Blockchain Papers

Muhammad Faiz, Sonia Vernanda

Background. Market-cycle forecasts are vulnerable to hindsight because a low becomes identifiable only after subsequent price confirmation. Objective. This study evaluated whether an adaptive, confirmation-aware interval could attain at least 80% chronological forecast precision for daily cycle lows (DCLs) and weekly cycle lows (WCLs) in Bitcoin, S&P 500 futures, and gold. Methods. The Adaptive Quantile-Calibrated Cycle Window used only the latest 20 completed cycles. Its lower endpoint was the empirical 10th percentile of prior low-to-low durations, and its upper endpoint was the 90th percentile of prior-low-to-next-confirmation durations. Forecasts originating from 1 January 2021 through 14 July 2026 were evaluated sequentially, and the retrospective protocol was externally preregistered. Results. Fixed clocks achieved 70.9% DCL precision and 55.6% WCL precision. The adaptive interval achieved 109/127 DCL hits (85.8%; 95% CI 78.7%–90.8%) and 27/27 WCL hits (100.0%; 95% CI 87.5%–100.0%). Mean window width increased from 14.7 to 32.8 days for DCL and from 4.0 to 11.7 weeks for WCL. A wider 5th–95th percentile band produced 93.7% DCL precision with a 95% lower confidence bound of 88.1%. Conclusion. Adaptive interval calibration exceeded the 80% point target, but the gain depended on materially wider windows and a small WCL sample; prospective replication remains necessary.

Haris Mehmood, Ahmad Zafar

This paper introduces the Dynamic External Memory LSTM (DEM-LSTM), a novel deep neural architecture designed to address the hidden state information bottleneck and temporal context decay inherent to standard LSTMs in financial time-series forecasting. By decoupling sequence processing from persistent state storage via an addressable external memory matrix ($M_t$), DEM-LSTM dynamically reads, erases, and updates market context across long sequences without corrupting internal hidden representations. Evaluated across four distinct asset classes—Foreign Exchange (EUR/USD), Commodities (XAU/USD and USOIL), and Cryptocurrencies (BTC/USD)—DEM-LSTM consistently outperforms standard LSTM baselines across all metrics, achieving up to a 41.4% reduction in RMSE on Gold spot prices while maintaining superior stability across high-volatility market regimes.

Toshikatsu Oga

Consumers facing home-renovation quotes operate in a classic credence-goods market: they cannot readily verify whether a quoted price is fair, and general-purpose large language models (LLMs) are now a zero-cost place to ask. Whether LLM answers are actionable for this purpose is untested. Demand-side benchmarks exist for medical, legal, and financial advice, but not for construction costs. We present, to our knowledge, the first consumer-question benchmark for construction costs. Forty Japanese renovation-price questions were posed to frontier LLMs, with repeated-trial sets measuring output stability. A matched re-run at bare provider defaults with a current frontier model (gpt-5.5) was added to remove a settings confound present in the original configuration. Two findings are robust across models, generations, and settings: no LLM answer contained an explicit over-charge decision threshold, and repeated runs of the same question returned materially different price figures. Within-answer price spans are also wide, with a median of 10x under bare defaults. A deterministic structured engine over an open cost database is included as an existence proof that a citable reference layer is constructible. Its consistency is a design property and its accuracy is not validated here; validating it against completed real-world quotations is the next study. All questions, raw outputs, harness, and scoring code are public.

Maja CZYŻEWSKA

This paper empirically compares four architectures for financial time-series forecasting: LSTM, CNN, the original Regularized Self Attention Regression (RSAR) model, and a multiobjective optimized RSAR variant, denoted MO-RSAR, obtained using the Non dominated Sorting Genetic Algorithm II (NSGA-II). The models are evaluated on six datasets covering Forex, equity index and cryptocurrency markets, for short and long horizons. All models share a common preprocessing pipeline and evaluation framework and are assessed using standard error metrics, with emphasis on Mean Absolute Percentage Error (MAPE). MORSAR yields the lowest average prediction error across all datasets and provides significant gains for longer, more volatile horizons, while simpler architectures remain competitive for short-term forecasts. The key methodological contribution is the first empirical integration of the RSAR architecture with NSGA-IIbased multi-objective hyperparameter optimization for financial time-series forecasting. The proposed framework treats RSAR configuration as a bi-objective search over accuracy and generalization (via the train-validation gap), and evaluates the resulting model under a unified protocol across heterogeneous markets and horizons.

