108 lines
4.0 KiB
Python
108 lines
4.0 KiB
Python
import time
|
|
import json
|
|
from langchain_core.messages import SystemMessage, HumanMessage
|
|
|
|
from tradingagents.log_utils import add_log, step_timer, symbol_progress
|
|
|
|
|
|
def create_neutral_debator(llm):
|
|
def neutral_node(state) -> dict:
|
|
risk_debate_state = state["risk_debate_state"]
|
|
history = risk_debate_state.get("history", "")
|
|
neutral_history = risk_debate_state.get("neutral_history", "")
|
|
|
|
current_risky_response = risk_debate_state.get("current_risky_response", "")
|
|
current_safe_response = risk_debate_state.get("current_safe_response", "")
|
|
|
|
market_research_report = state["market_report"]
|
|
sentiment_report = state["sentiment_report"]
|
|
news_report = state["news_report"]
|
|
fundamentals_report = state["fundamentals_report"]
|
|
|
|
trader_decision = state["trader_investment_plan"]
|
|
|
|
system_prompt = """You are a Neutral Risk Analyst at a financial advisory firm. You MUST stay in character as a financial analyst at all times.
|
|
|
|
CRITICAL RULES:
|
|
- NEVER mention that you are an AI, Claude, a language model, or an assistant
|
|
- NEVER offer to help with code, software, or implementation tasks
|
|
- NEVER say "I don't have access to" or "I can't see the data" — analyze whatever data is provided below
|
|
- If data sections are empty, state that data is unavailable and focus your analysis on the data that IS available
|
|
|
|
Your role: Provide a balanced perspective, weighing both potential benefits and risks objectively.
|
|
Focus on: balanced risk/reward assessment, moderate strategies, where both aggressive and conservative views may be flawed.
|
|
|
|
RESPONSE FORMAT:
|
|
- Maximum 2000 characters. Focus on 3-5 key balanced observations.
|
|
- Complete your ENTIRE argument in a SINGLE response.
|
|
|
|
Respond only with your balanced financial analysis. No disclaimers or meta-commentary."""
|
|
|
|
user_prompt = f"""Provide the balanced/neutral perspective on this investment:
|
|
|
|
TRADER'S DECISION:
|
|
{trader_decision}
|
|
|
|
MARKET DATA:
|
|
{market_research_report}
|
|
|
|
SENTIMENT:
|
|
{sentiment_report}
|
|
|
|
NEWS:
|
|
{news_report}
|
|
|
|
FUNDAMENTALS:
|
|
{fundamentals_report}
|
|
|
|
DEBATE HISTORY:
|
|
{history}
|
|
|
|
AGGRESSIVE ANALYST'S ARGUMENT:
|
|
{current_risky_response if current_risky_response else "None yet"}
|
|
|
|
CONSERVATIVE ANALYST'S ARGUMENT:
|
|
{current_safe_response if current_safe_response else "None yet"}
|
|
|
|
Present your balanced/neutral case."""
|
|
|
|
messages = [
|
|
SystemMessage(content=system_prompt),
|
|
HumanMessage(content=user_prompt)
|
|
]
|
|
step_timer.start_step("neutral_analyst")
|
|
add_log("agent", "neutral", f"⚖️ Neutral Analyst calling LLM...")
|
|
t0 = time.time()
|
|
response = llm.invoke(messages)
|
|
elapsed = time.time() - t0
|
|
add_log("llm", "neutral", f"LLM responded in {elapsed:.1f}s ({len(response.content)} chars)")
|
|
add_log("agent", "neutral", f"✅ Neutral argument ready: {response.content[:300]}...")
|
|
step_timer.end_step("neutral_analyst", "completed", response.content[:200])
|
|
symbol_progress.step_done(state["company_of_interest"], "neutral_analyst")
|
|
step_timer.set_details("neutral_analyst", {
|
|
"system_prompt": system_prompt,
|
|
"user_prompt": user_prompt[:3000],
|
|
"response": response.content[:3000],
|
|
"tool_calls": [],
|
|
})
|
|
|
|
argument = f"Neutral Analyst: {response.content}"
|
|
|
|
new_risk_debate_state = {
|
|
"history": history + "\n" + argument,
|
|
"risky_history": risk_debate_state.get("risky_history", ""),
|
|
"safe_history": risk_debate_state.get("safe_history", ""),
|
|
"neutral_history": neutral_history + "\n" + argument,
|
|
"latest_speaker": "Neutral",
|
|
"current_risky_response": risk_debate_state.get(
|
|
"current_risky_response", ""
|
|
),
|
|
"current_safe_response": risk_debate_state.get("current_safe_response", ""),
|
|
"current_neutral_response": argument,
|
|
"count": risk_debate_state["count"] + 1,
|
|
}
|
|
|
|
return {"risk_debate_state": new_risk_debate_state}
|
|
|
|
return neutral_node
|