Scholarlink

For traders to gain insight into market trends and recognize significant support and resistance levels, accurate forecasting and analysis across many timeframes are essential. The current methods need to be expanded to encompass various timeframes. This research focuses on developing a model that enables traders to predict the price of the following candles. Multiple timeframes (4 h, 1 h, and 30 min) are considered for candlestick data. The best model for the prediction was chosen after evaluating various models implemented with AutoSkLearn and TPOT auto-machine learning libraries. Historical price data and technical indicators are used to train the models. Further, the performance of the models is evaluated using the criteria such as Mean Absolute Error (MAE), R2 Score, and Mean Absolute Percentage Error (MAPE). According to the test findings, the performance of the TPOT is better than AutoSklearn library for all three timeframes considered in this study. The open, close, high, and low prices of 30-min candlestick data are predicted with MAE values of 5.52, 6.75, 6.88, and 6.539, respectively.

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

Unlocking Multi-Timeframe Insights: AutoML-Driven Ethereum Price Prediction

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Surinder Singh Khurana,
  • Parvinder Singh,
  • Naresh Kumar Garg

摘要

For traders to gain insight into market trends and recognize significant support and resistance levels, accurate forecasting and analysis across many timeframes are essential. The current methods need to be expanded to encompass various timeframes. This research focuses on developing a model that enables traders to predict the price of the following candles. Multiple timeframes (4 h, 1 h, and 30 min) are considered for candlestick data. The best model for the prediction was chosen after evaluating various models implemented with AutoSkLearn and TPOT auto-machine learning libraries. Historical price data and technical indicators are used to train the models. Further, the performance of the models is evaluated using the criteria such as Mean Absolute Error (MAE), R2 Score, and Mean Absolute Percentage Error (MAPE). According to the test findings, the performance of the TPOT is better than AutoSklearn library for all three timeframes considered in this study. The open, close, high, and low prices of 30-min candlestick data are predicted with MAE values of 5.52, 6.75, 6.88, and 6.539, respectively.