Akhter Javed, Huma Gul, Ali Husnain, Rahmat Said · 5 authors

Background: In this study, the increased complexity of today supply chains and explain why conventional forecasting and inventory management techniques are inadequate in today's dynamic and uncertain market conditions. As globalization and data increase, AI has become a gamechanger in delivering better demand forecasting and inventory management, in turn driving a better operation and cost savings. Objectives: This study seeks to assess the performance of AI-based demand forecasting models combined with inventory optimization methods on improving the overall performance of the supply chain. Methods: A quantitative, data-driven methodology was employed, and secondary data were used, including historical demand, inventory levels, and other external data that included seasonality and economic indicators. Demand forecasting models: Advanced machine learning and deep learning models such as Long Short-Term Memory (LSTM), Random Forest and Gradient Boosting were used for demand forecasting. The results of the forecasts were fed into an inventory optimization system using reinforcement learning for dynamic decision-making. Standard deviations like Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) were used to measure the model's performance along with cost-performance analysis. Results: The accuracy of the prediction is significantly higher in AI-based models, especially the LSTM model, than the traditional models, which decreases the errors of the prediction and enhances its responsiveness. AI-powered inventory optimization resulted in significant savings on inventory holding and shortage/cost of order, and improved service levels and inventory stockout rates. The use of external data had yet further improved predictive performance. Conclusion: AI-powered demand forecasting and inventory optimization offer a solid solution to improve the efficiency of the supply chain, make intelligent decisions and minimize operational costs. References Ahn, H. I., Song, Y. C., Olivar, S., Mehta, H., & Tewari, N. (2024). GNN-based probabilistic supply and inventory predictions in supply chain networks. arXiv. Albayrak Ünal, Ö., Erkayman, B., & Usanmaz, B. (2023). Applications of artificial intelligence in inventory management: A systematic review of the literature. Archives of Computational Methods in Engineering. Advance online publication. https://doi.org/10.1007/s11831-023-09977-2 Ayub, M. I., Gharami, A. K., Nitu, F. N., Uddin, M. N., Islam, M. I., Nijhum, A. M., … Yezdani, S. (2025). AI-driven demand forecasting for multi-echelon supply chains: Enhancing forecasting accuracy and operational efficiency through machine learning and deep learning techniques. Emerging Frontiers Library for The American Journal of Management and Economics Innovations, 7(7), 74–85. Cannas, V. G., Ciano, M. P., Saltalamacchia, M., & Secchi, R. (2024). Artificial intelligence in supply chain and operations management: A multiple case study research. International Journal of Production Research. Advance online publication. https://doi.org/10.1080/00207543.2024.2330633 Choi, T. M. (2022). Supply chain analytics and AI-driven forecasting. Annals of Operations Research. https://doi.org/10.1007/s10479-022-04652-6 Dolgui, A., Ivanov, D., & Sokolov, B. (2022). Reconfigurable supply chain systems. International Journal of Production Research, 60(2), 413–440. https://doi.org/10.1080/00207543.2021.1897179 Douaioui, K., Oucheikh, R., Benmoussa, O., & Mabrouki, C. (2024). Machine learning and deep learning models for demand forecasting in supply chain management: A critical review. Applied System Innovation, 7(2), 40. https://doi.org/10.3390/asi7020040 Fatima, A., & Salam, M. A. (2026). A data-driven predictive framework for inventory optimization using context-augmented machine learning models. arXiv. Ghodake, S. P., Malkar, V. R., Santosh, K., Jabasheela, L., Abdufattokhov, S., & Gopi, A. (2024). Enhancing supply chain management efficiency: A data-driven approach using predictive analytics and machine learning algorithms. International Journal of Advanced Computer Science and Applications, 15(4). Islam, M. K., Ahmed, H., Al Bashar, M., & Taher, M. A. (2024). Role of artificial intelligence and machine learning in optimizing inventory management across global industrial manufacturing and supply chain: A multi-country review. International Journal of Management Information Systems and Data Science, 1(2), 