Quick Links
PatchTST Mean Reversion Crypto Scanner
This model uses the PatchTST architecture (A Time Series is Worth 64 Words) to identify high-probability mean reversion setups in cryptocurrency perpetual futures. Unlike XGBoost which looks at single snapshots, this model analyzes a 48-step sequence of market conditions leading up to a signal.
- Downloads last month
- 9
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
# Load model directly from transformers import AutoTokenizer, PatchTSTForClassification tokenizer = AutoTokenizer.from_pretrained("mnigr/patchtst-mean-reversion-scanner") model = PatchTSTForClassification.from_pretrained("mnigr/patchtst-mean-reversion-scanner", device_map="auto")