88 lines
4.2 KiB
Python
88 lines
4.2 KiB
Python
import re
|
|
|
|
from langchain_core.messages import ToolMessage
|
|
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
|
|
from tradingagents.agents.utils.quant_tools import get_quant_analysis
|
|
from tradingagents.agents.utils.agent_utils import build_instrument_context, get_language_instruction
|
|
|
|
|
|
def _looks_like_raw_tool_call(text: str) -> bool:
|
|
"""Return True if the LLM output a tool call as plain text instead of via the API."""
|
|
stripped = re.sub(r"^```+\w*\s*", "", text.strip())
|
|
return bool(re.search(r'"name"\s*:', stripped) and re.search(r'"arguments"\s*:', stripped))
|
|
|
|
|
|
def create_quant_analyst(llm):
|
|
|
|
def quant_analyst_node(state):
|
|
current_date = state["trade_date"]
|
|
ticker = state["company_of_interest"]
|
|
instrument_context = build_instrument_context(ticker)
|
|
|
|
tools = [get_quant_analysis]
|
|
|
|
system_message = (
|
|
"You are a quantitative analyst. Use the get_quant_analysis tool to retrieve "
|
|
"statistical metrics for the stock. Interpret the results thoroughly: "
|
|
"assess risk (annualised volatility, semideviation, tail risk via skewness and kurtosis), "
|
|
"evaluate the return distribution (Jarque-Bera normality test), "
|
|
"analyse the market relationship (beta, alpha, R², rolling correlation with SPY), "
|
|
"and identify any structural concerns (fat tails, high downside deviation). "
|
|
"Provide a structured Markdown report with a clear summary table and actionable insights."
|
|
+ get_language_instruction()
|
|
)
|
|
|
|
prompt = ChatPromptTemplate.from_messages(
|
|
[
|
|
(
|
|
"system",
|
|
"You are a helpful AI assistant, collaborating with other assistants."
|
|
" Use the provided tools to progress towards answering the question."
|
|
" If you are unable to fully answer, that's OK; another assistant with different tools"
|
|
" will help where you left off. Execute what you can to make progress."
|
|
" If you or any other assistant has the FINAL TRANSACTION PROPOSAL: **BUY/HOLD/SELL** or deliverable,"
|
|
" prefix your response with FINAL TRANSACTION PROPOSAL: **BUY/HOLD/SELL** so the team knows to stop."
|
|
" You have access to the following tools: {tool_names}.\n{system_message}"
|
|
" For your reference, the current date is {current_date}. {instrument_context}",
|
|
),
|
|
MessagesPlaceholder(variable_name="messages"),
|
|
]
|
|
)
|
|
|
|
prompt = prompt.partial(system_message=system_message)
|
|
prompt = prompt.partial(tool_names=", ".join([tool.name for tool in tools]))
|
|
prompt = prompt.partial(current_date=current_date)
|
|
prompt = prompt.partial(instrument_context=instrument_context)
|
|
|
|
chain = prompt | llm.bind_tools(tools)
|
|
result = chain.invoke(state["messages"])
|
|
|
|
report = ""
|
|
if len(result.tool_calls) == 0:
|
|
content = result.content if isinstance(result.content, str) else ""
|
|
if _looks_like_raw_tool_call(content):
|
|
# The model output the tool call as plain text instead of using the
|
|
# function-calling API (common with weaker/local models). Call the
|
|
# tool directly so the quant data is always populated.
|
|
try:
|
|
raw_data = get_quant_analysis.invoke(
|
|
{"ticker": ticker, "analysis_date": current_date}
|
|
)
|
|
except Exception as e:
|
|
raw_data = f"Quant analysis unavailable: {e}"
|
|
# Feed the raw data back to the LLM so it can write the narrative report.
|
|
tool_msg = ToolMessage(content=raw_data, tool_call_id="fallback")
|
|
followup = chain.invoke(state["messages"] + [result, tool_msg])
|
|
fc = followup.content if isinstance(followup.content, str) else ""
|
|
report = fc if fc and not _looks_like_raw_tool_call(fc) else raw_data
|
|
result = followup
|
|
else:
|
|
report = content
|
|
|
|
return {
|
|
"messages": [result],
|
|
"quant_report": report,
|
|
}
|
|
|
|
return quant_analyst_node
|