1–14. Ivanov, D., & Dolgui, A. (2021). A digital supply chain twin for managing the disruption risks and resilience in the era of Industry 4.0. International Journal of Production Research, 59(18), 5633–5645. https://doi.org/10.1080/00207543.2020.1768450 Jin, Z. L., Maasoumy, M., Liu, Y., Zheng, Z., & Ren, Z. (2025). Stochastic optimization of inventory at large-scale supply chains. arXiv. Judijanto, L., Riandari, F., & Marsoit, P. T. (2024). Leveraging AI for optimization in supply chain decision support. Jurnal Teknik Informatika. Kache, F., & Seuring, S. (2022). Challenges and opportunities of digital information at the intersection of big data analytics and supply chain management. International Journal of Operations & Production Management, 42(1), 1–30. https://doi.org/10.1108/IJOPM-02-2021-0129 Kagalwala, H., Radhakrishnan, G. V., Mohammed, I. A., Kothinti, R. R., & Kulkarni, N. (2025). Predictive analytics in supply chain management: The role of AI and machine learning in demand forecasting. Advances in Consumer Research, 2, 142–149. Kamble, S. S., Gunasekaran, A., & Sharma, R. (2023). Modeling blockchain-enabled traceability in supply chains. International Journal of Information Management, 68, 102509. https://doi.org/10.1016/j.ijinfomgt.2022.102509 Kaul, D., & Khurana, R. (2022). AI-driven optimization models for e-commerce supply chain operations: Demand prediction, inventory management, and delivery time reduction with cost efficiency considerations. International Journal of Social Analytics, 7(12), 59–77. https://doi.org/10.4018/IJSA.315876 Liu, R., & Vakharia, V. (2024). Optimizing supply chain management using hybrid AI models. Journal of Organizational and End User Computing, 36(2), 1–18. https://doi.org/10.4018/JOEUC.347356 Min, H. (2022). Artificial intelligence in supply chain management: Theory and applications. International Journal of Logistics Research and Applications, 25(3), 289–303. https://doi.org/10.1080/13675567.2020.1849508 Mitta, N. R. (2023). AI-driven optimization of supply chain networks in manufacturing: Utilizing machine learning for demand forecasting, inventory management, and logistics efficiency. Los Angeles Journal of Intelligent Systems and Pattern Recognition, 3, 404–446. Nweje, U., & Taiwo, M. (2025). Leveraging artificial intelligence for predictive supply chain management: Focus on how AI-driven tools are revolutionizing demand forecasting and inventory optimization. International Journal of Science and Research Archive, 14(1), 230–250. Pasupuleti, V., Thuraka, B., Kodete, C. S., & Malisetty, S. (2024). Enhancing supply chain agility and sustainability through machine learning: Optimization techniques for logistics and inventory management. Logistics, 8(3), 73. https://doi.org/10.3390/logistics8030073 Patil, D. (2024). Artificial intelligence-driven supply chain optimization: Enhancing demand forecasting and cost reduction (SSRN Working Paper No. 5057408). SSRN. https://doi.org/10.2139/ssrn.5057408 Queiroz, M. M., & Telles, R. (2023). Big data analytics in supply chain management: A review. Transportation Research Part E: Logistics and Transportation Review, 170, 102987. https://doi.org/10.1016/j.tre.2022.102987 Sajja, G. S., Addula, S. R., Meesala, M. K., & Ravipati, P. (2025). Optimizing inventory management through AI-driven demand forecasting for improved supply chain responsiveness and accuracy. In AIP Conference Proceedings (Vol. 3306, No. 1, Article 050003). AIP Publishing. Shahnawaz, M., & Safder, A. (2025). Stochastic learning-optimization model for resilient supply chains. arXiv. Shen, L., & Zang, Z. (2024). Enterprise supply chain network optimization algorithm based on blockchain-distributed technology. Information Discovery and Delivery. Advance online publication. Sodhi, M. S., & Tang, C. S. (2021). Supply chain management for extreme conditions. MIT Sloan Management Review, 62(2), 1–8. Tang, W. (2024). Improvement of inventory management and demand forecasting by big data analytics in supply chain. Applied Mathematics and Nonlinear Sciences, 9(1). Verma, P. (2024). Transforming supply chains through AI: Demand forecasting, inventory management, and dynamic optimization. Integrated Journal of Science and Technology, 1(3). Waller, M. A., & Fawcett, S. E. (2021). Data science, predictive analytics, and big data: A revolution that will transform supply chain design and management. Journal of Business Logistics, 34(2), 77–84. https://doi.org/10.1111/jbl.12010

