使用大数据与模糊逻辑、支持向量机预测富时100指数

需积分: 9 93 浏览量更新于2024-09-08 收藏 772KB PDF 举报

"这篇文档是关于利用大数据分析技术在金融领域的应用，特别是针对富时100指数（FTSE100 Index）的短期交易策略。作者包括Thomas Amorgianiotis, Konstantinos Theofilatos, Sovan Mitra, Efstratios Georgopoulos和Spiros Likothanassis。该研究于2016年提交至HAL（一个多学科的开放存取档案），并遵循Creative Commons Attribution 4.0 International License进行分发，允许在保持署名的情况下自由使用和共享。文章探讨了如何集成高容量的金融数据集来实现有利润且可解释的短期交易策略。研究中涉及的关键技术包括模糊逻辑（Fuzzy Logic）和支持向量机（Support Vector Machines）。模糊逻辑是一种处理不确定性和模糊信息的方法，它允许在非精确或模糊的条件下做出决策。而支持向量机则是一种强大的机器学习算法，常用于分类和回归任务，尤其在小样本数据上表现优秀。在金融市场的预测中，这两种技术结合可能有助于处理复杂的市场动态，如价格波动、交易量变化和市场情绪等。通过模糊逻辑，可以对不确定的数据进行处理，转化为更易于理解和操作的规则。而支持向量机则可以利用历史数据来学习模式，预测未来市场走势，从而指导交易决策。在富时100指数的案例中，研究人员可能通过构建模型来预测短期价格变动，这些模型基于大量的金融数据，如股票价格、交易量、宏观经济指标以及可能的市场新闻。通过集成这些数据，他们可能提高了预测的准确性和交易策略的盈利能力。同时，由于研究强调了策略的可解释性，这意味着模型不仅提供了预测，还能够提供一定程度的理解，解释为何作出特定的交易决策，这对于风险管理至关重要。这篇研究展示了大数据分析如何与高级机器学习技术相结合，以解决金融市场的复杂问题。它对于金融分析师、数据科学家以及关注金融科技（FinTech）领域的专业人士具有很高的参考价值，特别是在开发智能交易系统和市场预测模型方面。"

their dynamic and noisy nature. The FTSE 100 index trades futures contracts were

utilized as a case study. A future contract is a contract between two parties and its

cash settlement is determined by calculating the difference between the traded price

and the closing price of the index on the expiration day of the contract. Most tradi-

tional methods and more complex machine learning ones have recently failed to cap-

ture the complexity and the nonlinearities that exist in financial time series during the

latest crisis period. Certain disadvantages have been identified on traditional model-

ling and trading methods including the difficulties in tuning the parameters of the

algorithm, the disability of linear methods to provide good prediction results, the

overfitting problem and the fact that modelling and trading are most of the times con-

sidered as different problems. Lately, several machine learning methods have been

proposed to solve these problems [1]. Despite the encouraging results of new hybrid

methodologies their performance could be further enhanced.

Considering modelling and trading of the FTSE100 index, many researchers have

been occupied with this problem. These methods are either based on simple tradition-

al methods such as ARMA [2], or on machine learning techniques. The machine

learning applications range from simple neural network techniques, such as Higher

Order Neural Networks (HONN) [3] to more elaborate techniques such as the hybrid

method combining Artificial Bee Colony Algorithm with Recurrent Neural Networks

[4] and the hybrid methodologies which combine Genetic Algorithms with Support

Vector Machines (SVM) [5].

Despite the promising methods of the aforementioned techniques most of them

deal with the FTSE100 index as being independent and cut off from the global stock

market. However, the reality is very different. As expected several articles indicated

dependencies between the FTSE100 index and several other financial indices [6, 7]

and these dependencies have not yet been studied thoroughly. Thus, in order to

achieve optimal prediction and trading results using the FTSE100 index, several in-

puts from other financial indices should be utilized alongside with the traditional au-

toregressive and technical indicator inputs. The size of the universe of financial indi-

ces which could possibly be dependent with FTSE100 is so high that makes this prob-

lem relevant to the “big data” [8] topic which has gained the attention of the scientific

community lately.

In the present paper, we have created an integrated dataset for modelling and daily

trading with the FTSE100 index. This dataset, includes inputs and technical indicators

from a variety of financial time-series including VIX, S&P500, DAXX, Euro Stoxx

50, and EURGBP exchange rate. The major problems for utilizing such integrated

datasets for modelling and trading are: a) the difficulty to solve effectively the dimen-

sionality reduction problem and b) the difficulty to achieve interpretability for extract-

ing useful knowledge about the uncovered dependencies.

In the present study, we propose the ESVM Fuzzy Inference Trader to solve both

these problems. This method is based on a Support Vector Machine methodology to

achieve high performance predictions and it proposes an advanced technique to ex-

tract a meaningful compact set of fuzzy prediction rules. Moreover, a simple genetic

algorithm finds the optimal feature subsets which should be used as inputs and opti-

mizes the parameters of the overall modelling procedure. The final extracted inter-

剩余10页未读，继续阅读

scorpionxh

粉丝: 0
资源: 1

使用大数据与模糊逻辑、支持向量机预测富时100指数

Integrating deep learning to achieve phase compensation for free-space orbital-angular-momentum-encoded quantum key distribution under atmospheric turbulence

Integrating Materials Management with Financial....zip

Integrating Case Frame into Japanese to Chinese Hierarchical Phrase-based Translation Model

the future of fintech - integrating finance and technology in financial .pdf

the future of fintech - integrating finance and technology in financial services (2017).pdf

[UVM]Integrating RAL to Agent

Integrating Active Learning Strategy to the Ensemble Kernel-based Method for Protein-Protein Interaction Extraction

Oracle Receivables Guide to Integrating Oracle Receivables with

A Field Guide to Integrating XML and Web Services

Service Oriented Architecture A Field Guide to Integrating XML and Web Services

最新资源