vault-dash/app/strategies/engine.py

from __future__ import annotations

from dataclasses import dataclass
from typing import Literal

from app.core.pricing.black_scholes import DEFAULT_GLD_PRICE, DEFAULT_RISK_FREE_RATE, DEFAULT_VOLATILITY
from app.models.portfolio import LombardPortfolio
from app.strategies.base import BaseStrategy, StrategyConfig
from app.strategies.laddered_put import LadderSpec, LadderedPutStrategy
from app.strategies.lease import LeaseStrategy
from app.strategies.protective_put import ProtectivePutSpec, ProtectivePutStrategy

RiskProfile = Literal["conservative", "balanced", "cost_sensitive"]

RESEARCH_PORTFOLIO_VALUE = 1_000_000.0
RESEARCH_LOAN_AMOUNT = 600_000.0
RESEARCH_MARGIN_CALL_THRESHOLD = 0.75
RESEARCH_GLD_SPOT = 460.0
RESEARCH_VOLATILITY = 0.16
RESEARCH_RISK_FREE_RATE = 0.045


@dataclass(frozen=True)
class StrategySelectionEngine:
    """Compare paper strategies and recommend the best fit by risk profile."""

    portfolio_value: float = RESEARCH_PORTFOLIO_VALUE
    loan_amount: float = RESEARCH_LOAN_AMOUNT
    margin_call_threshold: float = RESEARCH_MARGIN_CALL_THRESHOLD
    spot_price: float = RESEARCH_GLD_SPOT
    volatility: float = RESEARCH_VOLATILITY
    risk_free_rate: float = RESEARCH_RISK_FREE_RATE

    def _config(self) -> StrategyConfig:
        portfolio = LombardPortfolio(
            gold_ounces=self.portfolio_value / self.spot_price,
            gold_price_per_ounce=self.spot_price,
            loan_amount=self.loan_amount,
            initial_ltv=self.loan_amount / self.portfolio_value,
            margin_call_ltv=self.margin_call_threshold,
        )
        return StrategyConfig(
            portfolio=portfolio,
            spot_price=self.spot_price,
            volatility=self.volatility,
            risk_free_rate=self.risk_free_rate,
        )

    def _strategies(self) -> list[BaseStrategy]:
        config = self._config()
        return [
            ProtectivePutStrategy(config, ProtectivePutSpec(label="ATM", strike_pct=1.0, months=12)),
            ProtectivePutStrategy(config, ProtectivePutSpec(label="OTM_95", strike_pct=0.95, months=12)),
            ProtectivePutStrategy(config, ProtectivePutSpec(label="OTM_90", strike_pct=0.90, months=12)),
            LadderedPutStrategy(
                config,
                LadderSpec(label="50_50_ATM_OTM95", weights=(0.5, 0.5), strike_pcts=(1.0, 0.95), months=12),
            ),
            LadderedPutStrategy(
                config,
                LadderSpec(label="33_33_33_ATM_OTM95_OTM90", weights=(1 / 3, 1 / 3, 1 / 3), strike_pcts=(1.0, 0.95, 0.90), months=12),
            ),
            LeaseStrategy(config),
        ]

    def compare_all_strategies(self) -> list[dict]:
        comparisons: list[dict] = []
        for strategy in self._strategies():
            cost = strategy.calculate_cost()
            protection = strategy.calculate_protection()
            scenarios = strategy.get_scenarios()
            annual_cost = cost.get("annualized_cost", cost.get("lowest_annual_cost", 0.0))
            protection_ltv = protection.get("hedged_ltv_at_threshold")
            if protection_ltv is None:
                duration_rows = protection.get("durations", [])
                protection_ltv = min((row["hedged_ltv_at_threshold"] for row in duration_rows), default=1.0)
            comparisons.append(
                {
                    "name": strategy.name,
                    "cost": cost,
                    "protection": protection,
                    "scenarios": scenarios,
                    "score_inputs": {
                        "annual_cost": annual_cost,
                        "hedged_ltv_at_threshold": protection_ltv,
                    },
                }
            )
        return comparisons

    def recommend(self, risk_profile: RiskProfile = "balanced") -> dict:
        comparisons = self.compare_all_strategies()

        def score(item: dict) -> tuple[float, float]:
            annual_cost = item["score_inputs"]["annual_cost"]
            hedged_ltv = item["score_inputs"]["hedged_ltv_at_threshold"]
            if risk_profile == "conservative":
                return (hedged_ltv, annual_cost)
            if risk_profile == "cost_sensitive":
                return (annual_cost, hedged_ltv)
            return (hedged_ltv + (annual_cost / self.portfolio_value), annual_cost)

        recommended = min(comparisons, key=score)
        return {
            "risk_profile": risk_profile,
            "recommended_strategy": recommended["name"],
            "rationale": {
                "portfolio_value": self.portfolio_value,
                "loan_amount": self.loan_amount,
                "margin_call_threshold": self.margin_call_threshold,
                "spot_price": self.spot_price,
                "volatility": self.volatility,
                "risk_free_rate": self.risk_free_rate,
            },
            "comparison_summary": [
                {
                    "name": item["name"],
                    "annual_cost": round(item["score_inputs"]["annual_cost"], 2),
                    "hedged_ltv_at_threshold": round(item["score_inputs"]["hedged_ltv_at_threshold"], 6),
                }
                for item in comparisons
            ],
        }

    def sensitivity_analysis(self) -> dict:
        results: dict[str, list[dict]] = {"volatility": [], "spot_price": []}
        for volatility in (0.12, 0.16, 0.20):
            engine = StrategySelectionEngine(
                portfolio_value=self.portfolio_value,
                loan_amount=self.loan_amount,
                margin_call_threshold=self.margin_call_threshold,
                spot_price=self.spot_price,
                volatility=volatility,
                risk_free_rate=self.risk_free_rate,
            )
            recommendation = engine.recommend("balanced")
            results["volatility"].append(
                {
                    "volatility": volatility,
                    "recommended_strategy": recommendation["recommended_strategy"],
                }
            )
        for spot_price in (DEFAULT_GLD_PRICE * 0.9, DEFAULT_GLD_PRICE, DEFAULT_GLD_PRICE * 1.1):
            engine = StrategySelectionEngine(
                portfolio_value=self.portfolio_value,
                loan_amount=self.loan_amount,
                margin_call_threshold=self.margin_call_threshold,
                spot_price=spot_price,
                volatility=DEFAULT_VOLATILITY,
                risk_free_rate=DEFAULT_RISK_FREE_RATE,
            )
            recommendation = engine.recommend("balanced")
            results["spot_price"].append(
                {
                    "spot_price": round(spot_price, 2),
                    "recommended_strategy": recommendation["recommended_strategy"],
                }
            )
        return results