Cryptocurrency markets are highly volatile, creating challenges for accurate risk management and forecasting. As digital assets become more integrated into financial systems, understanding their volatility dynamics is essential for investors and policymakers. Previous research has primarily applied standard GARCH models to cryptocurrencies, often neglecting advanced specifications that capture asymmetry, regime-switching, and long-memory effects. This limits the accuracy of volatility forecasts and fails to reflect the unique behaviour of digital assets. This study aims to identify the most effective GARCH-class models for forecasting volatility in Bitcoin, Ethereum, Binance Coin, and Ripple. We analyse daily returns from August 2017 to December 2024, applying eight advanced GARCH specifications: EGARCH, GJR-GARCH, FIGARCH, HYGARCH, MSGARCH, CS-GARCH, and Log-GARCH. Hyperparameter tuning is conducted via grid search across lag orders (p, q ∈ [1, 5]), mean equations, and error distributions. Model performance is evaluated using AIC, BIC, RMSE, and MAE. Results show that MSGARCH and EGARCH outperform symmetric and short-memory models, highlighting the importance of regime-switching and leverage effects. FIGARCH provides the best fit for Bitcoin and Ethereum, confirming long-memory persistence. Skewed Student’s t and GED distributions improve accuracy by capturing heavy tails and asymmetry. These findings demonstrate the limitations of standard GARCH models and underscore the value of advanced specifications in modelling cryptocurrency volatility. The study offers practical insights for traders and risk managers, contributing to more robust forecasting in non-stationary markets. Advanced GARCH models significantly enhance volatility prediction for digital assets. Future research could extend this framework to other speculative instruments or integrate machine learning techniques to further improve performance.
Introduction This article presents a method for short-term cryptocurrency price forecasting utilizing news headlines. Methods The study analyzes the impact of news on asset prices within one hour of publication, employing machine learning-based classification with BERT and GPT models, as well as GloVe vector representations. Results The proposed cascade classifier model enhances prediction accuracy by initially assessing the strength of a news item and subsequently forecasting the direction of price movement. Experimental results demonstrate the effectiveness of the developed classification model. Discussion The model achieves an accuracy of 79% in predicting price movements, confirming the potential of leveraging news headlines to improve short-term forecasts in cryptocurrency markets.
Adnan Sagheer, Ali Raza, Muhammad Rizwan Rashid Rana, Faiza Kiran
The highly stochastic, nonlinear, and volatile nature of Bitcoin prices poses significant challenges for accurate forecasting using traditional statistical models. To address this, we propose a hybrid deep learning architecture that combines the strengths of Convolutional Neural Networks (CNNs) for spatial feature extraction with Recurrent Neural Networks (RNNs), specifically Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU), for capturing long-term temporal dependencies. This integrated framework effectively models both spatial and temporal patterns from historical Bitcoin price data. The model was trained and evaluated using real-world Bitcoin datasets.Experimental results demonstrate that the proposed CNN+LSTM model outperforms traditional machine learning and standalone deep learning approaches. Specifically, it achieves a Root Mean Square Error (RMSE) of 245.76, a Mean Absolute Error (MAE) of 11.45, a Mean Absolute Percentage Error (MAPE) of 15.68%, an R² score of 0.92, and a Mean Bias Error (MBE) of 6.94. These results highlight the effectiveness and reliability of the proposed hybrid model in enhancing the accuracy and stability of financial time series forecasting, providing valuable insights for traders, investors, and financial analysts.
Xinran Huang, Linzhi Tan, Haozhe Su, Jeremy Eng‐Tuck Cheah
ABSTRACT One of the critical risks associated with cryptocurrency assets is the so‐called downside risk, or tail risk. Conditional Value‐at‐Risk (CVaR) is a measure of tail risks that is not normally considered in the construction of a cryptocurrency portfolio. In this paper, we propose a new approach to portfolio construction based on a deep learning CVaR utility function. This approach is designed to address the issue of tail risk. We evaluate the performance of this approach in comparison to other portfolio construction techniques, including the naïve, minimum variance and mean‐variance portfolios. Our findings indicate that the proposed approach outperforms traditional optimisation models.
