Scholarlink

Labeling financial data can significantly influence the training of an algorithmic trading system. In this paper we present a new dynamic method for determining the label assigned at a given point in time which introduces two important parameters: (a) it takes into account the current standard deviation of the price changes (on a logarithmic scale) and (b) it yields a continuous value on the label. The new method is compared with traditional labelling methods using a simple softmax classifier.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dynamic Labelling of Financial Data

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Dimitrios S. Vlachos,
  • Adamantia N. Mavrogianni,
  • Konstantinos T. Kantoutsis

摘要

Labeling financial data can significantly influence the training of an algorithmic trading system. In this paper we present a new dynamic method for determining the label assigned at a given point in time which introduces two important parameters: (a) it takes into account the current standard deviation of the price changes (on a logarithmic scale) and (b) it yields a continuous value on the label. The new method is compared with traditional labelling methods using a simple softmax classifier.