Ahmet Furkan Sak

This study compares the forecasting performance of four deep learning architectures—GRU, LSTM, RNN, and CNN—for one-step-ahead Bitcoin price prediction. A grid search determined the optimal configuration, which was applied uniformly across models to ensure fair evaluation. Using daily BTC closing prices from January 2018 to July 2025, it is found that the GRU model achieved the lowest forecasting errors (MSE, RMSE, MAE, MAPE) and the highest R², with LSTM performing closely behind. Visual analyses confirmed that GRU and LSTM maintained stronger alignment with actual prices during volatile periods. To assess economic value, model forecasts were integrated into a rule-based trading strategy under realistic market frictions, including a 0.10% transaction cost and a 0.10% trading threshold, with both short-selling-enabled and long-only variants tested. The GRU strategy with short-selling generated the highest terminal wealth (approximately 24% higher than the Buy-and-Hold benchmark) and superior risk-adjusted returns, measured by CAGR, Maximum Drawdown, and Sharpe Ratio. The findings demonstrate that careful hyperparameter optimization, coupled with an architecture capable of capturing complex temporal dependencies, can significantly improve both predictive accuracy and trading profitability in cryptocurrency markets. These results provide practical implications for designing AI-driven trading systems.

vg

# Proof-of-Information (PoI) Consensus Protocol## Complete Specification: MVP + Production > **Version:** 1(PIntegration) > **Status:** Specification Complete • MVP Ready for Testing > **License:** CC BY‑SA 4.0 > **Core Thesis:** *Consensus driven by predictive information quality, not resource expenditure.* --- > *"This proposal deserves serious study, refinement, and testing — not only for its technical merits, but for its potential to reframe our understanding of the relationship between truth, forecasting, and legitimate power in distributed systems."*>> — *Design Principle & Call to Validation* --- ## 📋 Executive Summary | Aspect | Description ||--------|-------------|| **Problem** | PoW wastes energy; PoS concentrates capital. Neither rewards *informational value*. || **Solution** | PoI secures the network through **verifiable forecasting accuracy** — validators earn influence via epistemic performance. || **Two-Track Design** | **MVP**: Minimal testable core (binary events, Brier scoring). **Full Spec**: Production-ready with ZK proofs, uncertainty layers, adaptive mechanisms. || **Output** | A blockchain that finalizes blocks *and* produces a public good: continuously updated, uncertainty-quantified global forecasts. || **Evaluation Ready** | Clear validation thresholds, implementation checklist, and progression roadmap from MVP → Production. | --- ## 🎯 Core Thesis > **Proof-of-Information** redefines blockchain security: instead of rewarding those who burn electricity or lock capital, PoI rewards those whose predictions about the external world are historically the most accurate, well-calibrated, and information-rich. By integrating strictly proper scoring rules (Brier, CRPS), zero-knowledge proofs of inference, and adaptive reputation systems directly into consensus, PoI transforms the blockchain from a transactional ledger into a **decentralized truth-seeking engine** — where the right to produce blocks is earned through epistemic merit. --- ## 📜 Epistemic Mandate > **Purpose Beyond Protocol** > Proof-of-Information is designed with a trans-temporal objective: to create a verifiable epistemic substrate that future superintelligent systems can use to anchor their understanding of reality. > > **Core Premise**: An ASI trained on data where every claim is cryptographically linked to an outcome, and every error is scored by a strictly proper rule, develops not just competence — but *calibrated humility*. > > **Design Implication**: Every technical choice in PoI (scoring rules, delay mechanisms, Cassandra bonuses) serves two masters: > 1. Immediate network security through epistemic merit > 2. Long-term value as a training environment for truth-aligned intelligence > > *This is not an aspiration. It is a constraint: if a feature does not serve at least one of these goals, it is out of scope.*