Cryptocurrency markets are experiencing rapid growth, but this expansion comes with significant challenges, particularly in predicting cryptocurrency prices for traders in the U.S. In this study, we explore how deep learning and machine learning models can be used to forecast the closing prices of the XRP/USDT trading pair. While many existing cryptocurrency prediction models focus solely on price and volume patterns, they often overlook market liquidity, a crucial factor in price predictability. To address this, we introduce two important liquidity proxy metrics: the Volume-To-Volatility Ratio (VVR) and the Volume-Weighted Average Price (VWAP). These metrics provide a clearer understanding of market stability and liquidity, ultimately enhancing the accuracy of our price predictions. We developed four machine learning models, Linear Regression, Random Forest, XGBoost, and LSTM neural networks, using historical data without incorporating the liquidity proxy metrics, and evaluated their performance. We then retrained the models, including the liquidity proxy metrics, and reassessed their performance. In both cases (with and without the liquidity proxies), the LSTM model consistently outperformed the others. These results underscore the importance of considering market liquidity when predicting cryptocurrency closing prices. Therefore, incorporating these liquidity metrics is essential for more accurate forecasting models. Our findings offer valuable insights for traders and developers seeking to create smarter and more risk-aware strategies in the U.S. digital assets market.
This paper proposes Temporal Convolutional Networks (TCNs) for cryptocurrency forecasting at minute-resolution. TCNs show better accuracy to XGBoost, LightGBM, and LSTM. However, TCNs are less robust than the tree models regarding volatility. While TCNs require much more computations at inference than LightGBM, they run faster than LSTMs. The performance of TCNs is best configured using a TCN with dilated convolutions to capture the temporal patterns, and with residual connections for the stability.
Investment advisory services are now commonly offered by consulting firms with financial experts, typically for a monthly fee. Financial markets require specialized knowledge, but advancements in artificial intelligence have revolutionized this field. Deep learning algorithms, especially Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), are widely used to predict asset price trends in nonlinear time-series data. However, they demand large datasets and are prone to overfitting. Recently, combining deep learning with reinforcement learning has shown promise, though it requires intensive research and computational resources. This study introduces the BTC-PDPR (Bitcoin Price Direction Prediction Robot) model, which predicts Bitcoin's daily price direction using the Random Forest Regressor. As an ensemble-based machine learning model, it works effectively with smaller datasets and identifies key technical indicators influencing price trends. The model achieved a 99.20% accuracy rate on data from March 2018 to the present. It runs efficiently in Google Colab (v5e1 configuration), producing results in just 22 seconds. This paper outlines the methodology, reviews relevant studies from 2017 to 2024, highlights gaps in the literature, and emphasizes the study’s contributions to the field.
Cryptocurrencies like Bitcoin can be considered commodities under the Commodity Exchange Act (CEA) and the Commodity Futures Trading Commission (CFTC) has jurisdiction over cryptocurrencies considered to be commodities, particularly in the context of futures trading. This paper presents a method for long and short term trend prediction of certain cryptocurrencies which is predicated on an application of the Fractal Market Hypothesis. This is an area of market theory where the self-affine properties of a fractal stochastic field are used to model a financial time series. After an introduction to the underlying theory and mathematical modelling, a fundamental analysis of Bitcoin and Ethereum to U.S. Dollar exchange markets is conducted. This analysis is based on a consideration that a changes in polarity of the 'Beta-to-Volatility' and the 'Lyapunov-to-Volatility' ratios to indicate an impending change to the Bitcoin/Ethereum price trend signal. This is used to recommend a long, a short or a hold trading position for which algorithms are provided (coded in Matlab) and 'back-tested'. An optimisation of these algorithms is conducted, leading to a strategy for implementing an ideal range of the key parameters for 'driving' the algorithms developed. This is based on maximising the accuracy and profitability to assure a high level of confidence. The application of the trading strategy developed through this approach is demonstrated to provide useful information to aid cryptocurrency investments and quantify the likelihood that the market will become bull or bear dominant. Under stable conditions, Machine Learning (using the 'TuringBot') is shown to provide useful estimates of future price values and/or fluctuations over small event horizons in time. This minimises any \lq trading delay' caused by filtering the data and increases returns by providing optimal trade positions within a \lq micro-trend' that is too fast for detection otherwise. In certain cases, this increase can reach ~10%. The results presented confirm that Bitcoin and Ethereum exchanges are self-affine (fractal) stochastic fields with L\'evy distributions, displaying a Hurst Exponent of ~ 0.32, a Fractal Dimension of ~ 1.68 and Levy Index of ~1.22. They also confirm that the Fractal Market Hypothesis and its indices provide a suitable market model, that generates returns on investments that outperform all Buy and Hold strategies based on more standard market indices.
