model-tester / src /data_loader.py
Arkm20's picture
Create data_loader.py
ef957e4 verified
"""
data_loader.py — yfinance market data download with batching, caching,
and granular progress callbacks for the UI progress bar.
"""
import gc
import time
import logging
import hashlib
import pickle
from datetime import datetime, timedelta
from pathlib import Path
from typing import Callable, Optional
import numpy as np
import pandas as pd
logger = logging.getLogger("SniperData")
DATA_CACHE_DIR = Path("/tmp/sniper_data_cache")
DATA_CACHE_DIR.mkdir(parents=True, exist_ok=True)
BATCH_SIZE = 50
MAX_RETRIES = 3
RETRY_DELAY = 5
# ---------------------------------------------------------------------------
# Cache helpers
# ---------------------------------------------------------------------------
def _cache_key(tickers: list, start: str, end: str) -> str:
key = f"{'|'.join(sorted(tickers))}|{start}|{end}"
return hashlib.md5(key.encode()).hexdigest()[:16]
def _cache_path(ticker: str, start: str, end: str) -> Path:
safe = ticker.replace("^", "IDX_")
key = hashlib.md5(f"{ticker}{start}{end}".encode()).hexdigest()[:8]
return DATA_CACHE_DIR / f"{safe}_{key}.pkl"
def _load_cached(ticker: str, start: str, end: str) -> Optional[pd.DataFrame]:
p = _cache_path(ticker, start, end)
if p.exists():
try:
with open(p, "rb") as fh:
return pickle.load(fh)
except Exception:
p.unlink(missing_ok=True)
return None
def _save_cached(ticker: str, start: str, end: str, df: pd.DataFrame):
p = _cache_path(ticker, start, end)
try:
with open(p, "wb") as fh:
pickle.dump(df, fh)
except Exception as e:
logger.warning(f"Cache write failed for {ticker}: {e}")
# ---------------------------------------------------------------------------
# Download
# ---------------------------------------------------------------------------
def download_ticker_batch(
tickers: list[str],
start: str,
end: str,
progress_cb: Callable = None,
) -> dict[str, pd.DataFrame]:
"""
Download OHLCV data for a list of tickers (+ ^VIX, ^GSPC) in batches of 50.
Returns dict {ticker: DataFrame}.
Calls progress_cb(message, fraction_done) at each batch.
"""
import yfinance as yf
all_tickers = list(dict.fromkeys(tickers + ["^VIX", "^GSPC"]))
total = len(all_tickers)
results: dict[str, pd.DataFrame] = {}
to_download = []
# Check cache first
for ticker in all_tickers:
cached = _load_cached(ticker, start, end)
if cached is not None:
results[ticker] = cached
else:
to_download.append(ticker)
cached_count = total - len(to_download)
if cached_count > 0 and progress_cb:
progress_cb(f"Loaded {cached_count} tickers from cache. Downloading {len(to_download)} new...", 0.02)
n_batches = max(1, (len(to_download) + BATCH_SIZE - 1) // BATCH_SIZE)
for batch_i, batch_start in enumerate(range(0, len(to_download), BATCH_SIZE)):
batch = to_download[batch_start: batch_start + BATCH_SIZE]
frac = batch_start / max(1, len(to_download))
if progress_cb:
progress_cb(
f"Downloading batch {batch_i + 1}/{n_batches} ({len(batch)} tickers)...",
0.05 + 0.30 * frac,
)
for attempt in range(MAX_RETRIES):
try:
raw = yf.download(
batch,
start=start,
end=end,
group_by="ticker",
auto_adjust=True,
threads=True,
progress=False,
)
for ticker in batch:
try:
df = (raw[ticker] if len(batch) > 1 else raw).copy()
df = df.dropna(subset=["Close"])
if len(df) > 0:
df.index = pd.to_datetime(df.index)
results[ticker] = df
_save_cached(ticker, start, end, df)
except Exception:
pass
break
except Exception as e:
logger.warning(f"Batch {batch_i+1} attempt {attempt+1} failed: {e}")
if attempt < MAX_RETRIES - 1:
time.sleep(RETRY_DELAY * (attempt + 1))
if progress_cb:
progress_cb(f"Download complete. {len(results)} tickers loaded.", 0.35)
gc.collect()
return results
# ---------------------------------------------------------------------------
# Quality filter
# ---------------------------------------------------------------------------
def filter_ticker_data(
ticker_data: dict[str, pd.DataFrame],
min_history_days: int = 252,
min_dollar_volume: float = 2_000_000,
progress_cb: Callable = None,
) -> dict[str, pd.DataFrame]:
"""Remove tickers with insufficient history or low liquidity."""
valid = {}
skipped = []
for ticker, df in ticker_data.items():
if ticker.startswith("^"):
valid[ticker] = df
continue
if len(df) < min_history_days:
skipped.append((ticker, f"only {len(df)} days"))
continue
try:
avg_dv = (df["Close"] * df["Volume"]).rolling(20).mean().dropna()
if len(avg_dv) == 0 or avg_dv.iloc[-1] < min_dollar_volume:
skipped.append((ticker, "low dollar volume"))
continue
except Exception:
skipped.append((ticker, "volume check failed"))
continue
valid[ticker] = df
if progress_cb and skipped:
progress_cb(f"Filtered: {len(valid)} tickers kept, {len(skipped)} skipped.", 0.37)
return valid
# ---------------------------------------------------------------------------
# Extract VIX / SPX convenience helpers
# ---------------------------------------------------------------------------
def extract_market_series(ticker_data: dict) -> tuple:
"""Returns (vix_close_series, sp500_close_series) or (None, None)."""
vix = ticker_data.get("^VIX")
spx = ticker_data.get("^GSPC")
vix_s = vix["Close"] if vix is not None and not vix.empty else None
spx_s = spx["Close"] if spx is not None and not spx.empty else None
return vix_s, spx_s
# ---------------------------------------------------------------------------
# Regime detection (for live routing in backtester)
# ---------------------------------------------------------------------------
def get_current_regime(date, sp500_data: pd.Series, vix_data: pd.Series,
sma_period: int = 200, vix_threshold: float = 20.0) -> tuple[int, int]:
"""
Returns (market_regime, vix_regime) as (0/1, 0/1).
market_regime: 1=bull (above 200 SMA), 0=bear
vix_regime: 1=high VIX, 0=low VIX
"""
try:
sp_sma = sp500_data.rolling(sma_period).mean()
sp_idx = sp500_data.index.get_indexer([date], method="ffill")[0]
mkt = 1 if sp_idx >= 0 and sp500_data.iloc[sp_idx] > sp_sma.iloc[sp_idx] else 0
except Exception:
mkt = 1
try:
vix_idx = vix_data.index.get_indexer([date], method="ffill")[0]
vix = 1 if vix_idx >= 0 and vix_data.iloc[vix_idx] > vix_threshold else 0
except Exception:
vix = 0
return mkt, vix