr.2013.6611694","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611694","url":null,"abstract":"Support Vector Regression (SVR) algorithms have received increasing interest in forecasting, promising nonlinear, non-parametric and data driven regression capabilities for time series prediction. But despite evidence on the nonlinear properties of foreign exchange markets, applications of SVR in price or return forecasting have demonstrated only mixed results. However, prior studies were limited to using only autoregressive time series inputs to SVR. This paper evaluates the efficacy of SVR to predict the Euro-US Dollar exchange rate using input vectors enhanced with explanatory variables on mean-reversion movements derived from Bollinger Bands technical indicators. Using a rigorous empirical out-of-sample evaluation of multiple rolling forecast origins, we assess the accuracy of different SVR input vectors, including upper and lower BB, binary trading signals of BB, and combinations of the above. As a result, a local SVR model using autoregressive lags in conjunction with BB bands and BB indicators, and recalibrated yearly, outperforms the random walk on directional and all other error metrics, showing some promise for an SVR application.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131602460","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611693
A. Raudys, Esther Mohr, G. Schmidt
In this paper we analyze US stock market after-hours trading. This is a trading outside the regular trading hours of 09:30-16:00. During this time the market is thinly traded and the possibility of price (in)efficiency arises. Price spikes up or down sometimes reaching several percent can be observed. This pattern can be exploited by a simple automated trading strategy that buys low if market drops and closes the position high on the next day when the market reopens. An empirical study using the most liquid stocks and exchange traded funds listed in NASDAQ and NYSE exchanges for the years 2000 to 2012 is conducted. We create a portfolio of ~400 automated trading strategies. The average portfolio performance is a 23 percent per annum with a Sharpe ratio of 4. This shows that prices are inefficient during after-hours trading in the US stock market. To test for significance we run an out-of-sample test from 2012 onwards.
{"title":"How (in)efficient is after-hours trading?","authors":"A. Raudys, Esther Mohr, G. Schmidt","doi":"10.1109/CIFEr.2013.6611693","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611693","url":null,"abstract":"In this paper we analyze US stock market after-hours trading. This is a trading outside the regular trading hours of 09:30-16:00. During this time the market is thinly traded and the possibility of price (in)efficiency arises. Price spikes up or down sometimes reaching several percent can be observed. This pattern can be exploited by a simple automated trading strategy that buys low if market drops and closes the position high on the next day when the market reopens. An empirical study using the most liquid stocks and exchange traded funds listed in NASDAQ and NYSE exchanges for the years 2000 to 2012 is conducted. We create a portfolio of ~400 automated trading strategies. The average portfolio performance is a 23 percent per annum with a Sharpe ratio of 4. This shows that prices are inefficient during after-hours trading in the US stock market. To test for significance we run an out-of-sample test from 2012 onwards.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125691774","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611689
T. Mizuta, K. Izumi, S. Yoshimura
Financial exchanges sometimes employ a “price variation limit”, which restrict trades out of certain price ranges within certain time spans to avoid sudden large price fluctuations. We built an artificial market model implementing a learning process to replicate bubbles that has the continues double auction mechanism and investigated price variation limits. We surveyed an adequate limitation price range and an adequate limitation time span for the price variation limit and found a parameters' condition of the price variation limit to prevent bubbles. The price variation limits are expected to be an especially effective way to prevent bubbles, so the model should be able to replicate bubbles. When we gave a bubble-inducing trigger, which is a rapid increment of the fundamental value, a bubble occurred in the case in which the model implemented the learning process and did not occur in the case without the process. We also showed that a hazard rate enables verification of whether the models can replicate a bubble process or not.