I Putu Bramasta Priadinata, I Gede Iwan Sudipa, Ni Putu Suci Meinarni, I Made Leo Radhitya · 5 authors
Cryptocurrency is a highly volatile digital asset that requires accurate predictive methods. This study compares the performance of three deep learning architectures LSTM, GRU, and Bi-LSTM in forecasting the prices of Bitcoin (BTC), Ethereum (ETH), and Binance Coin (BNB) using univariate historical data. Evaluation was conducted through regression metrics (RMSE and MAPE) and classification of price movement into five categories, ranging from very bearish to very bullish, assessed using a confusion matrix. The results show that GRU performed best for BTC (RMSE 974.72, MAPE 1.18%), while Bi-LSTM outperformed others for ETH and BNB (RMSE 43.19 and 6.83; MAPE 1.16% and 1.08%) and achieved the highest classification accuracy (55% and 52%). However, overall classification accuracy remains low, reflecting the complexity of cryptocurrency price patterns. The study is limited by its univariate approach without incorporating external variables. Its contribution lies in combining regression and classification evaluation, and it recommends exploring multivariate and ensemble models in future research.
Frederic Haase, Tom Celig, Oliver Rath, Detlef Schoder
Abstract The emergence of cryptocurrencies and decentralized finance (DeFi) applications brings unique challenges, including high volatility, limited fundamental valuation methods, and significant informational reliance on social media. Consequently, traditional trading algorithms and decision support systems (DSS) often fall short in effectively capturing these dynamics, underscoring the need for tailored solutions. Recent research on sentiment analysis in cryptocurrency trading has provided mixed evidence regarding its predictive power, highlighting limitations in generalizability and reliability due to the inherent noise of social media content. Addressing these limitations, this study explores crowd-based trading signals, explicit buy and sell recommendations shared by users on social media platforms including X (formerly Twitter), Reddit, Stocktwits, and Telegram. We apply an event study methodology to analyze over 28,000 trading signals extracted using natural language processing (NLP) techniques based on large language models (LLMs). Our findings demonstrate that these explicit crowd-based signals significantly predict short-term cryptocurrency price movements, particularly for assets with lower market capitalization and recent negative returns. An out-of-sample trading strategy using these signals achieves superior risk-adjusted returns, outperforming both a standard cryptocurrency index (CCI30) and the S&P 500. Additionally, we uncover the role of automated accounts (signal bots) actively disseminating trading recommendations. This research advances literature by introducing a precise alternative to sentiment analysis, contributing to the understanding of social media as a distributed financial information environment, and raising theoretical considerations about algorithmic agency and trust. Practical implications span investors, social media platforms, and regulators.
Michael Smith, Valerie Kilders, Todd Kuethe, Nicole Olynk Widmar
We examine the relationship between market performance of leading cryptocurrencies (Bitcoin and Ethereum), meme-stocks (AMC, GameStop), and subjects of corporate boycotts (Bud Light) using weekly market price and volume data along with social media data of weekly mentions (which total 337 million in this dataset) and net sentiment. Using vector autoregression (VAR) time series analysis along with Granger causality testing and structural breaks, we successfully predict trade volume of these various assets using social media data and price data. We also find that closing price data and trade volume are reliable predictors of net sentiment about crypto in online and social media. However, we struggle to predict the closing price for the group of assets studied. We also employ impulse response functions, finding evidence of a dynamic relationship occurring between online and social media net sentiment and online media volume with closing price and trade volume. These functions show that investor sentiment operates with a short memory lasting around 3 weeks, additionally these functions show that price generates a shock on trade volume but that crypto and meme-stock markets experience this differently. Our findings reinforce the notion that meme-stock traders and herd investors do not trade on market fundamentals but are instead sensitive to herding (or sentiment) movements. Our findings also suggest that compared to these meme-stock investors, crypto markets have more traditional motivations of loss aversion.
