metadata
tags:
- crypto-prediction
- time-series
- bert
license: apache-2.0
datasets:
- custom-crypto-news
metrics:
- f1-score
model-index:
- name: crypto-trend-predictor
results:
- task:
type: text-classification
dataset:
name: custom-crypto-news
type: custom
metrics:
- name: F1-Score
type: f1
value: 0.85
Crypto Trend Predictor
Overview
This BERT-based model predicts cryptocurrency market trends (bearish, bullish, or neutral) based on news articles, tweets, or market summaries. It was fine-tuned on a dataset of historical crypto news and price movements.
Model Architecture
- Base Model: BERT-base-uncased
- Layers: 12
- Hidden Size: 768
- Attention Heads: 12
- Fine-tuned for multi-class classification (bearish/bullish/neutral)
Intended Use
Ideal for analyzing crypto-related text to forecast short-term market trends, assisting traders or analysts in decision-making.
Limitations
- Predictions are based on text sentiment and may not account for external factors like regulations or economic events.
- Performance degrades on non-English text or highly technical jargon not seen in training.
- Not financial advice; use at your own risk.
Example Code
from transformers import pipeline
predictor = pipeline("text-classification", model="user/crypto-trend-predictor")
result = predictor("Bitcoin surges after ETF approval.")
print(result)
# [{'label': 'BULLISH', 'score': 0.95}]