{"title":"Price variation limits and financial market bubbles: Artificial market simulations with agents' learning process","authors":"T. Mizuta, K. Izumi, S. Yoshimura","doi":"10.1109/CIFEr.2013.6611689","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611689","url":null,"abstract":"Financial exchanges sometimes employ a “price variation limit”, which restrict trades out of certain price ranges within certain time spans to avoid sudden large price fluctuations. We built an artificial market model implementing a learning process to replicate bubbles that has the continues double auction mechanism and investigated price variation limits. We surveyed an adequate limitation price range and an adequate limitation time span for the price variation limit and found a parameters' condition of the price variation limit to prevent bubbles. The price variation limits are expected to be an especially effective way to prevent bubbles, so the model should be able to replicate bubbles. When we gave a bubble-inducing trigger, which is a rapid increment of the fundamental value, a bubble occurred in the case in which the model implemented the learning process and did not occur in the case without the process. We also showed that a hazard rate enables verification of whether the models can replicate a bubble process or not.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130901341","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611704
P. Maguire, Philippe Moser, J. McDonnell, R. Kelly, Simon Fuller, R. Maguire
Existing risk-to-reward measures, such as the Sharpe ratio [1] or M2 [2], are based on the idea of quantifying the excess return per unit of deviation in an investment. In this preliminary article we introduce a new probabilistic measure for evaluating investment performance. Randomness Deficiency Coefficient (RDC) expresses the likelihood that the observed excess return of an investment has been generated by chance. Some of the advantages of RDC over existing measures are that it can be used with small historical datasets, is time-frame independent, and can be easily adjusted to take into account the familywise error rate which results from selection bias. We argue that RDC captures the fundamental relationship between risk and reward and prove that it converges with Sharpe's ratio.
{"title":"A probabilistic risk-to-reward measure for evaluating the performance of financial securities","authors":"P. Maguire, Philippe Moser, J. McDonnell, R. Kelly, Simon Fuller, R. Maguire","doi":"10.1109/CIFEr.2013.6611704","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611704","url":null,"abstract":"Existing risk-to-reward measures, such as the Sharpe ratio [1] or M2 [2], are based on the idea of quantifying the excess return per unit of deviation in an investment. In this preliminary article we introduce a new probabilistic measure for evaluating investment performance. Randomness Deficiency Coefficient (RDC) expresses the likelihood that the observed excess return of an investment has been generated by chance. Some of the advantages of RDC over existing measures are that it can be used with small historical datasets, is time-frame independent, and can be easily adjusted to take into account the familywise error rate which results from selection bias. We argue that RDC captures the fundamental relationship between risk and reward and prove that it converges with Sharpe's ratio.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133035197","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611703
William W. Y. Hsu, Cheng-Yu Lu, M. Kao, Jan-Ming Ho
This paper presents an efficient algorithm for finding the optimal k-cuts of a nondecreasing array of size n that produces the maximum area under the points. The naïve approach uses a dynamic programming algorithm which requires O(kn2) time, where n is the size of the array. This algorithm is time consuming for large n or k and thus inappropriate. We design faster algorithms by discovering and proving some nice properties of the nondecreasing arrays, finding convex hull, and by continuous-to-discrete transformation. We believe that an O(kn) time algorithm exists and show a heuristic algorithm.
{"title":"Optimum quantizing of monotonic nondecreasing arrays","authors":"William W. Y. Hsu, Cheng-Yu Lu, M. Kao, Jan-Ming Ho","doi":"10.1109/CIFEr.2013.6611703","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611703","url":null,"abstract":"This paper presents an efficient algorithm for finding the optimal k-cuts of a nondecreasing array of size n that produces the maximum area under the points. The naïve approach uses a dynamic programming algorithm which requires O(kn2) time, where n is the size of the array. This algorithm is time consuming for large n or k and thus inappropriate. We design faster algorithms by discovering and proving some nice properties of the nondecreasing arrays, finding convex hull, and by continuous-to-discrete transformation. We believe that an O(kn) time algorithm exists and show a heuristic algorithm.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125109418","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611708
M. Manuguerra, G. Sofronov, M. Tani, G. Heller
Spatio-temporal regression models are well developed in disciplines such as, for example, climate and geostatistics, but have had little application in the modelling of economic phenomena. In this study we have modelled migrations of skilled workers and firms across the European Union during the period 1998-2010. The data set has been extracted from Eurostats Labour Force Survey (LFS) and contains information stratified by European region. We investigate whether the spatial component in the migration patterns is based either on neighbourhood or on some other metric (such as the existence of a flight connection). The complete spatio-temporal model has been implemented using conditional autoregressive (CAR) random effects in the Bayesian framework. In recent years, Bayesian methods have been widely applied to spatio-temporal modelling since they enable the use of Markov chain Monte Carlo (MCMC) samplers to estimate model parameters. In this paper, we consider the Bayesian Adaptive Independence Sampler (BAIS) for estimation, and compare different computing schemes. The results suggest that the regions with a stronger increase of skilled workers are more likely to have similarities with other advanced regions which they are connected to by flight connections, than with the regions at their border. The conclusion of this study is that graphical proximity is not a sufficient condition to reduce differences in skill endowments between regions.