For any meaningful instructional delivery to take place, the teacher must clearly understand who the learners are: their strengths, weaknesses, environment, the goal of instruction, the pace to mention but a few.This process is better referred to as instructional analysis.This paper posits that instructional analysis, the foundational phase of instructional design, serves as the indispensable basis for achieving high-quality and impactful instructional delivery.It explores the multifaceted components of instructional analysis, including learner analysis, context analysis, content/task analysis, and performance analysis, demonstrating how insights derived from these processes directly inform strategic decisions regarding instructional strategies, media selection, and assessment design.Drawing upon established instructional design models and contemporary research, this paper highlights the benefits of thorough instructional analysis in optimizing learning outcomes, enhancing engagement, and ensuring the relevance and efficiency of educational interventions.It also addresses practical challenges in conducting instructional analysis, offering considerations for educators and designers in diverse learning environments, particularly within the evolving nature of education in the 21st century.
This study investigates the effectiveness of a hybrid forecasting model that combines Generalized Autoregressive Conditional Heteroskedasticity (GARCH) with Long Short-Term Memory (LSTM) neural networks, specifically applied to the cryptocurrency market, focusing on Bitcoin and Ethereum.The inherent volatility of cryptocurrencies presents substantial challenges for accurate price prediction, necessitating advanced methodologies that can adapt to fluctuating market conditions.We first utilize GARCH models to analyze and capture the time-varying volatility in the returns of Bitcoin and Ethereum, enabling a comprehensive understanding of the underlying market dynamics.Following this, we implement LSTM networks to exploit their capability to model complex, non-linear relationships in sequential data, enhancing the predictive power of the model.The performance of the GARCH-LSTM framework is rigorously evaluated using historical price data for Bitcoin and Ethereum, employing key metrics such as Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) to assess forecasting accuracy.The results demonstrate that the hybrid approach significantly outperforms traditional forecasting methods, providing more reliable predictions and insights into market trends.This study contributes to the growing body of literature on cryptocurrency forecasting by illustrating the potential of combining econometric techniques with advanced machine learning methods, offering valuable implications for traders and investors in the cryptocurrency ecosystem.However, the experimental results revealed that the LSTM model outperformed the other eight methods in terms of forecasting performance measures, the RMSPE validation is 0.112561, and the RMSE validation is 0.011456.
This systematic meta-review analyzes over 75 papers (2020-2025) applying deep learning (DL) techniques to cryptocurrency trading, adhering to PRISMA guidelines. It evaluates various DL architectures, including LSTM, GRU, CNN, and Transformers, and finds that DL methods outperform traditional approaches in managing the high volatility and non-linear patterns of crypto markets. Key findings highlight the promise of hybrid and ensemble models, the benefits of integrating blockchain data, sentiment analysis, and macroeconomic factors for improved predictions, and the potential of deep reinforcement learning for developing autonomous trading strategies with risk-adjusted returns. However, challenges such as model interpretability, nonstationary data, and real-world deployment persist. The review emphasizes emerging directions like explainable AI (XAI) for transparent decision-making and high-frequency trading applications, providing a critical synthesis of methodologies, empirical results, and research gaps to inform both academic research and practical trading system development.
Dimitris Kastoris, Dimitris Papadopoulos, Konstantinos C. Giotopoulos
Mathematical modeling plays a crucial role in supporting decision-making across a wide range of scientific disciplines. These models often involve multiple parameters, the estimation of which is critical to assessing their reliability and predictive power. Recent advancements in artificial intelligence have made it possible to efficiently estimate such parameters with high accuracy. In this study, we focus on modeling the dynamics of cryptocurrency market shares by employing a Lotka-Volterra system. We introduce a methodology based on a deep neural network (DNN) to estimate the parameters of the Lotka-Volterra model, which are subsequently used to numerically solve the system using a fourth-order Runge-Kutta method. The proposed approach, when applied to real-world market share data for Bitcoin, Ethereum, and alternative cryptocurrencies, demonstrates excellent alignment with empirical observations. Moreover, our method outperforms ARIMA models in terms of accuracy, showcasing its effectiveness for crypto market forecasting. The entire framework, including neural network training and Runge-Kutta integration, was implemented in MATLAB.
