crypto-sentiment-lstm

ETH Sentiment Model

Predict next-day ETH sentiment (Bearish=0, Neutral=1, Bullish=2) with confidence score.

Install locally

python -m pip install git+https://huggingface.co/jtassos2025/crypto-sentiment-lstm

Quick start example

from huggingface_hub import hf_hub_download
from eth_sentiment import EthSentimentModel

daily_model_path = hf_hub_download("jtassos2025/crypto-sentiment-lstm", "eth_lstm_sentiment_model.pth")
daily_scaler_path = hf_hub_download("jtassos2025/crypto-sentiment-lstm", "eth_scaler.pkl")
weekly_model_path = hf_hub_download("jtassos2025/crypto-sentiment-lstm", "eth_lstm_weekly_model.pth")
weekly_scaler_path = hf_hub_download("jtassos2025/crypto-sentiment-lstm", "eth_weekly_scaler.pkl")
csv_path = hf_hub_download("jtassos2025/crypto-sentiment-lstm", "eth_combined_updated.csv")

import pandas as pd

csv_path = hf_hub_download("jtassos2025/crypto-sentiment-lstm", "eth_combined_updated.csv")
df = pd.read_csv(csv_path)
# print(df.dtypes)
# print(df.head())

# Daily
model_daily = EthSentimentModel(daily_model_path, daily_scaler_path, csv_path)
result_daily = model_daily.predict({
    "Start": "2025-10-07",
    "End": "2025-10-08",
    "Open": 4687.59,
    "High": 4759.16,
    "Low": 4432.47,
    "Close": 4449.8,
    "Volume": 251172.82,
    "Market Cap": 1117668830.61,
    "Source": "Polygon"
})
print("Daily:", result_daily)

# Weekly
model_weekly = EthSentimentModel(weekly_model_path, weekly_scaler_path, csv_path, seq_len=26, mode="weekly")
result_weekly = model_weekly.predict({
    "Start": "2025-10-01",
    "End": "2025-10-07",
    "Open": 4687.59,
    "High": 4759.16,
    "Low": 4432.47,
    "Close": 4449.8,
    "Volume": 251172.82,
    "Market Cap": 1117668830.61,
    "Source": "Polygon"
})
print("Weekly:", result_weekly)

output: {"sentiment": Bullish, "score": 0.53}
sentiment: Bearish, Neutral, or Bullish
score: model confidence for predicted class (0–1)