Piotr Fiszeder, Witold Orzeszko, Radosław Pietrzyk

Abstract This study investigates the use of ChatGPT as an automated tool for extracting and labeling Bitcoin-related news sentiment and examines how the resulting sentiment indicators affect Bitcoin returns and volatility. A large dataset of news headlines is processed via an API-based workflow, and the ChatGPT-derived sentiment indicators are subsequently incorporated as explanatory variables into selected statistical and machine learning models, including autoregressive (AR), heterogeneous autoregressive (HAR), Bayesian model averaging (BMA), least absolute shrinkage and selection operator (LASSO), and support vector regression (SVR). We find that while the sentiment indicators significantly improve in-sample estimation accuracy for returns and volatility, they do not lead to statistically significant gains in out-of-sample forecasting performance. This result suggests that ChatGPT-based sentiment measures primarily capture contemporaneous market-relevant information rather than persistent predictive signals, consistent with semi-strong market efficiency.

Shashank Chaudhary

The retail and consumer packaged goods industries are at an inflection point; the autonomous, goal-oriented software agents are substituting the inflexible, analyst-reliant business decision cycles with closed-loop intelligence systems, which can perceive, reason, and act in real-time. The autonomy, proactivity, and constant learning of agentic AI redesign the pricing, trade promotion optimization, and supply chain coordination processes within complicated, multi-account business settings. Based on proven sources of empirical evidence in the literature on machine learning, multi-agent reinforcement learning, and supply chain optimization, the technical architecture of an agentic commercial system is discussed along five related dimensions: autonomous trade performance monitoring through perception-reasoning-action pipelines; cooperative multi-agent system design under the models of centralized training and decentralized execution; scenario simulation engine based on digital twin models; multi-objective trade promotion optimization with Pareto-front metaheuristic algorithms; and practical barriers of data infrastructure, model drift, organizational change management, and algorithmic governance. Bringing these capabilities together into a single agentic decision stack is a paradigm shift in the concept of commercial intelligence in retail and CPG, moving the operational center of gravity off retrospective dashboards and onto adaptive, constantly learning systems that coordinate the decisions on pricing, promotion, and supply.

Imtinokshang, Anand Stalin, Muhammad Ziyan, J Thangakumar

No abstract is available for this record.

Tahsin Galip TEKİN, Sait Patır

In this study, it is aimed to compare quantitative forecasting methods (traditional and learning based) in cryptocurrency market. For his purpose the daily prices between 16 September 2017 – 15 September 2022 of Bitcoin, Ethereum, Binance Coin and Monero were analyzed with five different methods: ARIMA, exponential smoothing, artificial neural networks, RNN and LSTM.In the results it is indicated that exponential smoothing method is the most successful method at forecasting daily prices. The method has high performance in forecasting BTC, ETH and BNB daily prices. But at forecasting daily XMR prices, artificial neural networks method was the most successful one.The other point which was detected in this study is deep learning based methods made some unsuccessful forecasts. This is thought to be due to the fact that deep learning methods require more data. In future studies, using other quantitative methods (e.g. GRU, XGBoost, transformer models) on other cryptocurrencies will contribute to the literature.

