Analyser la combinaison de deux stratégies complementaires: - FamaFrench Alpha: Rotation des 5 factor ETFs (VLUE, MTUM, SIZE, QUAL, USMV) basee sur momentum risk-adjuste - AllWeather Alpha: Allocation statique macro (SPY/IEF/GLD/XLP = 30/30/30/10)
Lecon apprise de MomentumRegime: PAS d’overlap entre les univers. - FamaFrench: equity factors (facteurs Fama-French) - AllWeather: traditional assets (equities, bonds, gold, staples) - Vraie diversification: facteurs de marche + allocation macro
# Setup QuantBook
from AlgorithmImports import *
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import warnings
warnings.filterwarnings('ignore')
plt.style.use('seaborn-v0_8-darkgrid')
plt.rcParams['figure.figsize'] = (14, 5)
qb = QuantBook()
print("QuantBook initialise.")QuantBook initialise.On charge les 5 factor ETFs Fama-French et les 4 assets AllWeather.
# FamaFrench universe: 5 factor ETFs
ff_tickers = ["VLUE", "MTUM", "SIZE", "QUAL", "USMV"]
# AllWeather universe: 4 traditional assets
aw_tickers = ["SPY", "IEF", "GLD", "XLP"]
# Benchmark pour regime detection
all_tickers = ff_tickers + aw_tickers
symbols = {}
for ticker in all_tickers:
symbols[ticker] = qb.add_equity(ticker, Resolution.DAILY).symbol
# Charger l'historique (2010-2026 pour inclure une periode complete)
start = datetime(2010, 1, 1)
end = datetime(2026, 1, 1)
history = qb.history(list(symbols.values()), start, end, Resolution.DAILY)
print(f"Donnees chargees: {len(history)} lignes")Donnees chargees: 4024 lignesPivot de la série ‘close’ en DataFrame large, avec remapping des colonnes Symbol → ticker pour Framework_Composite_FamaFrenchAllWeather.
# Pivoter les donnees
closes = history['close'].unstack(level=0)
symbol_to_ticker = {str(v): k for k, v in symbols.items()}
closes.columns = [symbol_to_ticker.get(str(c), str(c)) for c in closes.columns]
closes = closes.dropna()
print(f"Periode: {closes.index[0].date()} a {closes.index[-1].date()}")
print(f"Donnees: {len(closes)} jours de trading")
print(f"Tickers: {list(closes.columns)}")Periode: 2010-01-04 a 2025-12-31
Donnees: 4024 jours de trading
Tickers: ['SPY']def compute_risk_adjusted_momentum(closes, tickers, lookback=252, skip_days=21, vol_window=63):
"""Calcule le momentum risk-adjuste pour les factor ETFs."""
scores = pd.DataFrame(index=closes.index, columns=tickers)
for ticker in tickers:
if ticker not in closes.columns:
continue
prices = closes[ticker]
scores[ticker] = np.nan
for i in range(lookback, len(prices)):
# Momentum 12m-1m (skip dernier mois)
past_price = prices.iloc[i - lookback]
skip_idx = max(i - skip_days, lookback)
skip_price = prices.iloc[skip_idx]
if past_price <= 0:
continue
momentum_return = (skip_price / past_price) - 1
# Volatilite realisee 63j
recent_prices = prices.iloc[max(0, i - vol_window):i]
if len(recent_prices) < 20:
continue
daily_rets = recent_prices.pct_change().dropna()
realized_vol = daily_rets.std() * np.sqrt(252)
if realized_vol <= 0.001:
continue
scores[ticker].iloc[i] = momentum_return / realized_vol
return scores
def compute_famafrench_signals(closes, tickers, scores_df, use_sma200=False):
"""Genere les signaux FamaFrench (mensuels)."""