Makine öğrenmesi ve derin öğrenme, son yıllarda finansal piyasalarda tahminleme süreçlerinde sıklıkla başvurulan yöntemler arasında yer almaktadır. Bu yöntemler, özellikle yüksek oynaklığa sahip kripto para piyasalarında, yatırım kararlarını destekleyici araçlar olarak öne çıkmaktadır. Denetimli öğrenme algoritmaları, geçmiş verilere dayanarak gelecekteki fiyatları tahmin etmeye yönelik güçlü çözümler sunarken; derin öğrenme yaklaşımları, özellikle sıralı veri yapılarındaki karmaşık ilişkileri yakalamada avantaj sağlamaktadır. Bitcoin gibi dijital varlıkların ekonomik değişkenlerle olan ilişkisini anlamak, hem bireysel yatırımcılar hem de finansal kurumlar açısından stratejik bir gereklilik haline gelmiştir. Bu bağlamda açıklanabilir yapay zeka (XAI) teknikleri, tahmin modellerinin iç mantığını şeffaf biçimde ortaya koyarak karar destek sistemlerine katkı sunmaktadır. Bu çalışmada, Bitcoin’in (BTC) günlük kapanış fiyatları, altın, USDX, VIX ve Brent petrol gibi ekonomik göstergeler dikkate alınarak SVR, LSTM, XGBoost ve ANN algoritmaları ile tahmin edilmiştir. 2014–2024 dönemini kapsayan veri setiyle gerçekleştirilen analizde, MAPE, MAE ve R² gibi performans ölçütleri üzerinden karşılaştırma yapılmıştır. Sonuçlara göre, en yüksek doğruluk oranını SVR modeli göstermiştir. Ayrıca SHAP analizi kullanılarak BTCY (yükseliş), BTCD (düşüş) ve BTCA (açılış) değişkenlerinin tahmin sürecinde pozitif katkı sunduğu belirlenmiştir. Buna karşılık, AK (altın) ve BrPK (petrol) gibi dışsal faktörlerin olumsuz etkiler yarattığı gözlemlenmiştir. Elde edilen bulgular ile hem model doğruluğunu hem de değişkenlerin etkisini şeffaf biçimde ortaya koyarak literatüre önemli katkılar sunulmuştur.
In this note, we make a comparison between a novel machine learning method, Long Short-Term Memory (LSTM), and two trading strategies using technical analysis: Exponential Moving Average (EMA) crossing and Moving Average Convergence/Divergence with Average Directional Index (MACD+ADX). The purpose is to use trading signals to maximize profits in the Bitcoin digital commodity. The comparison was motivated by the approval of the first spot Bitcoin exchange-traded funds (ETFs) by the U.S. Securities and Exchange Commission (SEC) on January 9, 2024. The results show that the LSTM algorithm delivers a cumulative return of approximately 65.23% over a testing period of less than nine months, significantly outperforming both the EMA and MACD+ADX strategies, as well as the baseline buy-and-hold approach typically followed by fundamental investors. Our work highlights the potential for further integration between machine learning and technical analysis in the evolving landscape of cryptocurrency markets.
This paper proposes a comprehensive deep-learning framework, SentiStack, for Bitcoin price forecasting and trading strategy evaluation by integrating multimodal data sources, including market indicators, macroeconomic variables, and sentiment information extracted from financial news and social media. The model architecture is based on a Stacking-LSTM ensemble, which captures complex temporal dependencies and non-linear patterns in high-dimensional financial time series. To enhance predictive power, sentiment embeddings derived from full-text analysis using the DeepSeek language model are fused with traditional numerical features through early and late data fusion techniques. Empirical results demonstrate that the proposed model significantly outperforms baseline strategies, including Buy & Hold and Random Trading, in cumulative return and risk-adjusted performances. Feature ablation experiments further reveal the critical role of sentiment and macroeconomic inputs in improving forecasting accuracy. The sentiment-enhanced model also exhibits strong performance in identifying high-return market movements, suggesting its practical value for data-driven investment decision-making. Overall, this study highlights the importance of incorporating soft information, such as investor sentiment, alongside traditional quantitative features in financial forecasting models.
