Short term prediction of market prices using tick data

2018 Fiscal Year Final Research Report

• Project/Area Number: 16K12492
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Intelligent informatics
• Research Institution: Keio University
• Principal Investigator: Sakurai Akito 慶應義塾大学, 理工学部(矢上), 教授 (00303339)
• Research Collaborator: Serjam Chanakya
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: 外国為替証拠金取引 / 収益分布 / ティックデータ / 金融時系列

Outline of Final Research Achievements

The purpose of the research is to improve accuracy of short and medium horizon (minutes to hours) prediction of asset prices in financial markets, specifically focused on minutely and tick movement of exchange rates in foreign exchange markets (mainly EUR/USD and USD/JPY).
(1) We successfully modeled 1 to 90 minutely returns of the rates and also showed that the long-term predictability of the model is based on the fractal-like property of return reversals.
(2) We successfully built models to explain the return reversals and regress the distribution of the minutely and tick returns, which is supported by simulations for 2001 to 2015 data.

Free Research Field

機械学習

Academic Significance and Societal Importance of the Research Achievements

金融市場における資産価格変動の重要な例である外国為替市場におけるbid値の変動に関し、収益反転の存在を明確に示し、それを反映した、価格変化である収益の分布をこれまでにない精度で記述できるモデルが構築できたことは、為替変動のリスク評価の精緻化、他の資産の価格変動のモデル化の研究に、大きく寄与すると考えられる。