signals = pd.DataFrame(index=closes.index, columns=tickers)
sma200 = closes['SPY'].rolling(200).mean() if 'SPY' in closes.columns else None
for i in range(252, len(closes)):
# Filtre SMA200 (optionnel)
if use_sma200 and sma200 is not None:
spy_price = closes['SPY'].iloc[i]
spy_sma = sma200.iloc[i]
if pd.notna(spy_sma) and spy_price <= spy_sma:
# Risk-off: USMV seulement
for t in tickers:
signals[t].iloc[i] = 1 if t == "USMV" else 0
continue
# Calculer les scores actuels
current_scores = {}
for t in tickers:
if t in scores_df.columns:
score = scores_df[t].iloc[i]
if pd.notna(score):
current_scores[t] = score
if len(current_scores) < 2:
continue
# Selectionner tous les facteurs avec momentum positif
positive_factors = [t for t, s in current_scores.items() if s > 0]
if len(positive_factors) == 0:
# Tous negatifs → USMV seulement
for t in tickers:
signals[t].iloc[i] = 1 if t == "USMV" else 0
else:
# Equal weight sur les facteurs positifs
for t in tickers:
signals[t].iloc[i] = 1 if t in positive_factors else 0
return signals
# Calculer les scores de momentum
ff_scores = compute_risk_adjusted_momentum(closes, ff_tickers)
# Signaux SANS SMA200 (design recommande)
ff_signals_no_sma = compute_famafrench_signals(closes, ff_tickers, ff_scores, use_sma200=False)
# Signaux AVEC SMA200 (pour comparaison)
ff_signals_with_sma = compute_famafrench_signals(closes, ff_tickers, ff_scores, use_sma200=True)
print("Scores risk-adjusted momentum (derniers 5 jours):")
print(ff_scores.iloc[-5:])
print(f"\nSignaux FamaFrench sans SMA200 (derniers 5 jours):")
print(ff_signals_no_sma.iloc[-5:])Scores risk-adjusted momentum (derniers 5 jours):
VLUE MTUM SIZE QUAL USMV
time
2025-12-24 13:00:00 NaN NaN NaN NaN NaN
2025-12-26 16:00:00 NaN NaN NaN NaN NaN
2025-12-29 16:00:00 NaN NaN NaN NaN NaN
2025-12-30 16:00:00 NaN NaN NaN NaN NaN
2025-12-31 16:00:00 NaN NaN NaN NaN NaN
Signaux FamaFrench sans SMA200 (derniers 5 jours):
VLUE MTUM SIZE QUAL USMV
time
2025-12-24 13:00:00 NaN NaN NaN NaN NaN
2025-12-26 16:00:00 NaN NaN NaN NaN NaN
2025-12-29 16:00:00 NaN NaN NaN NaN NaN
2025-12-30 16:00:00 NaN NaN NaN NaN NaN
2025-12-31 16:00:00 NaN NaN NaN NaN NaNLe score risk-adjuste = (rendement 12m-1m) / (volatilite 63j).
Les facteurs selectionnes varient mensuellement, creant une rotation dynamique.
Fonction de backtest combinant FamaFrench + AllWeather avec allocation variable.
def backtest_composite(closes, ff_signals, ff_alloc=0.20, aw_alloc=0.80, rebal_freq=21):
"""
Backtest du composite FamaFrench + AllWeather.
Args:
ff_alloc: Allocation FamaFrench (ex: 0.20 = 20%)
aw_alloc: Allocation AllWeather (ex: 0.80 = 80%)
rebal_freq: Frequence rebalance (21 = mensuel ~)
"""
returns_df = closes.pct_change()
portfolio_values = [1.0]
warmup = 252
counter = 0
# AllWeather holdings (statiques)
aw_weights = {"SPY": 0.30, "IEF": 0.30, "GLD": 0.30, "XLP": 0.10}
# FamaFrench holdings (dynamiques)
ff_holdings = set()
for i in range(warmup, len(closes)):
# Rebalance mensuel
counter += 1
if counter >= rebal_freq:
counter = 0
# Update FamaFrench holdings
ff_holdings = set()
for t in ff_tickers:
if t in ff_signals.columns and ff_signals[t].iloc[i] == 1:
ff_holdings.add(t)
# Calcul du return
port_return = 0.0
# Contribution FamaFrench
if len(ff_holdings) > 0:
weight_per_stock = ff_alloc / len(ff_holdings)
for t in ff_holdings:
if t in returns_df.columns:
port_return += weight_per_stock * returns_df[t].iloc[i]
# Contribution AllWeather
for t, w in aw_weights.items():
if t in returns_df.columns:
port_return += aw_alloc * w * returns_df[t].iloc[i]
portfolio_values.append(portfolio_values[-1] * (1 + port_return))
# Metriques
returns = np.diff(portfolio_values) / np.array(portfolio_values[:-1])
cum_returns = pd.Series(portfolio_values[1:], index=closes.index[warmup:])
total_ret = (portfolio_values[-1] / portfolio_values[0]) - 1
years = len(returns) / 252
cagr = (1 + total_ret) ** (1 / years) - 1 if years > 0 else 0
vol = np.std(returns) * np.sqrt(252)
sharpe = (cagr - 0.03) / vol if vol > 0.001 else 0
running_max = cum_returns.expanding().max()
drawdown = (cum_returns - running_max) / running_max
max_dd = drawdown.min()
return {
'cum': cum_returns,
'sharpe': sharpe,
'cagr': cagr,
'max_dd': max_dd,
'vol': vol,
'final_value': portfolio_values[-1]
}
print("Fonction de backtest composite definie.")Fonction de backtest composite definie.On teste différentes allocations FF/AW de 10/90 a 50/50.
