https://github.com/happydasch/bt_llm_advisory
Extension of LLM Advisory for Backtrader strategies, enabling real-time trading insights from multiple LLM-based advisors directly within strategy execution.
https://github.com/happydasch/bt_llm_advisory
adivsory ai-agent ai-trading backtrader candle-patterns langchain llm strategy technical-analysis trading trend-analysis
Last synced: 3 months ago
JSON representation
Extension of LLM Advisory for Backtrader strategies, enabling real-time trading insights from multiple LLM-based advisors directly within strategy execution.
- Host: GitHub
- URL: https://github.com/happydasch/bt_llm_advisory
- Owner: happydasch
- Created: 2025-05-03T16:16:39.000Z (5 months ago)
- Default Branch: master
- Last Pushed: 2025-06-25T10:02:55.000Z (3 months ago)
- Last Synced: 2025-06-25T10:43:30.124Z (3 months ago)
- Topics: adivsory, ai-agent, ai-trading, backtrader, candle-patterns, langchain, llm, strategy, technical-analysis, trading, trend-analysis
- Language: Python
- Homepage:
- Size: 119 KB
- Stars: 4
- Watchers: 3
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Backtrader LLM Advisory
`Backtrader LLM Advisory` is using `LLM Advisory` for providing advisors to backtrader strategies. It allows the strategy to access different "advisors" which have access to the strategy and its current state.
inside the strategy init call `llm_advisory.init(self)` from next call `llm_advisory.get_advice()`
- Generates automatically data from a strategy and uses this for advisory data
- Control lookback period of data to use
## Usecase scenarios
- Confirm trend, identify sideways trends using the trend advisor
- Interpret current indicator values by the technical analysis advisor
- Get trading signals by asking the technical analysis advisor to check current indicator values
- Identify candle patterns by using the candle pattern advisor
- Confirm a signal from your strategy code by advisory
- Get feedback about your strategy and the current trading session
## Available Advisors
An overview of all available advisors with a short explaination.
### BacktraderStrategyAdvisor
Common advisor for strategies. Takes available information from strategy and provides an advice based on this data.
```python
from bt_llm_advisory.advisors import BacktraderStrategyAdvisor
strategy_advisor = BacktraderStrategyAdvisor()
```
### BacktraderTrendAdvisor
Returns a signal based on data from the current strategy. The signal is bullish for a up-trend, bearish for a down-trend, neutral for no-trend and none if unable to identify.
This advisor uses different indicators to identify the current trend.
```python
from bt_llm_advisory.advisors import BacktraderTrendAdvisor
trend_advisor = BacktraderTrendAdvisor(
short_ma_period: int = 10, # period for short moving average
long_ma_period: int = 25, # period for long moving average
lookback_period: int = 5, # lookback period for trend data
add_all_data_feeds: bool = False, # should all data feeds be included
)
```
### BacktraderTechnicalAnalysisAdvisor
Returns a signal based on a technical analysis of the strategy indicators.
```python
from bt_llm_advisory.advisors import BacktraderTechnicalAnalysisAdvisor
technial_analysis_advisor = BacktraderTechnicalAnalysisAdvisor()
```
### BacktraderCandlePatternAdvisor
Returns a signal from OHLC candles if it recognizes a candlestick pattern.
```python
from bt_llm_advisory.advisors import BacktraderCandlePatternAdvisor
candle_pattern_advisor = BacktraderCandlePatternAdvisor(
lookback_period: int = 5, # lookback period for ohlc data
add_all_data_feeds: bool = False, # should all data feeds be included
)
```
### BacktraderFeedbackAdvisor
Provides feedback about the strategies data.
```python
from bt_llm_advisory.advisors import BacktraderFeedbackAdvisor
feedback_advisor = BacktraderFeedbackAdvisor()
```
### BacktraderPersonaAdvisor
Strategy trading advisor with additional defined personaily or more detailed instructions.
```python
from bt_llm_advisory.advisors import BacktraderPersonaAdvisor
persona_advisor = BacktraderPersonaAdvisor(
# name of the persona, can also be a famous person
person_name: str,
# description of the personality, which also can contain informations about needed knowledge
personality: str
)
```
## Examples
## Frequently Asked Questions
TODO
- add possibility to instruct another llm to create / update the used strategy based on advisors input
- Update code of strategy by asking a llm for a change
- Trade based on advisors advices