main.py.finished

import prediction_market_agent_tooling
import prediction_market_agent_tooling.deploy
import prediction_market_agent_tooling.deploy.agent
from dotenv import load_dotenv
from langchain_core.output_parsers import PydanticOutputParser
from langchain_core.prompts import ChatPromptTemplate
from langchain_openai import ChatOpenAI
from prediction_market_agent_tooling.config import APIKeys
from prediction_market_agent_tooling.loggers import logger
from prediction_market_agent_tooling.deploy.agent import (
    AgentMarket,
    ProbabilisticAnswer,
)
from prediction_market_agent_tooling.deploy.agent import MarketType
from prediction_market_agent_tooling.deploy.betting_strategy import (
    BettingStrategy,
    KellyBettingStrategy,
)
from prediction_market_agent_tooling.markets.agent_market import AgentMarket
from prediction_market_agent_tooling.markets.data_models import ProbabilisticAnswer
from prediction_market_agent_tooling.tools.tavily.tavily_search import tavily_search

load_dotenv()


class MyAgent(prediction_market_agent_tooling.deploy.agent.DeployableTraderAgent):
    """
    TODO: Implement market betting agent.

    Step-by-step implementations can be found at https://github.com/gnosis/gnosis-ai-hackathon-starter/tree/step-4/trader_quickstart/agents.

    Full-fledged agents that Gnosis AI runs can be found in https://github.com/gnosis/prediction-market-agent.
    """

    def load(self) -> None:
        super().load()
        keys = APIKeys()
        self.agent = ChatOpenAI(
            model="gpt-4o",
            api_key=keys.openai_api_key_secretstr_v1,
        )

    def answer_binary_market(self, market: AgentMarket) -> ProbabilisticAnswer | None:
        # Search on Tavily for recent news
        response = tavily_search(market.question)
        tavily_content = "\n".join(r.content for r in response.results)
        logger.info(tavily_content)

        # Prepare prompt for the LLM
        parser = PydanticOutputParser(pydantic_object=ProbabilisticAnswer)
        prediction_prompt = ChatPromptTemplate.from_template(
            template="""INTRODUCTION:
Your primary task is to accurately estimate the probabilities for the outcome of a 'market question',
found in 'MARKET_QUESTION'. The market question is part of a prediction market, where users can place bets on the outcomes of market questions and earn rewards if the selected outcome occurrs. The 'market question'
in this scenario has only two possible outcomes: `Yes` or `No`. 
Each market has a closing date at which the outcome is evaluated. 
This date is stated within the market question.

MARKET_QUESTION:
```
{market_question}
```

MARKET_CONTEXT:
```
{tavily_content}
```

OUTPUT_FORMAT:
{format_instructions}
""",
        )

        # And call the model
        prompt_and_model = prediction_prompt | self.agent
        llm_output = prompt_and_model.invoke(
            {
                "format_instructions": parser.get_format_instructions(),
                "market_question": market.question,
                "tavily_content": tavily_content,
            }
        )
        structured_output: ProbabilisticAnswer = parser.invoke(llm_output)

        return structured_output

    def get_betting_strategy(self, market: AgentMarket) -> BettingStrategy:
        return KellyBettingStrategy(max_bet_amount=0.001)


MyAgent().run(market_type=MarketType.OMEN)