{"title":"Monte Carlo methods in spatio-temporal regression modeling of migration in the EU","authors":"M. Manuguerra, G. Sofronov, M. Tani, G. Heller","doi":"10.1109/CIFEr.2013.6611708","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611708","url":null,"abstract":"Spatio-temporal regression models are well developed in disciplines such as, for example, climate and geostatistics, but have had little application in the modelling of economic phenomena. In this study we have modelled migrations of skilled workers and firms across the European Union during the period 1998-2010. The data set has been extracted from Eurostats Labour Force Survey (LFS) and contains information stratified by European region. We investigate whether the spatial component in the migration patterns is based either on neighbourhood or on some other metric (such as the existence of a flight connection). The complete spatio-temporal model has been implemented using conditional autoregressive (CAR) random effects in the Bayesian framework. In recent years, Bayesian methods have been widely applied to spatio-temporal modelling since they enable the use of Markov chain Monte Carlo (MCMC) samplers to estimate model parameters. In this paper, we consider the Bayesian Adaptive Independence Sampler (BAIS) for estimation, and compare different computing schemes. The results suggest that the regions with a stronger increase of skilled workers are more likely to have similarities with other advanced regions which they are connected to by flight connections, than with the regions at their border. The conclusion of this study is that graphical proximity is not a sufficient condition to reduce differences in skill endowments between regions.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130687476","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611709
Steve Y. Yang, R. Cogill
Graph similarity measurement has been used in many applications, such as computational biology, text mining, pattern recognition, and computer vision. In this paper, we apply similarity measurement on graphs to measure structural differences in financial statements. Unconventional financial statement structures may potentially reveal deceptive intention of hiding certain information while making technically “correct” financial statements. Furthermore, unconventional financial statements may also lead to investment opportunities if legitimacy is not questioned. We construct an algorithm based on the metric of string edit distance as an approximation of graph similarity, and apply the Levenshtein algorithm with modified string edit costs to measure string edit distance. We demonstrate the effectiveness of this algorithm in capturing the sensitive changes of balance sheet structures by applying the algorithm in two experiments. The first experiment shows the algorithm is sensitive to all three basic edits (namely deletion, insertion and substitution) on a particular balance sheet, and the second experiment shows more than 90% clustering accuracy on real balance sheets.
{"title":"Balance sheet outlier detection using a graph similarity algorithm","authors":"Steve Y. Yang, R. Cogill","doi":"10.1109/CIFEr.2013.6611709","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611709","url":null,"abstract":"Graph similarity measurement has been used in many applications, such as computational biology, text mining, pattern recognition, and computer vision. In this paper, we apply similarity measurement on graphs to measure structural differences in financial statements. Unconventional financial statement structures may potentially reveal deceptive intention of hiding certain information while making technically “correct” financial statements. Furthermore, unconventional financial statements may also lead to investment opportunities if legitimacy is not questioned. We construct an algorithm based on the metric of string edit distance as an approximation of graph similarity, and apply the Levenshtein algorithm with modified string edit costs to measure string edit distance. We demonstrate the effectiveness of this algorithm in capturing the sensitive changes of balance sheet structures by applying the algorithm in two experiments. The first experiment shows the algorithm is sensitive to all three basic edits (namely deletion, insertion and substitution) on a particular balance sheet, and the second experiment shows more than 90% clustering accuracy on real balance sheets.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115514266","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611690
Tomohiro Nakada, K. Takadama
An agent-based simulation developed as a tool to analyze economic system and social systems since the 1990s. Previous paper reported that the simulation results indicated that the number of agents affects the trading prices and their distributions. To analyze the effect of the number of agents, this paper analyzes the relationship between the number of agents and simulation results using XCS agents for artificial trading. We report the market price fluctuation and population size of internal model by the number of agents. The revealed the following remarkable implications: (1) increasing number of XCS agents does not affect the convergence of population size of all agents; and (2) all agents converge towards approximately form 15 % to 20 %of population size by learning classifier system of XCS agents; and (3) increasing number of XCS agents reduce the variance of the market price.
