LONA Trading Assistant

Creates a new trading strategy from existing Backtrader Python code.

Code requirements:

• The code imports backtrader: import backtrader as bt
• Defines a class inheriting from bt.Strategy
• Parameters use tuple format: params = (('name', value), ...)
• Indicators are initialized in __init__, trading logic in next()

Example:

import backtrader as bt

class Strategy(bt.Strategy):
    params = (('period', 20),)

    def __init__(self):
        self.sma = bt.indicators.SMA(period=self.p.period)

    def next(self):
        if self.data.close[0] > self.sma[0]:
            self.buy()

Creates a new trading strategy from a natural language description. An AI pipeline generates Backtrader Python code based on the description.

This is an asynchronous operation. Returns a jobId immediately. The generation runs in the background (typically 3-5 minutes). Progress can be checked with lona_get_strategy_creation_status.

Pipeline stages: PENDING -> GENERATING -> SAVING -> COMPLETED (or FAILED at any point).

Example descriptions:

• "A momentum strategy that buys when RSI crosses above 30 and sells below 70"
• "Trend following using 20 and 50 period moving averages with 2% stop loss"
• "Mean reversion with Bollinger Bands - buy at lower band, sell at upper band"

Downloads cryptocurrency market data from supported exchanges and saves it as a new symbol in the user's Lona account. Returns a symbol ID usable as a data_id in lona_run_backtest.

Supported intervals: 1m, 5m, 15m, 30m, 1h, 4h, 1d

Practical data limits:

• 1m data: up to ~6 months
• 5m data: up to ~2 years
• 1h+ data: several years of history

Retrieves comprehensive backtest results by ID, including detailed metrics, trade history, and per-symbol performance breakdown.

Retrieves a backtest report summary by ID. Returns key performance metrics including total return, Sharpe ratio, maximum drawdown, win rate, and number of trades.

Retrieves chart data for a completed backtest report. Returns an interactive candlestick chart with buy/sell markers showing trade entry and exit points.

Checks the execution status of a backtest report. Returns current status (PENDING, EXECUTING, PROCESSING, COMPLETED, or FAILED) and whether the backtest is complete.

Retrieves metadata for a single trading strategy by ID. Returns the strategy name, description, version, language, and timestamps.

Retrieves the Python source code of a trading strategy by ID. Returns the Backtrader code that implements the strategy logic.

Checks the progress of an async strategy creation job started by lona_create_strategy_from_description.

Returns current status: PENDING, GENERATING, SAVING, COMPLETED, or FAILED. When COMPLETED, includes the strategyId, generated code, explanation, and a review score. When FAILED, includes error details and suggestions.

Retrieves metadata for a single market data symbol by ID. Returns the symbol name, description, data type, and whether it is a global or user-uploaded dataset. The metadata field may include exchange, asset_class, frequency, quote_currency, and timeframe availability.

Retrieves OHLCV (Open, High, Low, Close, Volume) time-series price data for a symbol by ID.

Lists the user's backtest reports with status, performance metrics (return, Sharpe ratio, trades), and timestamps. Supports filtering by strategy_id or status (PENDING, EXECUTING, PROCESSING, COMPLETED, FAILED) and pagination via skip/limit.

Lists the user's saved trading strategies. Returns strategy names, IDs, descriptions, and version info. Supports pagination via skip/limit parameters.

Lists available market data symbols. Returns user-uploaded data files by default, or pre-loaded global datasets (US equities, crypto, forex) when is_global=true. Each symbol has an ID usable as a data_id in lona_run_backtest.

Provisions a session token for API-key mode authentication. Not needed for OAuth-authenticated sessions (e.g., ChatGPT users). Returns a short-lived token for subsequent tool calls.

Executes a trading strategy backtest against historical market data. Runs asynchronously and returns a report_id for checking progress with lona_get_report_status.

Required inputs:

• strategy_id: a Lona strategy ID (from lona_list_strategies, lona_create_strategy, or lona_create_strategy_from_description)
• data_ids: array of Lona symbol IDs (from lona_list_symbols, lona_download_market_data, or lona_upload_market_data)

Configurable parameters:

• initial_cash: Starting capital (default: 100000)
• commission: Trading commission as decimal (default: 0.001 = 0.1%)
• leverage: Maximum leverage allowed (default: 1)
• parameters: Override strategy parameters (passed to the strategy's params tuple)

Modifies an existing trading strategy. Accepts updates to the name, description, or code individually or all at once.

This operation uses immutable versioning: the update creates a new version of the strategy and returns a new strategy ID. The previous version is replaced in listings. The returned strategyId represents the latest version.

Uploads user-provided CSV market data and creates a new symbol in the user's Lona account. Returns a symbol_id usable as a data_id in lona_run_backtest.

Accepts data via the file parameter (when user attaches a CSV) or the csv_content parameter (raw CSV string).

This tool accepts OHLCV (Open-High-Low-Close-Volume) price data in CSV format.

CSV Requirements:

• Header row with column names
• Minimum 10 data rows
• Maximum ~15MB
• Must contain at minimum: a timestamp/date column and a close price column
• Additional columns (open, high, low, volume) are optional but recommended

Supported timestamp formats (auto-detected):

• ISO 8601: 2024-01-15T00:00:00Z
• Date-time: 2024-01-15 00:00:00
• Date only: 2024-01-15
• Slash format: 01/15/2024 or 15/01/2024 (auto-detected)
• Unix timestamps: 1705276800 (seconds), 1705276800000 (ms)

Standard column names (auto-detected, case-insensitive):

• Timestamp: "timestamp", "date", "time", "datetime"
• Price: "open", "high", "low", "close" (or "adj close", "price")
• Volume: "volume", "vol"

If your CSV uses non-standard column names, provide the column_mapping parameter.