Paper: arXiv 2411.13586
Authors: Rahul Arulkumaran, Suyash Kumar, Shikha Tomar, Manideep Gongalla, Harshitha
Abstract
Cryptocurrencies are highly volatile financial instruments with more and more new retail investors joining the scene with each passing day. Bitcoin has always proved to determine in which way the rest of the cryptocurrency market is headed towards. As of today Bitcoin has a market dominance of close to 50 percent. Bull and bear phases in cryptocurrencies are determined based on the performance of Bitcoin over the 50 Day and 200 Day Moving Averages. The aim of this paper is to foretell the performance of bitcoin in the near future by employing predictive algorithms. This predicted data will then be used to calculate the 50 Day and 200 Day Moving Averages and subsequently plotted to establish the potential bull and bear phases.
Complexity vs Empirical Score
- Math Complexity: 3.5/10
- Empirical Rigor: 6.0/10
- Quadrant: Street Traders — empirical and applied, lower math complexity
Why this score: The paper uses basic ML techniques (linear regression, LSTM) and standard technical indicators, resulting in low mathematical complexity. However, it details data collection, preprocessing, and model validation on real cryptocurrency data, indicating a practical, implementation-focused approach.
Research Flowchart
flowchart TD A["Research Goal: Predict<br>Bitcoin Market Phases"] --> B["Data Input:<br>Historical Bitcoin Price Data"] B --> C["Computational Process:<br>Apply Predictive Algorithms"] C --> D["Generate Forecast:<br>Future Bitcoin Price Data"] D --> E["Compute Technical Indicators:<br>50 & 200 Day Moving Averages"] E --> F["Define Market Phases:<br>Bull & Bear Phase Detection"] F --> G["Key Outcome:<br>Advance Detection of<br>Bull and Bear Markets"]
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 to predict near‑future Bitcoin prices with machine‑learning models (MLR and LSTM) and use those predictions to compute 50‑day and 200‑day moving averages for advance detection of bull and bear phases.
Source sentence
The aim of this paper is to foretell the performance of bitcoin in the near future by employing predictive algorithms.
Assumes 4
- Bitcoin price movements can be captured by technical indicators derived from OHLCV data. data
Source sentence
The dataset did not have all the features required like technical indicators. However, the OHLCV data was sufficient to manually calculate the technical indicators required.
- Future closing prices for the next 21 days are available for training the models. data
Source sentence
prices of 21 days in the future were also added to each and as a result 21 new columns were created for this purpose.
- The model can be evaluated by visual comparison of predicted and actual moving‑average curves rather than statistical fit metrics. model
Source sentence
Instead, the authors decided to plot and check the predicted moving average graphs and see how closely related they are to the actual moving average graphs during that duration and whether they followed trends or not.
- R² is not a suitable accuracy metric because technical indicators are highly correlated. statistical
Source sentence
The authors chose not to use R2 as their accuracy indicator because most of the technical indicators are derived from OHLC price points and, therefore, were highly correlated.
Data
- Bitcoin OHLCV daily data market-prices public
Source sentence
Data for Bitcoin was gathered from an open API where the Open, High, Low, Close and Volume data were available for each day from 1st January 2012.
Costs
Not applicable
Finds 2
- LSTM models were found to better capture time‑sequence relationships and achieve greater success in detecting market phases than MLR. performance
Source sentence
Here, LSTMs have been used to extract functionality from time‑sequences, with the hypothesis and expectation that more intricate functional relationships between combinatorial patterns can be extracted, leading to achieving greater success in the above objective.
- Visual comparison of predicted and actual 50‑day and 200‑day SMAs showed that the models could follow the true moving‑average curves, mitigating over‑fitting concerns. predictive-accuracy
Source sentence
The authors plotted the actual 50day SMA, 200day SMA and predicted 50day SMA, 200day SMA to check how was the accuracy of the model and if the predicted values were following the same curve as the actual values.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 14 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.