EVALUATING THE EFFECTIVENESS OF MACHINE LEARNING ALGORITHMS IN PREDICTING CRYPTOCURRENCY PRICES UNDER MARKET VOLATILITY: A STUDY BASED ON THE USA FINANCIAL MARKET

Аннотация

INTRODUCTION considerable challenges for investors, traders, and Background and Motivation financial institutions. As per Shamshad (2024), the According to Samson (2024), the cryptocurrency cryptocurrency world is unpredictable, and market has been among the most dynamic and influenced in a very complex way by a variety of volatile sectors within the global financial factors, from regulatory changes and adoption to ecosystem, with significant changes to the new technological breakthroughs and investment landscape in the US—financial market. macroeconomic conditions. This enhances not only Over the past ten years, cryptocurrencies have the risk of an investment in cryptocurrencies but gone from niche digital assets to mainstream also makes the job of prediction quite challenging. investment opportunities, including Bitcoin, In this respect, price prediction is an indispensable Ethereum, and many more. However, their need for any stakeholder interested in this intrinsic price instability provides quite emerging asset class (Khan et al, 2024; Shil et https://www.theamericanjournals.com/index.php/tajmei 16 THE USA JOURNALS THE AMERICAN JOURNAL OF MANAGEMENT AND ECONOMICS INNOVATIONS (ISSN- 2693-0811) VOLUME 06 ISSUE12 al.2024). Looking at algorithm performance in periods of high volatility and low volatility tries to reveal how Peng et al. (2024), posit that Machine learning ML algorithms react in moments of sudden changes algorithms have immense potential to in the market, which then depicts how adaptable revolutionize conventional financial forecasting and robust those very algorithms are. It forms a methods by offering advanced capabilities related very important layer when it comes to to the analysis of big datasets, pattern understanding the practical utility of algorithms: identification, and prediction. Contrary to real-world scenarios being light years away from conventional statistical models, which often cannot the stably structured academic ones. capture the nonlinear dynamic nature of cryptocurrency markets, Saha et al. (2023), LITERATURE REVIEW asserted that Machine Learning algorithms can Cryptocurrency Market Dynamics learn from such complexities and provide more Dudek et al. (2024), The cryptocurrency market reliable and actionable insights. Especially apt for has made a name for itself as a very peculiar and cryptocurrency price estimation amid volatility, it volatile part of the global financial topography. enshrines the ability to learn from historical data to Unlike more conventional financial assets, such as cater to changing market conditions. Therefore, stocks and bonds, cryptocurrencies are digital investors and researchers in turn have a growing assets that are decentralized and operate on interest in how the use of ML tools can aid a better blockchain technology. This decentralization, while forecast with a greater reduction in financial risks it promotes transparency and security, contributes (Islam et al. 2023). to significant price instability due to a lack of Research Objectives regulatory oversight and susceptibility to market The key objective of this research project is to sentiment. Akyildirim et al (2021), argued that investigate the effectiveness of various machine some of the most striking characteristics of the learning algorithms in the prediction of virtual currency market include high volatility, cryptocurrency prices within the volatile US with wild swings in prices sometimes in a very financial market. This research will pinpoint which short period. All these are influenced by several ML techniques provide the most accurate and factors including technological advances, reliable predictions under different market regulatory announcements, macroeconomic conditions, with a full understanding of their trends, and speculative trading behaviors (Alam et strengths and limitations. The framework al., 2024). therefore looks to analyze different algorithms, Basher & Sadorsky (2022), asserted that the prices such as linear regression, decision trees, support of cryptocurrencies have fluctuated wildly since vector machines, and neural networks, to provide a the creation of Bitcoin back in 2009, reflecting how comparative assessment performance framework dynamic and speculative this asset class is. Early for cryptocurrency price forecasting. Equally historic trends in the cryptocurrency markets were important, other key objectives include studying driven mostly by technological innovation and the the impact of market volatility on the accuracy of adoption of blockchain technology. Bitcoin-the first machine learning-based predictions. Volatility, as decentralized cryptocurrency-only received one might expect with cryptocurrencies, defines modest growth during its formative years and was these markets, where most traditional predictive generally relegated to niche communities of early models turn up their toes and refuse to perform. adopters. But with increased awareness, its value h