main.py

import questionary
from utils.data_fetcher import (
    fetch_market_data,
    fetch_news,
    fetch_onchain_data,
    fetch_nvt,
    fetch_mvrv,
    fetch_historical_prices,
    calculate_rsi,
    calculate_correlations,
    calculate_var,
)
from agents.valuation_agent import valuation_agent
from agents.sentiment_agent import sentiment_agent
from agents.fundamentals_agent import fundamentals_agent
from agents.technicals_agent import technicals_agent
from agents.strategy_agent import strategy_agent
from agents.risk_manager import risk_manager
from agents.portfolio_manager import portfolio_manager
from utils.portfolio import Portfolio
from termcolor import colored

# all available agents
AGENT_CHOICES = [
    {"name": "Valuation Agent (analyzes market valuation)", "value": "valuation"},
    {"name": "Sentiment Agent (analyzes news sentiment)", "value": "sentiment"},
    {"name": "Fundamentals Agent (analyzes on-chain metrics)", "value": "fundamentals"},
    {"name": "Technicals Agent (analyzes price charts)", "value": "technicals"},
    {"name": "Strategy Agent (synthesizes signals into trading strategies)", "value": "strategy"},
]

# all available cryptocurrencies
CRYPTO_CHOICES = [
    {"name": "Binance Coin (BNB)", "value": "binancecoin"},
    {"name": "Bitcoin (BTC)", "value": "bitcoin"},
    {"name": "Ethereum (ETH)", "value": "ethereum"},
    {"name": "Ripple (XRP)", "value": "xrp"},
    {"name": "USD Coin (USDC)", "value": "usd-coin"}
]

# all available models
MODEL_CHOICES = [
    {"name": "Qwen QwQ 32B (most accurate, may run into loop)", "value": "qwen-qwq-32b|131072|0.6|0.95"},
    {"name": "Deepseek R1 Distil Llama 70B (even more accurate)", "value": "deepseek-r1-distill-llama-70b|131072|0.6|0.95"},
    {"name": "Qwen 2.5 32B (more accurate)", "value": "qwen-2.5-32b|131072|0.7|0.8"},
    {"name": "Llama 3.3 70B (fastest)", "value": "llama-3.3-70b-specdec|8192|0.6|0.9"},
]


def main():
    print(colored("Welcome to the AI Crypto Fund", "red", attrs=["bold"]))

    # user inputs
    selected_agents = questionary.checkbox(
        "Select agents to use:",
        choices=AGENT_CHOICES,
        validate=lambda x: len(x) > 0 or "Please select at least one agent.",
    ).ask()

    selected_cryptos = questionary.checkbox(
        "Select cryptocurrencies to trade:",
        choices=CRYPTO_CHOICES,
        validate=lambda x: len(x) > 0 or "Please select at least one cryptocurrency.",
    ).ask()

    selected_model = questionary.select(
        "Select model to use:", choices=MODEL_CHOICES
    ).ask()

    # Extract the values (e.g., "valuation", "bitcoin") from the selected choices
    selected_agents = selected_agents
    selected_cryptos = selected_cryptos

    available_money = questionary.text("How much money do you have to invest?").ask()
    if (
        available_money is None
        or not available_money.isdigit()
        or float(available_money) <= 0
    ):
        print(colored("No money, no honey!", "red"))
        return

    portfolio = Portfolio(initial_cash=float(available_money))

    # fetch market data
    print(colored("\nFetching market data...", "blue"))
    market_data = fetch_market_data(selected_cryptos)

    # fetch news articles
    print(colored("Fetching news articles...", "blue"))
    news = fetch_news()

    # fetch on-chain data
    print(colored("Fetching on-chain data...", "blue"))
    onchain_data = {crypto: fetch_onchain_data(crypto) for crypto in selected_cryptos}

    # fetch historical prices
    print(colored("Fetching historical prices...", "blue"))
    historical_prices = {
        crypto: fetch_historical_prices(crypto) for crypto in selected_cryptos
    }

    # calculate technical indicators
    print(colored("Calculating technical indicators...", "blue"))
    technical_indicators = {
        crypto: {"RSI": calculate_rsi(historical_prices[crypto])}
        for crypto in selected_cryptos
    }

