Paper: arXiv 2306.08157
Authors: Rasoul Amirzadeh, Dhananjay Thiruvady, Asef Nazari, Mong Shan Ee
Abstract
Cryptocurrencies have gained popularity across various sectors, especially in finance and investment. Despite their growing popularity, cryptocurrencies can be a high-risk investment due to their price volatility. The inherent volatility in cryptocurrency prices, coupled with the effects of external global economic factors, makes predicting their price movements challenging. To address this challenge, we propose a dynamic Bayesian network (DBN)-based approach to uncover potential causal relationships among various features including social media data, traditional financial market factors, and technical indicators. Six popular cryptocurrencies, Bitcoin, Binance Coin, Ethereum, Litecoin, Ripple, and Tether are studied in this work. The proposed model’s performance is compared to five baseline models of auto-regressive integrated moving average, support vector regression, long short-term memory, random forests, and support vector machines. The results show that while DBN performance varies across cryptocurrencies, with some cryptocurrencies exhibiting higher predictive accuracy than others, the DBN significantly outperforms the baseline models.
Complexity vs Empirical Score
- Math Complexity: 5.5/10
- Empirical Rigor: 6.5/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper uses advanced probabilistic models (Dynamic Bayesian Networks) which requires mathematical sophistication, while also implementing a full comparison with multiple baselines and diverse datasets (six cryptocurrencies, multiple feature sets), indicating strong empirical testing.
Research Flowchart
flowchart TD A["Research Goal<br>Predict Crypto Price Directions<br>& Uncover Causal Relationships"] --> B["Data Collection & Preprocessing<br>6 Cryptos + Social Media + Financial + Technical Indicators"] B --> C["Model Development<br>Dynamic Bayesian Network DBN"] C --> D["Computational Process<br>Time Series Analysis & Causal Inference"] D --> E["Baseline Comparison<br>ARIMA, SVR, LSTM, RF, SVM"] D --> E E --> F["Key Findings/Outcomes<br>DBN Outperforms Baselines<br>Causal Relationships Identified"]
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 dynamic Bayesian network (DBN)-based approach for cryptocurrency price direction prediction that uncovers causal relationships among social media, market, and technical features.
Source sentence
we propose a dynamic Bayesian network (DBN)-based approach to uncover potential causal relationships among various features including social media data, traditional financial market factors, and technical indicators.
Assumes 3
- The system follows a first-order Markov process, where the state at time t depends only on the state at t‑1. model
Source sentence
a basic assumption in constructing DBN is that the system follows a first-order Markov process, in which the state of the system at time t only depends on its state at the previous time slice t −1.
- Each feature group is independent of the others, with no inter‑slice arcs between consecutive timeframes. model
Source sentence
In this model, we assume that each feature group is independent of one another, meaning no inter-slice arcs between two consecutive timeframes.
- The analysis uses five years of historical intraday closing‑price data for the selected cryptocurrencies. data
Source sentence
five years of historical intraday closing-price data.
Data
- Historical intraday closing-price data market-prices unknown
Source sentence
five years of historical intraday closing-price data.
Yahoo Finance Binance
Costs
Not applicable
Finds 4
- DBNs consistently outperform ARIMA, SVR, LSTM, RF, and SVM models, showing an average precision improvement of at least 15%. performance
Source sentence
DBNs consistently outperform ARIMA, SVR, LSTM, RF, and SVM models, showing an average precision improvement of at least 15%.
- Combining OHLCV price information with technical indicators yields the highest average precision (70.05%) across all cryptocurrencies, while using only OHLCV gives the lowest (66.57%). predictive-accuracy
Source sentence
We see that on average, the combination of OHLCV and technical indicators yields the best results for all cryptocurerncies (average precision 70.05%). Conversely, DBNs built solely on OHLCV achieve the lowest performance (average precision 66.57%).
- Adding more features does not always improve performance; in some cases it decreases precision. negative
Source sentence
adding more features to the DBN models does not always result in further improvements, but sometimes decreases the performance.
- For Binance Coin, Litecoin, and Ripple, introducing technical indicators substantially improves performance; for Ethereum and Tether it leads to a slight decrease. performance
Source sentence
The influence of technical indicators on DBN performance varies among cryptocurrencies. For Binance Coin, Litecoin, and Ripple, introducing technical indicators to the basic DBN model (OHLCV) substantially
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 29 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.