{"title":"Analysis on the number of XCS agents in agent-based computational finance","authors":"Tomohiro Nakada, K. Takadama","doi":"10.1109/CIFEr.2013.6611690","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611690","url":null,"abstract":"An agent-based simulation developed as a tool to analyze economic system and social systems since the 1990s. Previous paper reported that the simulation results indicated that the number of agents affects the trading prices and their distributions. To analyze the effect of the number of agents, this paper analyzes the relationship between the number of agents and simulation results using XCS agents for artificial trading. We report the market price fluctuation and population size of internal model by the number of agents. The revealed the following remarkable implications: (1) increasing number of XCS agents does not affect the convergence of population size of all agents; and (2) all agents converge towards approximately form 15 % to 20 %of population size by learning classifier system of XCS agents; and (3) increasing number of XCS agents reduce the variance of the market price.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125632047","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-04-16DOI: 10.1109/CIFEr.2013.6611702
A. Garg, S. Sriram, K. Tai
Genetic programming (GP) and its variants have been extensively applied for modeling of the stock markets. To improve the generalization ability of the model, GP have been hybridized with its own variants (gene expression programming (GEP), multi expression programming (MEP)) or with the other methods such as neural networks and boosting. The generalization ability of the GP model can also be improved by an appropriate choice of model selection criterion. In the past, several model selection criteria have been applied. In addition, data transformations have significant impact on the performance of the GP models. The literature reveals that few researchers have paid attention to model selection criterion and data transformation while modeling stock markets using GP. The objective of this paper is to identify the most appropriate model selection criterion and transformation that gives better generalized GP models. Therefore, the present work will conduct an empirical analysis to study the effect of three model selection criteria across two data transformations on the performance of GP while modeling the stock indexed in the New York Stock Exchange (NYSE). It was found that FPE criteria have shown a better fit for the GP model on both data transformations as compared to other model selection criteria.
{"title":"Empirical analysis of model selection criteria for genetic programming in modeling of time series system","authors":"A. Garg, S. Sriram, K. Tai","doi":"10.1109/CIFEr.2013.6611702","DOIUrl":"https://doi.org/10.1109/CIFEr.2013.6611702","url":null,"abstract":"Genetic programming (GP) and its variants have been extensively applied for modeling of the stock markets. To improve the generalization ability of the model, GP have been hybridized with its own variants (gene expression programming (GEP), multi expression programming (MEP)) or with the other methods such as neural networks and boosting. The generalization ability of the GP model can also be improved by an appropriate choice of model selection criterion. In the past, several model selection criteria have been applied. In addition, data transformations have significant impact on the performance of the GP models. The literature reveals that few researchers have paid attention to model selection criterion and data transformation while modeling stock markets using GP. The objective of this paper is to identify the most appropriate model selection criterion and transformation that gives better generalized GP models. Therefore, the present work will conduct an empirical analysis to study the effect of three model selection criteria across two data transformations on the performance of GP while modeling the stock indexed in the New York Stock Exchange (NYSE). It was found that FPE criteria have shown a better fit for the GP model on both data transformations as compared to other model selection criteria.","PeriodicalId":226767,"journal":{"name":"2013 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2013-04-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117212971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：