Decentralized Finance (DeFi) faces critical security challenges due to its pseudonymous and permissionless nature, which exposes it to fraud and market instability. Existing approaches, such as single-agent reinforcement learning (RL) and static graph-based fraud detection, struggle to capture dynamic multi-agent interactions and evolving financial risks. This study proposes an integrated framework combining Multi-Agent Reinforcement Learning (MARL) and Graph Neural Networks (GNNs) to address adaptive decision-making and real-time fraud detection in DeFi. MARL agents, trained in DeepMind’s Melting Pot environment, optimize trading, liquidity provisioning, and arbitrage strategies, while GNNs analyze transaction graphs to detect anomalous patterns. Experimental results demonstrate that MARL agents achieve a 210% increase in average profit per trade and a 57% improvement in market adaptation, alongside a 120% rise in liquidity utilization. The GNN model attains a converged loss below 0.10, reducing false positives by 29%. The integrated system enhances market stability, achieving a stability impact score of 175 within 10 training episodes. This work establishes a scalable, intelligent framework for fraud-resistant trading, cross-chain compliance, and decentralized risk management, advancing the security and efficiency of DeFi ecosystems.
Predicting the price of Bitcoin is crucial, primarily because of the market’s rapid volatility and non-linear environment. For enhanced prediction of the price of Bitcoin, this research proposed a novel interpretable hybrid technique that combines long short-term memory (LSTM) networks with convolutional neural networks (CNN). Deep variational autoencoders (VAE) are used in the stage of preprocessing to determine noticeable patterns in datasets by learning features from historical Bitcoin price data. The CNN-LSTM model additionally implies Shapley additive explanations (SHAP) to promote interpretability and clarify the role of various features. For better performance, the methodology used data cleaning, preprocessing, and effective machine-learning techniques. The hybrid CNN + LSTM model, in collaboration with VAE, obtains a mean squared Error (MSE) of 0.0002, a mean absolute error (MAE) of 0.008, and an R-squared (R2) of 0.99, based on the experimental results. These results show that the proposed model is a good financial forecast method since it effectively reflects the complex dynamics of primary changes in the price of Bitcoin. The combination of deep learning and explainable artificial intelligence improves predictive accuracy as well as transparency, thus qualifying the model as highly useful for investors and analysts.
Purpose: This study aims to examines advanced portfolio management techniques using Long Short-Term Memory (LSTM) networks, the study was applied to investing in cryptocurrencies whose markets are characterized by high-frequency trading, and using behavioral finance models based on the concept of return-risk and deep learning based on the work of artificial neural networks (ANN) and long-term memory (LSTM) algorithms Design/Methodology/Approach: This study adopts quantitative approach. Moreover, A random portfolio consisting of 25 cryptocurrencies was selected based on the database of the website: https://finance.yahoo.com/crypto/ during the period 2021-2024 AD and programming the Python language. And an attempt to evaluate the performance of the models used in accurately predicting the optimal relative weights of the investment portfolio, which proved the relative effectiveness of deep learning models by estimating the values of the mean square error (MSE) at a level of 0.0218% to predict the optimal portfolio weights for 5 days based on training 80% and testing 20% of the study data. Findings: The second hypothesis of this study was accepted, which states the effectiveness of deep learning algorithms to predict the weights of optimal portfolios with a return estimated at 1.7239% and a risk of 1.1219% and a Sharpe index value estimated at 1.5365%, while the Markowitz return-risk model portfolio came with a return rate estimated at 31.15% and a risk of 39.05%. With no diversification of investment on all portfolio assets and a Sharpe index value of 0.7978%. Practical Implications: This study provides important insights that machine learning offers significant advantages in portfolio optimization, from improved forecasting of asset returns to dynamic rebalancing, better risk management, and automation. The ability to handle high-dimensional, non-linear, and non-stationary data makes ML an ideal tool for optimizing portfolios in complex and fast-moving markets; especially in cryptocurrency markets. However, challenges like data quality, overfitting, and interpretability must be addressed to ensure effective deployment of ML in real-world portfolio. Originality/Value: This study provides an original and timely contribution to understanding the use of deep learning for portfolio optimization represents a significant advancement over traditional financial models by offering several original and valuable benefits. These include the ability to capture complex non-linear relationships, dynamic rebalancing in response to real-time data, processing of unstructured data (like sentiment analysis), advanced risk management, and the integration of high-dimensional data. The combination of these capabilities enables more accurate, adaptive, and robust portfolio optimization, ultimately enhancing portfolio performance and reducing risk.
