config.py

"""Configuration constants for Stock Scanner."""

import os

import pytz

# HF Dataset repo for remote data
HF_DATASET_REPO = os.getenv("HF_DATASET_REPO", "Arrechenash/stocks")

# Direct HF CDN URL for DuckDB (public datasets, no auth needed)
MERGED_DATA = f"https://huggingface.co/datasets/{HF_DATASET_REPO}/resolve/main/alpaca_merged.parquet"

# Local paths (small files, kept in repo)
CACHE_DIR = "data/cache"
RESULTS_DIR = "results"
CALENDAR_DIR = "data/market_calendar.csv"
CORPORATE_ACTIONS_DIR = "data/corporate_actions.csv"
METADATA_DIR = "data/symbol_metadata.csv"
DAILY_DATA = "data/raw/daily_data.parquet"
COMPUTED_DATA = "data/computed/computed_data.parquet"
METADATA_DB = "data/metadata.db"

# Timezone
NY = pytz.timezone("America/New_York")

# Default backtest parameters
DEFAULT_INITIAL_CAPITAL = 100
DEFAULT_COMMISSION_PER_SHARE = 0.005
DEFAULT_SLIPPAGE_PER_SHARE = 0.01
DEFAULT_SHARES_PER_UNIT = 1

# Delay when fetching Alpaca SIP data to avoid free-tier subscription errors
ALPACA_SIP_DELAY_MINUTES = 21