Blockchain Papers

Follow blockchain research across journals, conferences, and preprint repositories.

2 papersLast indexed Aug 31, 2026
Search papers

Paper index

2 results · page 1 of 1

Clear filters
Aug 21, 2026·Enigma in Economics
0 cites
Adaptive Quantile Calibration of Daily and Weekly Cycle-Low Forecasts in Bitcoin, S&P 500 Futures, and Gold

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.

Open access
2 source records
Circadian rhythm and melatonin
Forecasting Techniques and Applications
Climate Change and Health Impacts
Original source
Aug 8, 2026·Zenodo (CERN European Organization for Nuclear Research)
0 cites
Overcoming Context Bottlenecks in Financial Time-Series Forecasting via Dynamic External Memory Augmented LSTMs

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.

Open access
2 source records
Stock Market Forecasting Methods
Time Series Analysis and Forecasting
Forecasting Techniques and Applications
Original source