TL;DR: A four-model forecasting comparison using Bitcoin history from January 2012 through March 2021.
About This Document
The authors compare two deep-learning models, LSTM and bidirectional LSTM, with Facebook Prophet and Silverkite forecasting approaches. Their dataset spans roughly nine years of historical Bitcoin information. The paper reports Bi-LSTM error measures of 7.073 MAE and 3.639 RMSE and presents that model as the strongest result in its experiment. Those values depend on the paper's preprocessing, scale, and test design.
Why Bitcoin People May Care
This is useful as a reproducible period example of how researchers frame Bitcoin forecasting, choose model families, and compare error metrics. It does not show that the winning model can predict future markets outside the tested data. Price models age quickly, and none of the reported results should be treated as investment advice or a profit promise.
Document Details
• Title: Comparing Exponential Smoothing and Deep Learning for Bitcoin Price Prediction
• Author / organization: Nrusingha Tripathy et al.
• Year: 2025
• Language: English
• Document type: Research paper
• Pages: 9
Download access
Unlock a direct PDF download of this document. The download is hosted externally. CacheRat provides researched identification and access; purchase does not transfer ownership or expand the document’s existing usage rights.
#pdf #bitcoin #price_forecasting #deep_learning
About This Document
The authors compare two deep-learning models, LSTM and bidirectional LSTM, with Facebook Prophet and Silverkite forecasting approaches. Their dataset spans roughly nine years of historical Bitcoin information. The paper reports Bi-LSTM error measures of 7.073 MAE and 3.639 RMSE and presents that model as the strongest result in its experiment. Those values depend on the paper's preprocessing, scale, and test design.
Why Bitcoin People May Care
This is useful as a reproducible period example of how researchers frame Bitcoin forecasting, choose model families, and compare error metrics. It does not show that the winning model can predict future markets outside the tested data. Price models age quickly, and none of the reported results should be treated as investment advice or a profit promise.
Document Details
• Title: Comparing Exponential Smoothing and Deep Learning for Bitcoin Price Prediction
• Author / organization: Nrusingha Tripathy et al.
• Year: 2025
• Language: English
• Document type: Research paper
• Pages: 9
Download access
Unlock a direct PDF download of this document. The download is hosted externally. CacheRat provides researched identification and access; purchase does not transfer ownership or expand the document’s existing usage rights.
#pdf #bitcoin #price_forecasting #deep_learning
Sats-powered reveal
Unlock Download
One payment unlocks this reveal. Save your secret to return. No account needed.
Unlock with Secret
Views: 3
Categories: Bitcoin & Blockchain [c26], Law & Regulation [c31], PDFs & Documents [c17]