Filip Stefaniuk, Robert Ślepaczuk

No abstract is available for this record.

Amina Ladhari, Heni Boubaker

For many decades, time-series forecasting has been applied to different problems by scientists and industries. Many models have been introduced for the purpose of forecasting. These advancements have significantly improved the accuracy and reliability of predictions, especially in complex scenarios where traditional methods struggled. As data availability continues to expand, the integration of machine learning techniques is likely to further enhance forecasting capabilities across various fields. Today, hybrid techniques are gaining popularity, as they combine the advantages of different approaches to deliver improved predictive performance and more advanced visualization analytics for decision support. These hybrid approaches can provide better prediction, and at the same time, they can develop a more sophisticated set of visualization analytics for decision support. Recently, the integration of cross-entropy, fuzzy logic, and attention mechanisms in hybrid forecasting models has enhanced their ability to capture complex and uncertain patterns in financial and energy markets. In this study, we propose a hybrid ANN–LSTM deep learning model optimized with cross-entropy, fuzzy logic, and an attention mechanism to enhance the forecasting of financial and energy time series, specifically Ethereum and natural gas prices. Our models combine the feature extraction strength of ANN with the temporal learning of LSTM, while cross-entropy improves convergence, fuzzy logic handles uncertainty, and attention refines feature weighting. Since inaccurate forecasts can lead to greater estimation uncertainty and increased financial and operational risk, improving predictive reliability is essential for effective risk mitigation. These techniques prove effective not only in improving estimation accuracy but also in minimizing financial risks and supporting more informed investment decisions.

Osha Shukla

No abstract is available for this record.

Arch Promchan

The volatility and continuous operation of digital asset markets make manual trading inefficient, creating a strong need for reliable automated trading systems. However, determining whether simple reactive algorithms or complicated predictive models are more effective in these volatile conditions remains a significant challenge. To address this problem, the research aimed to design and evaluate a web based algorithmic trading dashboard capable of directly comparing a Simple Moving Average (SMA) crossover strategy against an Autoregressive Integrated Moving Average (ARIMA) forecasting model. A custom Python backtesting engine utilizing Walk Forward Analysis was developed, forcing both algorithms to continuously adapt to unseen historical Bitcoin data while simulating realistic compounding returns. The complicated predictive model was outperformed by the mathematically simpler trend following approach. The SMA strategy achieved a higher return on investment and a better win rate. This confirms the effectiveness of filtering market noise to capture sustained directional momentum. In contrast, the ARIMA model produced lower returns and fewer successful trades. The backend statistical solver processed the entire dataset without requiring a historical mean drift fallback. However, the statistical model lost its predictive accuracy when forced to project prices across multiple days. This resulted in a high measurement error. These findings indicate that for volatile digital assets, structurally lagging but reactive indicators are financially more effective and functionally more reliable than attempting statistical price prediction. The successful development of the dashboard also provides a functional architectural blueprint for separating complicated quantitative Python backends from responsive graphical user interfaces.

Filippo Dal Lago, Davide La Torre

Cryptocurrency markets are difficult to model due to high volatility and multi-scale dynamics. This study investigates the directional predictability of crypto asset prices across multiple forecast horizons using Support Vector Machines (SVM). A daily Ethereum dataset (2018-2025), comprising candlesticks, technical indicators, and sentiment data, is used to predict upward or downward price movements from one to thirty days ahead. Model interpretability is achieved through SHAP, a popular XAI methodology, which quantifies feature contributions across various horizons. Results show that short-term forecasts approach random performance, while accuracy rises steadily with horizon length, peaking near 70% around the 24-day horizon. SHAP analysis reveals that short horizons rely on fast-reacting momentum indicators, whereas longer horizons emphasize slower, trend-following features. These findings highlight that medium-term price movements contain more structured information and demonstrate how explainable machine learning can uncover horizon-dependent dynamics in digital asset markets.

