Paper: arXiv 2401.00534
Authors: Shun Liu, Kexin Wu, Chufeng Jiang, Bin Huang, Danqing Ma
Abstract
In the realm of cryptocurrency, the prediction of Bitcoin prices has garnered substantial attention due to its potential impact on financial markets and investment strategies. This paper propose a comparative study on hybrid machine learning algorithms and leverage on enhancing model interpretability. Specifically, linear regression(OLS, LASSO), long-short term memory(LSTM), decision tree regressors are introduced. Through the grounded experiments, we observe linear regressor achieves the best performance among candidate models. For the interpretability, we carry out a systematic overview on the preprocessing techniques of time-series statistics, including decomposition, auto-correlational function, exponential triple forecasting, which aim to excavate latent relations and complex patterns appeared in the financial time-series forecasting. We believe this work may derive more attention and inspire more researches in the realm of time-series analysis and its realistic applications.
Complexity vs Empirical Score
- Math Complexity: 3.5/10
- Empirical Rigor: 2.0/10
- Quadrant: Philosophers — conceptual discussion, lower math and empirical scores
Why this score: The paper introduces standard time-series techniques and basic machine learning models with minimal advanced mathematics, and it lacks concrete backtesting, statistical metrics, or code/data availability, placing it in the Philosophers quadrant.
Research Flowchart
flowchart TD
A["Research Goal<br>Predict Bitcoin Prices & Enhance Interpretability"] --> B["Data Preparation<br>Historical Bitcoin Price Data"]
B --> C{"Hybrid ML Models Comparison"}
C --> D["Linear Regression<br>OLS/LASSO"]
C --> E["Long-Short Term Memory<br>LSTM"]
C --> F["Decision Tree Regressor"]
D --> G["Best Performance<br>Linear Regressor"]
E --> G
F --> G
B --> H["Interpretability Analysis<br>Time-series Statistics"]
H --> I["Decomposition<br>ACF<br>Exponential Triple Forecasting"]
G --> J["Key Outcome<br>Synergizing Performance & Interpretability"]
I --> J
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a comparative study of hybrid machine learning algorithms for Bitcoin price prediction, emphasizing model interpretability.
Source sentence
This paper propose a comparative study on hybrid machine learning algorithms and leverage on enhancing model interpretability.
Assumes 2
- The time series is assumed to be stationary when applying ACF and PACF analyses. statistical
Source sentence
They also assume underlying strationarity of time series, which is typically judged by hypothesis checking: if p −value is greateer than a critical value of 0.05, then we claim that the data has a unit root, hence non-stationary; otherwise the data is stationary.
- The random component (residual) is assumed to be independent and identically distributed (i.i.d.) with mean zero. statistical
Source sentence
Random also known as the residual or error term, represents the unpredictable fluctuations or noise in the time series. It includes all the factors that cannot be explained by the trend or seasonality. Random variations are typically assumed to be independent and identically distributed (i.i.d.) and have a mean of zero
Data
- Bitcoin closing prices market-prices unknown
Source sentence
Predicted value versus real value of bitcoin closing prices by configured Linear Regression(left) and Lasso(right), spanning from 2013-04 to 2021-07, note that Lasso outperforms other rivalries in this task.
Costs
Not applicable
Finds 3
- Lasso Regression achieves the lowest MAE (91.80) among candidate models, outperforming others. performance
Source sentence
Experimental results are displayed in the Table 1. It’s obvious that Lasso Regression outperforms other methods in terms of MAE(mean absolute error), further prove our assumption in Section 3.3.
- Linear Regression obtains the highest confidence score (0.9998) despite higher MAE. performance
Source sentence
Linear Regression owns the largest confidence score 0.9998.
- Lasso outperforms other rivalries in predicting Bitcoin closing prices over the test period. predictive-accuracy
Source sentence
Lasso outperforms other rivalries in this task.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 16 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (26)
- Research and Implementation of Traffic Sign Recognition Algorithm Model Based on Machine Learning (2023)
- Research and Implementation of Cancer Gene Data Classification Based on Deep Learning (2023)
- Autonomous navigation at unsignalized intersections: A coupled reinforcement learning and model predictive control approach (2023)
- Comprehensive Transformer-Based Model Architecture for Real-World Storm Prediction (2023)
- Utility-based route choice behavior modeling using deep sequential models (2023)
- The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis (1998)
- Impact of autocorrelation function model on the probability of failure (2019)
- Forecasting future climate boundary maps (2021–2060) using exponential smoothing method and GIS (2022)
and 18 more
- Developing a deep learning framework with two-stage feature selection for multivariate financial time series forecasting (2020)
- A new hybrid financial time series prediction model (2020)
- Holt–Winters smoothing enhanced by fruit fly optimization algorithm to forecast monthly electricity consumption (2020)
- Energy management optimisation using a combined long short-term memory recurrent neural network – particle swarm optimisation model (2021)
- Forecasting call center arrivals using temporal memory networks and gradient boosting algorithm (2023)
- Predicting energy consumption: A multiple decomposition-ensemble approach (2019)
- Real-time prediction of docker container resource load based on a hybrid model of ARIMA and triple exponential smoothing (2022)
- The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis (1998)
- Impact of autocorrelation function model on the probability of failure (2019)
- Forecasting future climate boundary maps (2021–2060) using exponential smoothing method and GIS (2022)
- Developing a deep learning framework with two-stage feature selection for multivariate financial time series forecasting (2020)
- A new hybrid financial time series prediction model (2020)
- Holt–Winters smoothing enhanced by fruit fly optimization algorithm to forecast monthly electricity consumption (2020)
- Energy management optimisation using a combined long short-term memory recurrent neural network – particle swarm optimisation model (2021)
- Forecasting call center arrivals using temporal memory networks and gradient boosting algorithm (2023)
- Predicting energy consumption: A multiple decomposition-ensemble approach (2019)
- Real-time prediction of docker container resource load based on a hybrid model of ARIMA and triple exponential smoothing (2022)
- Statistical and exponential triple smoothing approach to estimate the current and future deaths of Pakistani coal miners from 2010 to 2050 (2021)
0 of 26 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.