store

name	store
description	Use when tasks need local Parquet market data storage, pool management, research artifacts, backtest records, or model metadata.

Store

Use this skill when a task needs local file management for research data and artifacts. The public project uses DataManager as the storage boundary.

DataManager

from skills.store.data_manager import DataManager, DataQualityReport, validate_ohlcv

DataManager resolves its root from QUANTSPACE_DATA_ROOT, falling back to the repository data/ directory.

Supported Layout

• data/market/{frequency}/{symbol}.parquet
• data/adj_factor/{symbol}.parquet
• data/pools/{pool_id}.json
• data/factors/{pool_id}/
• data/factor_test/{pool_id}/
• data/correlation/
• data/backtest/
• data/models/
• data/export/

Main Methods

Market data

• read_symbol(symbol, frequency="1d")
• read_symbols(symbols, frequency="1d")
• save_symbol(symbol, df, frequency="1d", source="unknown")
• import_symbol_csv(csv_path, symbol, frequency="1d")
• import_combined_csv(csv_path, frequency="1d")
• list_symbols(frequency="1d")

Pools

• create_pool(pool_id, symbols, description="", frequency="1d")
• get_pool_symbols(pool_id)
• get_pool_frequency(pool_id)
• load_pool_data(pool_id)
• check_pool_coverage(pool_id)
• list_pools()

Research artifacts

• save_factor, read_factor
• save_factor_test, read_factor_test_summary
• save_factor_correlation, read_factor_correlation
• save_backtest_run, read_backtest_summary, read_backtest_run
• list_models, read_model_metadata

Recipes

Save PandaData bars

import pandas as pd

from skills.ingest import PandaDataClient
from skills.store.data_manager import DataManager

raw = PandaDataClient().fetch_market_data("SHSE.600000", "20230101", "20231231")
bars = raw.copy()
bars["eob"] = pd.to_datetime(bars["date"])
bars = bars.set_index("eob")[["open", "high", "low", "close", "volume"]].sort_index()

DataManager().save_symbol("SHSE.600000", bars, frequency="1d", source="panda_data")

Load a panel

from skills.store.data_manager import DataManager

dm = DataManager()
panel = dm.load_pool_data("sample_etf_rotation")

Load explicit symbols

from skills.store.data_manager import DataManager

dm = DataManager()
panel = dm.read_symbols(["CFFEX.IF99", "SHFE.CU99"], frequency="1d")

read_symbols returns a MultiIndex (symbol, eob) panel and reports all missing symbols in one FileNotFoundError.