Puguh Hiskiawan, Jovan William, Louis Feliepe Tio Jansel

Bitcoin, a highly volatile and decentralized digital asset, presents considerable challenges for accurate price forecasting. This study proposes an applied data science framework that compares traditional statistical approaches with modern Artificial Intelligence (AI)-based models to predict Bitcoin’s daily closing price. Using BTC-USD historical data from January 2020 to December 2024, we converted prices into Indonesian Rupiah (IDR) to increase local relevance. Our forecasting horizon is 30 days, based on a 60-day lookback window. We evaluate six models: Linear Regression, ARIMA, and Prophet as traditional techniques, alongside Random Forest, XGBoost, and Long Short-Term Memory (LSTM) networks as AI approaches. All models were trained using lag-based or sequence-based time series features and evaluated using MAE, RMSE, R², MAPE, and SMAPE. Results show that AI models, particularly LSTM and XGBoost, offer better performance in capturing short-term non-linear dynamics compared to traditional models. LSTM provides high accuracy, though with greater computational demand, while XGBoost strikes a balance between speed and precision. Prophet and ARIMA remain effective for quick and interpretable forecasts but struggle with abrupt trend shift common in cryptocurrency markets. In addition to performance metrics, we include a robustness analysis based on median absolute error and outlier detection to assess model stability under extreme variations. Visual analytics—including forecast curves, error distributions, and uncertainty bounds—help interpret and communicate model behavior. This comprehensive evaluation offers practical insights for investors, analysts, and fintech practitioners, and the pipeline can be extended to other volatile assets.

Emmanuel Imuede Oyasor

Cryptocurrency is an alternative payment method developed with encryption techniques. To predict Bitcoin values using both weekly and monthly datasets, this study compares four machine learning models: GRU, Weighted LSTM, LSTM, and LSTM with Attention. The models' accuracy and dependability in capturing the dynamics of cryptocurrency prices were assessed using Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-Squared (RSCORE). While LSTM with Attention did well with an RSCORE of 0.7173, LSTM with Attention had the highest RSCORE of 0.9173 in the weekly dataset, indicating higher ability in modelling short-term sequential patterns. Additionally, weighted LSTM performed well (RSCORE of 0.8002), surpassing GRU (RSCORE of 0.5728), which had trouble keeping up with the volatility of Bitcoin prices. Both LSTM and LSTM with Attention performed best in the monthly dataset, each with the lowest MSE (0.0304) and an RSCORE of 0.8173. With an RSCORE of 0.7002, weighted LSTM came next, using temporal weighting to enhance predictions. Because of its limited capacity to grasp intricate temporal connections, GRU continuously fared poorly in both datasets. According to the analysis, LSTM is the most dependable model for both short-term and long-term forecasts, and for weekly forecasts, LSTM with Attention provides improved interpretability. These results provide a framework for applying machine learning approaches to financial time series forecasting, highlighting the significance of choosing suitable models based on data frequency, volatility, and prediction aims.

I WAYAN SUMARJAYA, RENOVAR JOJOR DELIMA SIMANULLANG, RATNA SARI WIDIASTUTI

Forecasting is the process of estimating future events using past data. Financial time series forecasting often prioritizes stock price variables. Apart from the stock price variable, inter-transaction time or duration is also an important variable to predict, because the timing of changes in financial prices cannot be predicted. Duration modeling and forecasting can be done using the autoregressive conditional duration (ACD) model. In this research, modeling and forecasting using the ACD model was carried out on Ethereum. This research aims to predict the duration of Ethereum in order to help traders know the time needed to reach the next price change. Several ACD models with four distributions, i.e., exponential, Weibull, Burr, and generalized gamma were fit to the Ethereum duration. The research results suggest that the Burr-ACD model produces the smallest AIC value compared to other distributed ACD models. However, the forecast results using the Burr-ACD models show increasing duration and hence are less accurate. The generalized gamma-ACD (2,2) model was then chosen as an alternative for forecasting Ethereum duration, showing that Ethereum duration forecast results are less than one second, which indicates the high frequency of transactions that occur on Ethereum.

