Overall Statistics
Total Orders
227
Average Win
3.01%
Average Loss
-1.97%
Compounding Annual Return
27.127%
Drawdown
48.700%
Expectancy
0.670
Start Equity
100000.0
End Equity
422389.94
Net Profit
322.390%
Sharpe Ratio
0.84
Sortino Ratio
0.899
Probabilistic Sharpe Ratio
30.732%
Loss Rate
34%
Win Rate
66%
Profit-Loss Ratio
1.53
Alpha
0.119
Beta
1.047
Annual Standard Deviation
0.236
Annual Variance
0.056
Information Ratio
0.762
Tracking Error
0.161
Treynor Ratio
0.189
Total Fees
$900.38
Estimated Strategy Capacity
$7600000.00
Lowest Capacity Asset
BND TRO5ZARLX6JP
Portfolio Turnover
4.77%
Drawdown Recovery
507
 
# region imports
from AlgorithmImports import *

from sklearn.tree import DecisionTreeRegressor
from sklearn.preprocessing import StandardScaler
# endregion


class StockBondRotationAlgorithm(QCAlgorithm):
    """
    Machine Learning based dynamic allocation between equities and bonds.

    Based on:
    https://www.quantconnect.com/research/18049/reimagining-the-60-40-portfolio-in-an-era-of-ai-and-falling-rates/p1
    """

    def initialize(self):
        # keep in mind data snooping bias:
        # always leave some (typically more recent) time period for testing!
        self.set_start_date(2017, 1, 1)
        self.set_end_date(2022, 12, 31)

        # key settings
        tickers = ['SPY', 'GLD', 'BND']
        ticker_crypto = "BTCUSD"
        fred_factors = ['VIXCLS', 'T10Y3M', 'DFF']
        self._lookback_years = 4
        self._max_bitcoin_weight = 0.25
        tree_depth = 12
        
        self.settings.daily_precise_end_time = False
        
        self._equities = [self.add_equity(ticker).symbol for ticker in tickers]
        self._bitcoin = self.add_crypto(ticker_crypto, market=Market.BITFINEX, leverage=2).symbol
        self._factors = [self.add_data(Fred, ticker, Resolution.DAILY).symbol for ticker in fred_factors]
        
        self._model = DecisionTreeRegressor(max_depth=tree_depth, random_state=1)
        self._scaler = StandardScaler()
        self.schedule.on(self.date_rules.month_start(self._equities[0]), self.time_rules.after_market_open(self._equities[0], 1), self.rebalance)
    

    def rebalance(self):
        # economic factors history
        factors = self.history(self._factors, timedelta(self._lookback_years*365), Resolution.DAILY)['value'].unstack(0).dropna()
        
        # monthly returns history
        label = self.history(self._equities + [self._bitcoin], timedelta(self._lookback_years*365), Resolution.DAILY, data_normalization_mode=DataNormalizationMode.TOTAL_RETURN)['close'].unstack(0).dropna().pct_change(21).shift(-21).dropna()
        
        prediction_by_symbol = pd.Series()
        for symbol in self._equities + [self._bitcoin]:
            asset_labels = label[symbol].dropna()
            idx = factors.index.intersection(asset_labels.index)
            self._model.fit(self._scaler.fit_transform(factors.loc[idx]), asset_labels.loc[idx])
            prediction = self._model.predict(self._scaler.transform([factors.iloc[-1]]))[0]
            if prediction > 0:
                prediction_by_symbol.loc[symbol] = prediction
        weight_by_symbol = 1.5 * prediction_by_symbol / prediction_by_symbol.sum()
        
        # respect the limit of cryptocurrency exposure
        if self._bitcoin in weight_by_symbol and weight_by_symbol.loc[self._bitcoin] > self._max_bitcoin_weight:
            weight_by_symbol.loc[self._bitcoin] = self._max_bitcoin_weight
            if len(weight_by_symbol) > 1:
                equities = [symbol for symbol in self._equities if symbol in weight_by_symbol]
                weight_by_symbol.loc[equities] = 1.5 * weight_by_symbol.loc[equities] / weight_by_symbol.loc[equities].sum()
        
        # adjust portfolio holdings
        self.set_holdings([PortfolioTarget(symbol, weight) for symbol, weight in weight_by_symbol.items()], True)