Investigating the effectiveness of Twitter sentiment in cryptocurrency close price prediction by using deep learning
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Abstract In recent years, cryptocurrencies' price prediction has attracted the interest of many people including investors, researchers and practitioners. In this study, we proposed a hybrid model for predicting the daily close price of cryptocurrencies based on different neural networks such as long short‐term memory, convolutional neural network and attention mechanism. Using an ensemble of three pre‐trained language models, we extracted sentiment of cryptocurrency‐related tweets posted between 1 January 2021 and 31 December 2021. We constructed 20 different versions of our model and evaluated their performance on data of 27 most traded cryptocurrencies using a history of previous days' sentiment data along with close prices as input data. The flexible input layer of our model enables different ways of feeding data into the model to adjust it for different cryptocurrencies to obtain better predictions. Our analysis revealed several important findings. We showed that longer sequences of input data achieve most accurate predictions on average. More specifically, using a history of 14‐ and 21‐days' data results in lowest RMSE values on average compared to using a history of 7 days. However, there is no significant difference between the results related to the input sequences with lengths of 14 and 21. In addition, our findings suggest that sentiment data can be useful in predicting prices for more than 70% of the studied cryptocurrencies. Thus, peoples' emotions, opinions, and sentiment that are expressed through their posts on Twitter platform play a significant role in prediction of cryptocurrencies' prices.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.024 | 0.011 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it