# Test differentes allocations (SANS SMA200)
allocations = [
(0.10, 0.90, "FF10/AW90"),
(0.20, 0.80, "FF20/AW80"),
(0.30, 0.70, "FF30/AW70"),
(0.40, 0.60, "FF40/AW60"),
(0.50, 0.50, "FF50/AW50"),
]
results_no_sma = {}
print(f"{'Allocation':<15} {'Sharpe':>8} {'CAGR':>8} {'MaxDD':>8} {'Vol':>8}")
print("-" * 50)
for ff_alloc, aw_alloc, name in allocations:
r = backtest_composite(closes, ff_signals_no_sma,
ff_alloc=ff_alloc, aw_alloc=aw_alloc)
results_no_sma[name] = r
print(f"{name:<15} {r['sharpe']:>8.3f} {r['cagr']:>7.1%} {r['max_dd']:>7.1%} {r['vol']:>7.1%}")
best_no_sma = max(results_no_sma.items(), key=lambda x: x[1]['sharpe'])
print(f"\nMeilleure allocation (sans SMA200): {best_no_sma[0]} (Sharpe={best_no_sma[1]['sharpe']:.3f})")Allocation Sharpe CAGR MaxDD Vol
--------------------------------------------------
FF10/AW90 -0.086 2.7% -10.0% 3.8%
FF20/AW80 -0.182 2.4% -8.9% 3.4%
FF30/AW70 -0.307 2.1% -7.8% 3.0%
FF40/AW60 -0.473 1.8% -6.7% 2.6%
FF50/AW50 -0.707 1.5% -5.6% 2.1%
Meilleure allocation (sans SMA200): FF10/AW90 (Sharpe=-0.086)L’allocation FF20/AW80 est le point de depart du sweep. Verifier si une autre allocation surperforme.
Verdict: L’allocation avec le Sharpe le plus eleve est optimale.
Comparaison: avec vs sans filtre SMA200 (AllWeather gere le defensif).
# Test AVEC SMA200
results_with_sma = {}
for ff_alloc, aw_alloc, name in allocations:
r = backtest_composite(closes, ff_signals_with_sma,
ff_alloc=ff_alloc, aw_alloc=aw_alloc)
results_with_sma[name] = r
# Comparaison
print(f"{'Allocation':<15} {'Sharpe NoSMA':>15} {'Sharpe WithSMA':>15} {'Delta':>10}")
print("-" * 55)
for name in results_no_sma.keys():
s_no_sma = results_no_sma[name]['sharpe']
s_with_sma = results_with_sma[name]['sharpe']
delta = s_no_sma - s_with_sma
print(f"{name:<15} {s_no_sma:>15.3f} {s_with_sma:>15.3f} {delta:>+10.3f}")
print(f"\nVerdict: ", end="")
if s_no_sma > s_with_sma:
print("SANS SMA200 est superieur (AllWeather gere le defensif)")
else:
print("AVEC SMA200 est superieur (double protection defensif)")Allocation Sharpe NoSMA Sharpe WithSMA Delta
-------------------------------------------------------
FF10/AW90 -0.086 -0.086 +0.000
FF20/AW80 -0.182 -0.182 +0.000
FF30/AW70 -0.307 -0.307 +0.000
FF40/AW60 -0.473 -0.473 +0.000
FF50/AW50 -0.707 -0.707 +0.000
Verdict: AVEC SMA200 est superieur (double protection defensif)fig, axes = plt.subplots(1, 2, figsize=(16, 5))
# Gauche: Sans SMA200
ax = axes[0]
for name, r in results_no_sma.items():
ax.plot(r['cum'].values, label=f"{name} (S={r['sharpe']:.2f})", linewidth=1.5)
ax.set_title('Allocation FF/AW (sans SMA200)', fontsize=12, fontweight='bold')
ax.set_ylabel('Valeur du portefeuille')
ax.legend(fontsize=9)
ax.grid(True, alpha=0.3)
# Droite: Impact du SMA200
ax = axes[1]
best_name = best_no_sma[0]
ax.plot(results_no_sma[best_name]['cum'].values,
label=f"Sans SMA200 (S={results_no_sma[best_name]['sharpe']:.2f})", linewidth=1.5)
ax.plot(results_with_sma[best_name]['cum'].values,
label=f"Avec SMA200 (S={results_with_sma[best_name]['sharpe']:.2f})", linewidth=1.5, linestyle='--')
ax.set_title(f'Impact SMA200 ({best_name})', fontsize=12, fontweight='bold')
ax.set_ylabel('Valeur du portefeuille')
ax.legend(fontsize=9)
ax.grid(True, alpha=0.3)
plt.tight_layout()
plt.savefig('composite_famafrench_allweather.png', dpi=150, bbox_inches='tight')
plt.show()
print("Graphique sauvegarde.")
