Performance Evaluation of a Multi-Timeframe Multi-Head Attention System for Futures Market Forecasting and Day Trading 


Vol. 51,  No. 2, pp. 363-373, Feb.  2026
10.7840/kics.2026.51.2.363


PDF Full-Text
  Abstract

This study presents a high-frequency trading system that combines a multi-head attention-based multi-timeframe transformer with a trend-driven labeling method. The labeling defines price changes within a threshold as continuous trends, producing more stable labels than conventional up/down schemes. Parallel processing of multi-timeframe data enables simultaneous short- and long-term dependency learning. Experiments show superior performance over existing models in accuracy, profit factor, and maximum drawdown, while reducing trades by about fivefold, enhancing both reliability and efficiency in futures markets

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Related Articles
  Cite this article

[IEEE Style]

W. Shon and J. Kim, "Performance Evaluation of a Multi-Timeframe Multi-Head Attention System for Futures Market Forecasting and Day Trading," The Journal of Korean Institute of Communications and Information Sciences, vol. 51, no. 2, pp. 363-373, 2026. DOI: 10.7840/kics.2026.51.2.363.

[ACM Style]

Woojin Shon and Jaeyun Kim. 2026. Performance Evaluation of a Multi-Timeframe Multi-Head Attention System for Futures Market Forecasting and Day Trading. The Journal of Korean Institute of Communications and Information Sciences, 51, 2, (2026), 363-373. DOI: 10.7840/kics.2026.51.2.363.

[KICS Style]

Woojin Shon and Jaeyun Kim, "Performance Evaluation of a Multi-Timeframe Multi-Head Attention System for Futures Market Forecasting and Day Trading," The Journal of Korean Institute of Communications and Information Sciences, vol. 51, no. 2, pp. 363-373, 2. 2026. (https://doi.org/10.7840/kics.2026.51.2.363)
Vol. 51, No. 2 Index

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880