Scholarlink

<p>In this research study, a machine learning based Adaptive Dollar Cost Averaging Technique (ADCAT) has been introduced. ADCAT is designed for trading highly volatile assets like cryptocurrencies such as Bitcoin and Ethereum. The methodology comprises two phases: model selection using AutoML libraries i.e. TPOT and AutoSkLearn and implementation of the ADCAT trading strategy. Historical price data of Bitcoin and Ethereum is used along with computed technical indicators to enhance predictive modeling. ADCAT uses optimal machine learning models for predicting next 4-hour candlestick’s close price and dynamically initiates trades based on these predictions to optimize profitability and manage risk. During the evaluation results demonstrated the exceptional performance of ADCAT. It achieved significantly higher profitability and improved risk management compared to other prominent approaches such as Buy &amp; Hold and OGCAT for Bitcoin and Ethereum as well. While trading Ethereum, ADCAT outperformed other approaches by achieving a profit of USD 1291.82, with a maximum drawdown of 4.6% and a Return over Maximum Drawdown of 28.11%. For Bitcoin trading, the best-performing model, ADCAT_1, generated a profit of USD 1110.98 with an RoMDD of 18.58%.</p>

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

ADCAT: Revolutionizing Dollar Cost Averaging with AutoML

  • 期刊论文

  • 发表日期：

    2025年12月22日

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

摘要

In this research study, a machine learning based Adaptive Dollar Cost Averaging Technique (ADCAT) has been introduced. ADCAT is designed for trading highly volatile assets like cryptocurrencies such as Bitcoin and Ethereum. The methodology comprises two phases: model selection using AutoML libraries i.e. TPOT and AutoSkLearn and implementation of the ADCAT trading strategy. Historical price data of Bitcoin and Ethereum is used along with computed technical indicators to enhance predictive modeling. ADCAT uses optimal machine learning models for predicting next 4-hour candlestick’s close price and dynamically initiates trades based on these predictions to optimize profitability and manage risk. During the evaluation results demonstrated the exceptional performance of ADCAT. It achieved significantly higher profitability and improved risk management compared to other prominent approaches such as Buy & Hold and OGCAT for Bitcoin and Ethereum as well. While trading Ethereum, ADCAT outperformed other approaches by achieving a profit of USD 1291.82, with a maximum drawdown of 4.6% and a Return over Maximum Drawdown of 28.11%. For Bitcoin trading, the best-performing model, ADCAT_1, generated a profit of USD 1110.98 with an RoMDD of 18.58%.