Editing AI for Wealth Management (section)

== <span style="color: #FFFFFF;">Applying</span> ==
'''Mean-variance portfolio optimization with ML expected returns:'''
<syntaxhighlight lang="python">
import numpy as np
import pandas as pd
from scipy.optimize import minimize
import yfinance as yf
from sklearn.linear_model import Ridge

# Step 1: Download historical price data
tickers = ['SPY', 'QQQ', 'BND', 'GLD', 'VNQ', 'EFA', 'EEM']
prices = yf.download(tickers, start='2018-01-01', end='2024-01-01')['Adj Close']
returns = prices.pct_change().dropna()

# Step 2: ML-enhanced return prediction (Black-Litterman style)
# Combine equilibrium returns with ML views
mkt_cap_weights = np.array([0.35, 0.18, 0.20, 0.05, 0.05, 0.12, 0.05])
cov_matrix = returns.ewm(span=252).cov().iloc[-len(tickers):].values  # Exponential weighting
risk_aversion = 2.5
# Implied equilibrium returns (reverse-engineered from market cap weights)
pi = risk_aversion * cov_matrix @ mkt_cap_weights

# ML momentum signal as view
momentum_12m = returns.rolling(252).mean().iloc[-1].values
ridge = Ridge(alpha=10)
# In practice: train ridge on momentum → forward returns; use predictions as views
ml_views = momentum_12m  # Simplified: use raw momentum as return expectation

# Black-Litterman: combine equilibrium + ML views
tau = 0.05
P = np.eye(len(tickers))  # View matrix (full views on all assets)
omega = np.diag(np.diag(tau * P @ cov_matrix @ P.T))  # Uncertainty of views
# BL formula: posterior returns
bl_returns = np.linalg.inv(np.linalg.inv(tau * cov_matrix) + P.T @ np.linalg.inv(omega) @ P) @ \
             (np.linalg.inv(tau * cov_matrix) @ pi + P.T @ np.linalg.inv(omega) @ ml_views)

# Step 3: Optimize portfolio weights (max Sharpe)
def neg_sharpe(weights, returns_est, cov, rf=0.04):
    port_ret = weights @ returns_est * 252  # Annualized
    port_vol = np.sqrt(weights @ cov @ weights * 252)
    return -(port_ret - rf) / port_vol

constraints = [{'type': 'eq', 'fun': lambda w: np.sum(w) - 1}]
bounds = [(0.02, 0.40)] * len(tickers)  # Min 2%, max 40% per asset
result = minimize(neg_sharpe, x0=mkt_cap_weights, args=(bl_returns, cov_matrix),
                  method='SLSQP', bounds=bounds, constraints=constraints)
optimal_weights = result.x
portfolio = dict(zip(tickers, optimal_weights))
print("Optimal Portfolio:", {k: f"{v:.1%}" for k, v in portfolio.items()})
</syntaxhighlight>

; Wealth management AI tools
: '''Robo-advisors''' → Betterment, Wealthfront, Schwab Intelligent, Vanguard Digital Advisor
: '''Direct indexing + TLH''' → Parametric (Morgan Stanley), Canvas (Wealthfront), Fidelity Managed FidFolios
: '''Advisor tools''' → Orion Portfolio Solutions, Riskalyze (Nitrogen), MoneyGuide Pro
: '''Conversational AI''' → Schwab Intelligent Portfolios Premium chat, Avantis (JPMorgan)
: '''ESG analytics''' → MSCI ESG Ratings, Sustainalytics, Bloomberg ESG
</div>

<div style="background-color: #8B4500; color: #FFFFFF; padding: 20px; border-radius: 8px; margin-bottom: 15px;">