Deep learning and hybrid deep learning models are widely regarded as some of the most effective predictive modeling techniques to date. Their hierarchical architecture enables them to capture complex, non-linear relationships among features and uncover hidden patterns within data, making them particularly powerful for tasks involving high-dimensional and unstructured inputs. But, these models are computationally intensive and require substantial processing time. Moreover, their predictive efficiency is highly dependent on the availability of large-scale datasets. In this study, meta learning model is employed for the prediction of two financial markets: equity market and crypto market. NASDAQ and S&P 500 index has been taken for equity market prediction. On the other hand, Bitcoin & Ethereum are considered for crypto market. Three deep learning models: LSTM, GRU and CNN are trained for the prediction of these four indices and a hybrid deep learning model of GRU and CNN is also developed. Based on RMSE, MAE and R 2 values, it is observed that meta learning yields best results among all trained models with minimum time and using scarce computation resources based on small dataset.
Najma Ali Soomro, Suresh Kumar Oad RAJPUT, Ishfaque Ahmed
Predictions regarding returns and price movements in financial markets can be made using online search engines, which track the sentiments of individual investors. This study aims to analyse how the sentiments of Bitcoin investors impact changes in the American stock market returns. The Bitcoin sentiment index was created to benchmark the sentiments of Bitcoin investors from 2013 to 2018. This index is built by analysing terms from leading business magazines and online journals. Such an index measures potential investors’ sentiments about Bitcoin and how those sentiments impact S&P returns. We use the ordinary least squares method to analyse this. It was found that BSI has a negative impact on S&P returns. Furthermore, the Vector Autoregressive (VAR) model is used to determine the relationship between these economic time series. VAR results indicated a significant positive impact of S&P returns on BSI, while BSI could not predict S&P returns. Consequently, it can be concluded that S&P returns cause changes in BSI. Recognising that Bitcoin sentiment can offer valuable insights and guidance for retail investors during market downturns, much like the S&P 500. By tracking changes in the S&P 500, analysts can anticipate shifts in cryptocurrency market sentiment and take preventative measures when needed. Understanding this relationship is crucial for assessing systemic risks, as volatility in traditional markets can impact the crypto space.
This study aims to develop a hybrid algorithm using the ARIMA model and LSTM-type recurrent neural networks to predict the closing prices of the cryptocurrencies BTC, LTC, and ETH. The methodology includes an exploratory data analysis, followed by the design, implementation, and evaluation of each individual algorithm as well as the combined hybrid algorithm. The results, after experimentation and evaluation of metrics on the test set, indicated that the ARIMA model was inefficient in predicting the closing prices of cryptocurrencies. On the other hand, the hybrid model for BTC showed significant statistical differences in the metrics, with MAE = $726.21 and MAPE = 1.75%, compared to the LSTM model, which achieved MAE = $729.35 and MAPE = 1.76%. These results indicate better performance from the hybrid model. Regarding the RMSE metric, the hybrid model scored 1157.47, while LSTM scored 1159.99; although statistically equivalent, the hybrid model was numerically better. For the remaining metrics and other cryptocurrencies, both methods were statistically equivalent. For five-day-ahead predictions, the hybrid algorithm continued to yield better results for LTC and ETH.