    # fetch valuation metrics
    print(colored("Fetching valuation metrics...", "blue"))
    valuation_metrics = {
        crypto: {"NVT": fetch_nvt(crypto), "MVRV": fetch_mvrv(crypto)}
        for crypto in selected_cryptos
    }

    # generate signals from selected agents
    signals = {}
    for crypto in selected_cryptos:
        signals[crypto] = {}
        for agent_name in selected_agents:
            prefix = colored(
                f"Analyzing {crypto} with {agent_name} agent...", "magenta"
            )
            print(prefix, end="", flush=True)
            current_price = market_data[crypto]["usd"]
            if agent_name == "valuation":
                signals[crypto][agent_name] = valuation_agent(
                    crypto,
                    current_price,
                    valuation_metrics[crypto],
                    prefix=prefix,
                    model=selected_model,
                )
            elif agent_name == "sentiment":
                signals[crypto][agent_name] = sentiment_agent(
                    crypto, news, prefix=prefix, model=selected_model
                )
            elif agent_name == "fundamentals":
                signals[crypto][agent_name] = fundamentals_agent(
                    crypto, onchain_data[crypto], prefix=prefix, model=selected_model
                )
            elif agent_name == "technicals":
                signals[crypto][agent_name] = technicals_agent(
                    crypto,
                    historical_prices[crypto],
                    technical_indicators[crypto],
                    prefix=prefix,
                    model=selected_model,
                )
            elif agent_name == "strategy":
                signals[crypto][agent_name] = strategy_agent(
                    crypto,
                    signals[crypto],
                    portfolio.get_state()["positions"],
                    prefix=prefix,
                    model=selected_model,
                )
            print()

    # calculate correlations
    print(colored("Calculating correlations...", "blue"))
    correlations = calculate_correlations(historical_prices)

    # calculate portfolio VaR
    print(colored("Calculating portfolio VaR...", "blue"))
    positions = portfolio.get_state()["positions"]
    total_value = sum(
        positions.get(crypto, {"long": 0, "short": 0})["long"]
        * market_data[crypto]["usd"]
        for crypto in selected_cryptos
    )
    if total_value == 0:  # if no positions, assume equal weights
        portfolio_weights = {
            crypto: 1 / len(selected_cryptos) for crypto in selected_cryptos
        }
    else:
        portfolio_weights = {
            crypto: (
                positions.get(crypto, {"long": 0, "short": 0})["long"]
                * market_data[crypto]["usd"]
            )
            / total_value
            for crypto in selected_cryptos
        }
    var = calculate_var(
        historical_prices, portfolio_weights, confidence_level=0.95, time_horizon=1
    )

    # risk management
    prefix = colored("Assessing portfolio risk...", "blue")
    print(prefix, end="", flush=True)
    risk_output = risk_manager(
        portfolio.get_state()["positions"],
        historical_prices,
        correlations,
        {"var": var},
        prefix=prefix,
        model=selected_model,
    )
    print()

    # make trading decisions
    prefix = colored("Making trading decisions...", "blue")
    print(prefix, end="", flush=True)
    current_prices = {crypto: market_data[crypto]["usd"] for crypto in selected_cryptos}
    decisions = portfolio_manager(
        signals,
        risk_output["max_amounts"],
        portfolio.cash,
        portfolio.get_state()["positions"],
        current_prices,
        margin_requirement=0.5,
        prefix=prefix,
        model=selected_model,
    )
    print()

    print(colored("\nTrading Decisions:", "green", attrs=["bold"]))
    for crypto in selected_cryptos:
        decision = decisions["decisions"].get(
            crypto,
            {
                "action": "hold",
                "quantity": 0.0,
                "confidence": 0.0,
                "reasoning": "No signal provided",
            },
        )
        action = decision["action"]
        quantity = decision["quantity"]
        confidence = decision["confidence"]
        reasoning = decision["reasoning"]
        color = (
            "green"
            if action in ["buy", "cover"]
            else "red" if action in ["sell", "short"] else "yellow"
        )
        print(
            colored(
                f"{crypto.capitalize()}: {action.upper()} {quantity:.2f} USD (Confidence: {confidence}%)",
                color,
            )
        )
        print(colored(f"Reasoning: {reasoning}", "white"))


if __name__ == "__main__":
    main()