THULab/stock_dataset

S&P 500 Stock Market (TsFile)

This dataset is an Apache TsFile conversion of Adilbai/stock-dataset.

Modalities: Time-series.

Overview

• Daily observations of S&P 500 companies with 73 engineered technical features over 5 years (620,095 rows).

• OHLCV plus moving averages, RSI, and other technical indicators.

• Each ticker is a device identified by the Ticker TAG.

• Converted observations: 620,095 rows across 1 TsFile file(s)

• Source format: parquet

TsFile schema

• Time — source Date (datetime), converted to INT64 milliseconds.

Column	Role	Type	Meaning
Time	TIME	INT64 (ms)	sample timestamp
Ticker	TAG	STRING	ticker symbol
Open	FIELD	FLOAT	open
High	FIELD	FLOAT	high
Low	FIELD	FLOAT	low
Close	FIELD	FLOAT	close
Volume	FIELD	FLOAT	volume
Dividends	FIELD	FLOAT	—
Stock_Splits	FIELD	FLOAT	—
SMA_5	FIELD	FLOAT	—
SMA_10	FIELD	FLOAT	—
SMA_20	FIELD	FLOAT	—
SMA_50	FIELD	FLOAT	—
EMA_12	FIELD	FLOAT	—
EMA_26	FIELD	FLOAT	—
MACD	FIELD	FLOAT	—
MACD_Signal	FIELD	FLOAT	—
MACD_Histogram	FIELD	FLOAT	—
RSI	FIELD	FLOAT	—
BB_Middle	FIELD	FLOAT	—
BB_Upper	FIELD	FLOAT	—
BB_Lower	FIELD	FLOAT	—
BB_Width	FIELD	FLOAT	—
BB_Position	FIELD	FLOAT	—
Volatility	FIELD	FLOAT	—
Price_Change	FIELD	FLOAT	—
Price_Change_5d	FIELD	FLOAT	—
High_Low_Ratio	FIELD	FLOAT	—
Open_Close_Ratio	FIELD	FLOAT	—
Volume_SMA	FIELD	FLOAT	—
Volume_Ratio	FIELD	FLOAT	—
Close_lag_1	FIELD	FLOAT	—
Close_lag_2	FIELD	FLOAT	—
Close_lag_3	FIELD	FLOAT	—
Close_lag_5	FIELD	FLOAT	—
Close_lag_10	FIELD	FLOAT	—
Volume_lag_1	FIELD	FLOAT	—
Volume_lag_2	FIELD	FLOAT	—
Volume_lag_3	FIELD	FLOAT	—
Volume_lag_5	FIELD	FLOAT	—
Volume_lag_10	FIELD	FLOAT	—
Price_Change_lag_1	FIELD	FLOAT	—
Price_Change_lag_2	FIELD	FLOAT	—
Price_Change_lag_3	FIELD	FLOAT	—
Price_Change_lag_5	FIELD	FLOAT	—
Price_Change_lag_10	FIELD	FLOAT	—
RSI_lag_1	FIELD	FLOAT	—
RSI_lag_2	FIELD	FLOAT	—
RSI_lag_3	FIELD	FLOAT	—
RSI_lag_5	FIELD	FLOAT	—
RSI_lag_10	FIELD	FLOAT	—
MACD_lag_1	FIELD	FLOAT	—
MACD_lag_2	FIELD	FLOAT	—
MACD_lag_3	FIELD	FLOAT	—
MACD_lag_5	FIELD	FLOAT	—
MACD_lag_10	FIELD	FLOAT	—
Volatility_lag_1	FIELD	FLOAT	—
Volatility_lag_2	FIELD	FLOAT	—
Volatility_lag_3	FIELD	FLOAT	—
Volatility_lag_5	FIELD	FLOAT	—
Volatility_lag_10	FIELD	FLOAT	—
Future_Return_1d	FIELD	FLOAT	—
Future_Up_1d	FIELD	FLOAT	—
Future_Category_1d	FIELD	FLOAT	—
Future_Return_5d	FIELD	FLOAT	—
Future_Up_5d	FIELD	FLOAT	—
Future_Category_5d	FIELD	FLOAT	—
Future_Return_10d	FIELD	FLOAT	—
Future_Up_10d	FIELD	FLOAT	—
Future_Category_10d	FIELD	FLOAT	—
Future_Return_20d	FIELD	FLOAT	—
Future_Up_20d	FIELD	FLOAT	—
Future_Category_20d	FIELD	FLOAT	—

Conversion notes

• Ticker kept as TAG; 72 numeric features kept as FLOAT/INT64 FIELDs.

Source & license

• Original dataset: https://huggingface.co/datasets/Adilbai/stock-dataset
• Author / publisher: Adilbai
• License: mit

Usage

Install the Apache TsFile Python SDK (pip install tsfile) and read a converted file:

from pathlib import Path
from tsfile import TsFileReader

path = Path("stock_dataset.tsfile")
with TsFileReader(str(path)) as reader:
    schemas = reader.get_all_table_schemas()
    print("tables:", list(schemas))
    table_name = next(iter(schemas))
    table = schemas[table_name]
    columns = [column.get_column_name() for column in table.get_columns()]
    print("columns:", columns)
    field_names = [
        column.get_column_name()
        for column in table.get_columns()
        if column.get_column_name() not in {"Time", "time"}
    ]
    if field_names:
        with reader.query_table(table_name, field_names[:3], batch_size=1024) as result:
            batch = result.read_arrow_batch()
            if batch is not None:
                print(batch.to_pandas().head())