Graphique sauvegarde.def backtest_famafrench_only(closes, signals):
"""Backtest FamaFrench uniquement."""
returns_df = closes.pct_change()
portfolio_values = [1.0]
warmup = 252
counter = 0
holdings = set()
for i in range(warmup, len(closes)):
counter += 1
if counter >= 21:
counter = 0
holdings = set()
for t in ff_tickers:
if t in signals.columns and signals[t].iloc[i] == 1:
holdings.add(t)
port_return = 0.0
if len(holdings) > 0:
for t in holdings:
if t in returns_df.columns:
port_return += (1.0 / len(holdings)) * returns_df[t].iloc[i]
portfolio_values.append(portfolio_values[-1] * (1 + port_return))
returns = np.diff(portfolio_values) / np.array(portfolio_values[:-1])
cagr = (1 + (portfolio_values[-1] / portfolio_values[0]) - 1) ** (1 / (len(returns) / 252)) - 1
vol = np.std(returns) * np.sqrt(252)
sharpe = (cagr - 0.03) / vol
return {'sharpe': sharpe, 'cagr': cagr, 'values': portfolio_values[warmup:]}
def backtest_allweather_only(closes):
"""Backtest AllWeather uniquement."""
returns_df = closes.pct_change()
aw_weights = {"SPY": 0.30, "IEF": 0.30, "GLD": 0.30, "XLP": 0.10}
portfolio_values = [1.0]
warmup = 21
for i in range(warmup, len(closes)):
port_return = 0.0
for t, w in aw_weights.items():
if t in returns_df.columns:
port_return += w * returns_df[t].iloc[i]
portfolio_values.append(portfolio_values[-1] * (1 + port_return))
returns = np.diff(portfolio_values) / np.array(portfolio_values[:-1])
cagr = (1 + (portfolio_values[-1] / portfolio_values[0]) - 1) ** (1 / (len(returns) / 252)) - 1
vol = np.std(returns) * np.sqrt(252)
sharpe = (cagr - 0.03) / vol
return {'sharpe': sharpe, 'cagr': cagr, 'values': portfolio_values[warmup:]}
# Backtester les strategies individuelles
ff_result = backtest_famafrench_only(closes, ff_signals_no_sma)
aw_result = backtest_allweather_only(closes)
print("Strategies individuelles:")
print(f" FamaFrench: Sharpe={ff_result['sharpe']:.3f}, CAGR={ff_result['cagr']:.1%}")
print(f" AllWeather: Sharpe={aw_result['sharpe']:.3f}, CAGR={aw_result['cagr']:.1%}")
print(f"\nComposite ({best_no_sma[0]}): Sharpe={best_no_sma[1]['sharpe']:.3f}, CAGR={best_no_sma[1]['cagr']:.1%}")Strategies individuelles:
FamaFrench: Sharpe=-inf, CAGR=0.0%
AllWeather: Sharpe=0.023, CAGR=3.1%
Composite (FF10/AW90): Sharpe=-0.086, CAGR=2.7%| Metrique | FamaFrench | AllWeather | Composite optimal |
|---|---|---|---|
| Sharpe | (a remplir) | (a remplir) | (a remplir) |
| CAGR | (a remplir) | (a remplir) | (a remplir) |
| Max DD | (a remplir) | (a remplir) | (a remplir) |
Allocation: [a remplir]
[a remplir: avec ou sans SMA200 est meilleur?]