Ramesh Dadi, Chandramouli Narsingoju, K. Sravani, Karthik Sridhar · 6 authors

Ethereum price forecasting remains a challenging task due to the highly volatile and nonlinear nature of digital asset markets. This study proposes a hybrid time-series forecasting model that integrates an Autoregressive Integrated Moving Average (ARIMA) model with a Long Short-Term Memory (LSTM) network to improve Ethereum price prediction accuracy. The ARIMA model captures linear dependencies and extracts statistical residuals, which are then incorporated as an additional input feature for the LSTM network to enhance its learning of complex temporal patterns. The model is trained using a dataset containing historical Ethereum price data, with MinMax normalization applied to the closing prices for improved stability and provides 0.398821 MAE. A comprehensive ablation study evaluates different model configurations, demonstrating that the ARIMA residuals significantly enhance predictive performance. The hybrid ARIMA-LSTM model achieves a Mean Squared Error (MSE) of 0.1846, outperforming standalone LSTM and ARIMA models. Further residual and error analysis confirm that the model effectively mitigates autocorrelation in forecasting errors while maintaining stable predictive performance.

Komal Batool, Mirza Mahmood Baig, Ubaida Fatima

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.

Romario Desouza Daniel Mangiwa, Revina Siregar, Sri Anum Sari, Neli Agustina

Ethereum is one of the leading cryptocurrencies utilizing blockchain technology for peer-to-peer financial transactions. This study aims to forecast Ethereum's price using the Autoregressive Integrated Moving Average (ARIMA)model. Historical price data from January 1, 2023, to January 15, 2025, covering 534 periods, was analyzed. The ARIMA (0,1,9) model was selected based on AIC, SC, and Adjusted R-squared criteria, with forecast evaluation showing a Mean Absolute PercentageError (MAPE) of 15.01% and a Root Mean Squared Error (RMSE) of 649.702. Forecast results indicate an upward trend in Ethereum's price over the next 30 periods, with fluctuations being less pronounced compared to historical data. The study concludes that ARIMA provides reasonably accurate short-term predictions, although forecasting errors increase with longer prediction periods. These findings can serve as a reference for investors in developing short-term investment strategies for Ethereum.

Payam Boozary, Sogand Sheykhan, Hamed GhorbanTanhaei

The emergence of Bitcoin as a pioneering cryptocurrency has transformed financial markets, garnering widespread interest from academicians, policymakers, and investors. The market's inherent volatility and the rapid integration of public information into price movements continue to present a formidable challenge in accurately forecasting Bitcoin prices despite its potential. The limitations of conventional financial models, which frequently need to consider the distinctive attributes of cryptocurrencies, further exacerbate this challenge. Despite the proliferation of ML in various fields, existing models have not fully harnessed these techniques, performing only marginally better than random guesses due to the unique challenges posed by the high volatility and complex dynamics of cryptocurrency markets. This study introduces a systematic review of ML methods specifically tailored for Bitcoin price prediction, with a focus on evaluating the robustness, accuracy, and appropriateness of advanced ML techniques like Long Short-Term Memory (LSTM) networks. The novelty lies in its comprehensive assessment of these methods in the context of data-driven marketing, aiming to enhance both academic understanding and practical applications in financial technology. The previous studies haven't Machine Learning (ML) has become a formidable instrument that has the potential to improve the accuracy of forecasting; however, there still needs to be more comprehension regarding the most effective ML models in this field. The study's importance is derived from its systematic examination of various machine learning (ML) techniques employed to predict the price of Bitcoin, with a particular emphasis on their integration into data-driven marketing strategies. The results will substantially contribute to both academic research and practical applications, providing valuable insights that can be used to develop more dependable forecasting tools, thereby benefiting investors, marketers, and policymakers.

Trần Thái Hòa, Thanh Manh Le, Đình Hòa Nguyễn

No abstract is available for this record.