Paper: arXiv 2404.19324

Authors: Hulusi Mehmet Tanrikulu, Hakan Pabuccu

Abstract

Forecasting cryptocurrencies as a financial issue is crucial as it provides investors with possible financial benefits. A small improvement in forecasting performance can lead to increased profitability; therefore, obtaining a realistic forecast is very important for investors. Successful forecasting provides traders with effective buy-or-hold strategies, allowing them to make more profits. The most important thing in this process is to produce accurate forecasts suitable for real-life applications. Bitcoin, frequently mentioned recently due to its volatility and chaotic behavior, has begun to pay great attention and has become an investment tool, especially during and after the COVID-19 pandemic. This study provided a comprehensive methodology, including constructing continuous and trend data using one and seven years periods of data as inputs and applying machine learning (ML) algorithms to forecast Bitcoin price movement. A binarization procedure was applied using continuous data to construct the trend data representing each input feature trend. Following the related literature, the input features are determined as technical indicators, google trends, and the number of tweets. Random forest (RF), K-Nearest neighbor (KNN), Extreme Gradient Boosting (XGBoost-XGB), Support vector machine (SVM) Naive Bayes (NB), Artificial Neural Networks (ANN), and Long-Short-Term Memory (LSTM) networks were applied on the selected features for prediction purposes. This work investigates two main research questions: i. How does the sample size affect the prediction performance of ML algorithms? ii. How does the data type affect the prediction performance of ML algorithms? Accuracy and area under the ROC curve (AUC) values were used to compare the model performance. A t-test was performed to test the statistical significance of the prediction results.

Complexity vs Empirical Score

  • Math Complexity: 2.5/10
  • Empirical Rigor: 6.0/10
  • Quadrant: Street Traders — empirical and applied, lower math complexity

Why this score: The paper employs standard machine learning algorithms and basic statistical tests (t-test) without advanced mathematical derivations, focusing on empirical evaluation of cryptocurrency prediction using real-world data sources like Google trends and tweets.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>How do data types & sample sizes<br>affect ML performance in<br>Bitcoin price prediction?"] --> B["Data Preparation"]
  B --> C["Input Features:<br>Technical Indicators, Google Trends, Tweets"]
  C --> D1["Data Type 1: Continuous<br>(1 & 7 Year Periods)"]
  C --> D2["Data Type 2: Trend<br>(Binarized Continuous Data)"]
  D1 & D2 --> E["ML Algorithms:<br>RF, KNN, XGBoost, SVM, NB, ANN, LSTM"]
  E --> F["Evaluation:<br>Accuracy & AUC Scores<br>+ Statistical Significance (t-test)"]
  F --> G["Key Outcomes:<br>1. Sample size significantly impacts performance<br>2. Trend data often outperforms continuous data<br>3. Tree-based models (RF, XGBoost) show high robustness"]

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 comprehensive methodology that constructs continuous and trend datasets over one- and seven-year periods and applies various machine learning algorithms to forecast Bitcoin price movements.

Source sentence
This study provided a comprehensive methodology, including constructing continuous and trend data using one and seven years periods of data as inputs and applying machine learning (ML) algorithms to forecast Bitcoin price movement.

Assumes

No explicit assumptions found in the text.

Data

  • Bitcoin price data crypto-onchain unknown
    one and seven years
    Source sentence
    constructing continuous and trend data using one and seven years periods of data as inputs
  • Technical indicators alternative unknown
    Source sentence
    the input features are determined as technical indicators, google trends, and the number of tweets
  • Bitcoin tweets text-news unknown
    Source sentence
    the number of daily tweets on Bitcoin

Recognised sources: Yahoo Finance

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • The LSTM model achieved the highest accuracy on all trend data, reaching 0.9593. performance
    Source sentence
    The best performance in all trend data was obtained with the LSTM machine learning method
  • Naive Bayes performed the worst on trend data with an accuracy of 0.9283. negative
    Source sentence
    Naive Bayes method provided the lowest performance with 0.9283.
  • On continuous all data, the ANN model achieved the highest accuracy of 0.8046. performance
    Source sentence
    Neural Networks (ANN) machine learning method was obtained with 0.8046 accuracy
  • Naive Bayes showed the lowest accuracy on continuous all data at 0.6110. negative
    Source sentence
    Naive Bayes machine learning method showed the lowest performance with an accuracy rate of 0.6110.
  • Trend datasets improved overall prediction performance across all models compared to continuous datasets. predictive-accuracy
    Source sentence
    the trend datasets improved the overall prediction performance in all estimated models.

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.