Skip to main content

​
Overview

The Dataloader Server loads OHLCV data, computes TA-Lib indicators, and persists derived tables. Use it as the market-data and indicator-calculation entry point.

​
Connection

Add this server to your MCP client configuration. 
{
  "mcpServers": {
    "dataloader": {
      "url": "https://market-data-loader-production.up.railway.app/mcp/dataloader"
    }
  }
}
Restart the client after changing MCP configuration. The server tools appear automatically after the connection is established.

​
Transport

PropertyValue
ProtocolMCP over Streamable HTTP
MCP URLhttps://market-data-loader-production.up.railway.app/mcp/dataloader
Health URLhttps://market-data-loader-production.up.railway.app/health
MCP path/mcp/dataloader
Request envelope{"request": {...}}
AuthOptional bearer token when enabled for the endpoint

​
Best-Fit Workflows

  • Load OHLCV data for one or more tickers.
  • Compute TA-Lib indicators on tabular OHLCV payloads.
  • Inspect supported indicators.
  • Export CSV, Parquet, or pandas pickle outputs.
  • fetch-ohlcvs
  • compute-indicators
  • list-supported-indicators
  • save-dataframe

​
Tools

ToolDescriptionReturns
list-supported-indicatorsReturns a sorted list of indicator names supported by TA-Lib, enabling developers to see which technical analysis functions are available. Params: nonelist[str]
compute-indicatorsCalculates TA-Lib technical indicators from OHLCV data, supporting single or multi-ticker inputs with optional parameter overrides. Params: df: OHLCV DataFrame; indicators: list of TA-Lib indicator names; params: dict of indicator-specific parameters (optional); ticker: prefix to filter multi-ticker data (optional)dict[str, Any]
fetch-ohlcvsLoads OHLCV data for one or multiple tickers from Yahoo or other sources, returning a combined DataFrame and metadata. This is the retrieval step, not the file-export step. If the user wants a local CSV/Parquet file, follow with save-dataframe instead of manually writing rows. Params: Inputs: tickers (list of symbols), timeframe (e.g., ‘1d’, ‘1h’, ‘5m’), start_date, end_date, source (optional), interval (legacy alias). Returns dict with ohlcv_df, ticker_details, row_count, columns.dict[str, pd.DataFrame | list[dict[str, str | list[str]]] | int | list[str]]
save-dataframePersist a dataframe to disk and return the absolute file path. This is the primary tool for user requests like ‘download this data’, ‘save to Downloads’, or ‘export to csv/parquet’. Prefer it over fetching rows and manually writing a file. ticker and dataset_name are required identifiers and must be provided explicitly; they are not inferred from the dataframe payload. Supports CSV, Parquet, and pandas pickle formats, with automatic filename generation based on ticker, timeframe, and date range. Params: Inputs: dataframe (Any), ticker (str), dataset_name (str), timeframe (str|None), start_date (str|None), end_date (str|None), file_format (str, default=‘csv’), data_kind (str, default=‘auto’), interval (str|None). Example: timeframe=‘1h’, file_format=‘parquet’, data_kind=‘ohlcv’.str

​
Examples

​
Load OHLCV data

{
  "request": {
    "tickers": [
      "AAPL",
      "MSFT"
    ],
    "timeframe": "1d",
    "start_date": "2024-01-01",
    "end_date": "2024-03-31",
    "source": "yahoo"
  }
}

​
Compute indicators

{
  "request": {
    "df": "$last",
    "indicators": [
      "SMA",
      "RSI"
    ],
    "params": {
      "SMA": {
        "timeperiod": 20
      },
      "RSI": {
        "timeperiod": 14
      }
    }
  }
}

​
Save a dataframe

{
  "request": {
    "dataframe": "$last",
    "ticker": "MSFT",
    "dataset_name": "ta_features",
    "timeframe": "1d",
    "file_format": "csv"
  }
}

​
Notes

  • If no data source is specified and the premium provider key is unavailable, the effective source falls back to Yahoo.
  • Large tabular outputs can be returned as result handles; use shared result/artifact tools when that happens.

Client setup

Configure this endpoint in Cursor, Claude Desktop, or a generic MCP client.

Shared tools

Use health, result, artifact, environment, and table helper tools.

​
Other Servers

Fama-French Replicate

Official and replicated Fama-French factors plus loadings and alpha estimation.

Statistical Factor Models

Stock-Watson, complete-panel, and dynamic statistical factor extraction.

Jump Models

JumpModel and SparseJumpModel regime fitting, online prediction, and backtesting.

Wavelet Mean Reversion

Wavelet-based mean reversion analysis for financial time series.

Parallax ExtremeHurst

ExtremeHurst signal generation from OHLCV data.

EP Ratio Screener

Fundamental stock screening based on earnings yield and balance-sheet quality.

Volatility Scaling Lab

Volatility targeting, EWMA volatility, Monte Carlo bands, and risk diagnostics.