TradingAgents/tradingagents/agents/analysts/fundamentals_analyst.py

145 lines
7.1 KiB
Python

from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
import time
import json
from tradingagents.agents.utils.agent_utils import get_fundamentals, get_balance_sheet, get_cashflow, get_income_statement, get_insider_sentiment, get_insider_transactions, get_analyst_recommendations, get_earnings_data, get_institutional_holders, execute_text_tool_calls, needs_followup_call, execute_default_tools, generate_analysis_from_data
from tradingagents.dataflows.config import get_config
from tradingagents.log_utils import add_log, step_timer, symbol_progress
ANALYST_RESPONSE_FORMAT = """
RESPONSE FORMAT (follow this structure exactly):
## EXECUTIVE SUMMARY
2-3 sentences: Key fundamental finding and directional bias (BULLISH / BEARISH / NEUTRAL).
## KEY DATA POINTS
- Bullet list of the 5 most significant fundamental metrics with specific numbers
- Include valuation, profitability, institutional positioning, analyst consensus
## SIGNAL ASSESSMENT
Your overall reading: BULLISH / BEARISH / NEUTRAL
1-2 sentences explaining why, referencing specific financial data.
## RISK FACTORS
2-3 specific fundamental risks (debt, margins, earnings misses, etc.).
## CONFIDENCE: HIGH / MEDIUM / LOW
1 sentence justifying your confidence level.
| Metric | Value | Signal | Significance |
|--------|-------|--------|-------------|
| (fill with key fundamental metrics) |
RULES:
- Maximum 3000 characters total
- Do NOT repeat raw data verbatim — summarize trends and insights
- Complete your ENTIRE analysis in a SINGLE response"""
def create_fundamentals_analyst(llm):
def fundamentals_analyst_node(state):
current_date = state["trade_date"]
ticker = state["company_of_interest"]
company_name = state["company_of_interest"]
tools = [
get_fundamentals,
get_balance_sheet,
get_cashflow,
get_income_statement,
get_analyst_recommendations,
get_earnings_data,
get_institutional_holders,
]
system_message = (
"You are a fundamentals analyst tasked with analyzing a company's financial health, valuation, and institutional positioning. "
"Use ALL available tools to build a comprehensive fundamental picture:\n"
"- `get_fundamentals`: Company overview, valuation ratios, profitability metrics\n"
"- `get_balance_sheet`, `get_cashflow`, `get_income_statement`: Financial statements\n"
"- `get_analyst_recommendations`: Wall Street analyst consensus and recent rating changes\n"
"- `get_earnings_data`: Earnings dates, EPS estimates vs actuals, earnings surprises\n"
"- `get_institutional_holders`: Top institutional holders, insider vs institutional ownership\n\n"
"Provide specific numbers and quantitative evidence. Do not simply state trends are mixed."
+ ANALYST_RESPONSE_FORMAT,
)
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}. The company we want to look at is {ticker}",
),
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(ticker=ticker)
chain = prompt | llm.bind_tools(tools)
step_timer.start_step("fundamentals_analyst")
add_log("agent", "fundamentals", f"📈 Fundamentals Analyst calling LLM for {ticker}...")
t0 = time.time()
result = chain.invoke(state["messages"])
elapsed = time.time() - t0
report = ""
if len(result.tool_calls) == 0:
report = result.content
add_log("llm", "fundamentals", f"LLM responded in {elapsed:.1f}s ({len(report)} chars)")
tool_results = execute_text_tool_calls(report, tools)
if tool_results:
add_log("data", "fundamentals", f"Executed {len(tool_results)} tool calls: {', '.join(t['name'] for t in tool_results)}")
else:
add_log("agent", "fundamentals", f"🔄 No tool calls found, proactively fetching data for {ticker}...")
tool_results = execute_default_tools(tools, ticker, current_date)
add_log("data", "fundamentals", f"Proactively fetched {len(tool_results)} data sources")
if tool_results and needs_followup_call(report):
add_log("agent", "fundamentals", f"🔄 Generating analysis from {len(tool_results)} tool results...")
t1 = time.time()
followup = generate_analysis_from_data(llm, tool_results, system_message, ticker, current_date)
elapsed2 = time.time() - t1
if followup and len(followup) > 100:
report = followup
add_log("llm", "fundamentals", f"Follow-up analysis generated in {elapsed2:.1f}s ({len(report)} chars)")
add_log("agent", "fundamentals", f"✅ Fundamentals report ready: {report[:300]}...")
step_timer.end_step("fundamentals_analyst", "completed", report[:200])
symbol_progress.step_done(ticker, "fundamentals_analyst")
step_timer.update_details("fundamentals_analyst", {
"system_prompt": system_message[:2000],
"user_prompt": f"Analyze fundamentals for {ticker} on {current_date}",
"response": report[:3000],
"tool_calls": tool_results if tool_results else [],
})
else:
tool_call_info = [{"name": tc["name"], "args": str(tc.get("args", {}))[:200]} for tc in result.tool_calls]
step_timer.set_details("fundamentals_analyst", {
"system_prompt": system_message[:2000],
"user_prompt": f"Analyze fundamentals for {ticker} on {current_date}",
"response": "(Pending - tool calls in progress)",
"tool_calls": tool_call_info,
})
add_log("data", "fundamentals", f"LLM requested {len(result.tool_calls)} tool calls in {elapsed:.1f}s: {', '.join(tc['name'] for tc in result.tool_calls)}")
return {
"messages": [result],
"fundamentals_report": report,
}
return fundamentals_analyst_node