Overall Statistics |
Total Orders 491 Average Win 1.78% Average Loss -1.33% Compounding Annual Return 23.474% Drawdown 29.100% Expectancy 0.348 Start Equity 1000000 End Equity 3016059.63 Net Profit 201.606% Sharpe Ratio 0.753 Sortino Ratio 0.764 Probabilistic Sharpe Ratio 28.831% Loss Rate 42% Win Rate 58% Profit-Loss Ratio 1.34 Alpha 0.104 Beta 0.61 Annual Standard Deviation 0.212 Annual Variance 0.045 Information Ratio 0.347 Tracking Error 0.196 Treynor Ratio 0.261 Total Fees $9533.34 Estimated Strategy Capacity $150000000.00 Lowest Capacity Asset ETN R735QTJ8XC9X Portfolio Turnover 5.38% |
#region imports from AlgorithmImports import * import numpy as np from collections import deque import statsmodels.api as sm import statistics as stat import pickle #endregion class MonthlyRebalancingWithEarlyStop(QCAlgorithm): def initialize(self): self.set_start_date(2019, 3, 1) # Set Start Date self.set_end_date(2024, 8, 1) # Set End Date self.initial_cash = 1000000 self.set_cash(self.initial_cash) # Set Strategy Cash self.set_security_initializer(BrokerageModelSecurityInitializer( self.BrokerageModel, FuncSecuritySeeder(self.GetLastKnownPrices) )) self.p_lookback = 252 self.p_num_coarse = 200 self.p_num_fine = 70 self.p_num_long = 5 self.p_adjustment_step = 1.0 self.p_n_portfolios = 1000 self.p_short_lookback = 63 self.p_rand_seed = 13 self.p_adjustment_frequency = 'monthly' # Can be 'monthly', 'weekly', 'bi-weekly' # Change resolution to minute self.universe_settings.resolution = Resolution.DAILY self.set_benchmark('SPY') self._momp = {} # Dict of Momentum indicator keyed by Symbol self._lookback = self.p_lookback # Momentum indicator lookback period self._num_coarse = self.p_num_coarse # Number of symbols selected at Coarse Selection self._num_fine = self.p_num_fine # Number of symbols selected at Fine Selection self._num_long = self.p_num_long # Number of symbols with open positions self._rebalance = False self.current_holdings = set() # To track current holdings self.target_weights = {} # To store target weights self.adjustment_step = self.p_adjustment_step # Adjustment step for gradual transition self.first_trade_date = None self.next_adjustment_date = None # Metrics for no trades and profit tracking self.no_trade_days = 0 self.highest_profit = 0 self.lowest_profit = float('inf') self.monthly_starting_equity = 0 self.last_logged_month = None # 用于记录上次输出日志的月份 self.global_stop_loss_triggered = False # 标志是否已经触发了全局止损 self.add_universe(self._coarse_selection_function, self._fine_selection_function) def _coarse_selection_function(self, coarse): '''Drop securities which have no fundamental data or have too low prices. Select those with highest by dollar volume''' if self.next_adjustment_date and self.time < self.next_adjustment_date: return Universe.UNCHANGED self._rebalance = True if not self.first_trade_date: self.first_trade_date = self.time self.next_adjustment_date = self.get_next_adjustment_date(self.time) self._rebalance = True selected = sorted([x for x in coarse if x.has_fundamental_data and x.price > 5], key=lambda x: x.dollar_volume, reverse=True) return [x.symbol for x in selected[:self._num_coarse]] def _fine_selection_function(self, fine): '''Select security with highest market cap''' selected = sorted(fine, key=lambda f: f.market_cap, reverse=True) return [x.symbol for x in selected[:self._num_fine]] def on_data(self, data): # Update the indicator for symbol, mom in self._momp.items(): mom.update(self.time, self.securities[symbol].close) # Initialize starting equity if not already set (based on last adjustment) if self.monthly_starting_equity == 0: self.monthly_starting_equity = self.Portfolio.TotalPortfolioValue # Calculate the net profit percentage relative to the last adjustment current_portfolio_value = self.Portfolio.TotalPortfolioValue if self.monthly_starting_equity != 0: current_profit_pct_to_start = ((current_portfolio_value - self.monthly_starting_equity) / self.monthly_starting_equity) * 100 else: current_profit_pct_to_start = 0 # Track the highest profit percentage since the last adjustment self.highest_profit = max(self.highest_profit, current_profit_pct_to_start) # Calculate the percentage drop from the highest profit if self.highest_profit != 0: drop_pct = ((self.highest_profit - current_profit_pct_to_start) / self.highest_profit) * 100 else: drop_pct = 0 # Check for stop-loss based on the loss from the last adjustment (using minute data) if current_profit_pct_to_start <= -12 and not self.global_stop_loss_triggered: # 只在第一次触发止损时输出日志 current_date = self.Time.strftime('%Y-%m-%d %H:%M:%S') self.Debug(f"{current_date}: Liquidating all holdings due to a portfolio loss of {current_profit_pct_to_start:.2f}% (stop-loss from last adjustment).") self.Liquidate() self._rebalance = False self.global_stop_loss_triggered = True # 标记已触发全局止损 self.highest_profit = 0 self.monthly_starting_equity = 0 self.next_adjustment_date = self.get_next_adjustment_date(self.time) return # Check for take-profit based on the drop from the highest profit (using minute data) if self.highest_profit > 10 and drop_pct >= 10: # 如果从最高点的下降幅度达到或超过10% current_date = self.Time.strftime('%Y-%m-%d %H:%M:%S') self.Debug(f"{current_date}: Liquidating all holdings due to a {drop_pct:.2f}% drop in profit (take-profit).") self.Debug(f"{current_date}: Highest Net Profit: {self.highest_profit:.2f}% (from last adjustment)") self.Debug(f"{current_date}: Current Net Profit: {current_profit_pct_to_start:.2f}% (from last adjustment)") total_profit_pct = ((current_portfolio_value - self.initial_cash) / self.initial_cash) * 100 self.Debug(f"{current_date}: Total Net Profit: {total_profit_pct:.2f}% (from inception)") self.Liquidate() self._rebalance = False self.global_stop_loss_triggered = True # 同样在此标记为已触发,避免重复清仓 self.highest_profit = 0 self.monthly_starting_equity = 0 # 重置为下一个周期的初始 self.next_adjustment_date = self.get_next_adjustment_date(self.time) return # Log portfolio status on the first trading day of each month if self.Time.day == 1 and (self.Time.month != self.last_logged_month): current_date = self.Time.strftime('%Y-%m-%d %H:%M:%S') portfolio_value = self.Portfolio.TotalPortfolioValue net_profit = portfolio_value - self.initial_cash holdings_value = sum([sec.HoldingsValue for sec in self.Portfolio.Values if sec.Invested]) unrealized_profit = self.Portfolio.TotalUnrealizedProfit return_pct = (net_profit / self.initial_cash) * 100 self.Debug(f"{current_date}: Equity: ${portfolio_value:.2f} | Holdings: ${holdings_value:.2f} | Net Profit: ${net_profit:.2f} | Unrealized: ${unrealized_profit:.2f} | Return: {return_pct:.2f}%") self.last_logged_month = self.Time.month # Update the last logged month if not self._rebalance: return # Rebalance the portfolio and reset the stop-loss trigger if self._rebalance: self.global_stop_loss_triggered = False # 重置全局止损标志 # Rebalance logic... self._rebalance = False # Selects the securities with highest momentum sorted_mom = sorted([k for k,v in self._momp.items() if v.is_ready], key=lambda x: self._momp[x].current.value, reverse=True) selected = sorted_mom[:self._num_long] new_holdings = set(selected) # Only rebalance if the new selection is different from current holdings if new_holdings != self.current_holdings or self.first_trade_date == self.time: if len(selected) > 0: optimal_weights = self.optimize_portfolio(selected) self.target_weights = dict(zip(selected, optimal_weights)) self.current_holdings = new_holdings self.adjust_portfolio() self._rebalance = False self.next_adjustment_date = self.get_next_adjustment_date(self.time) def on_securities_changed(self, changes): # Clean up data for removed securities and Liquidate for security in changes.RemovedSecurities: symbol = security.Symbol if self._momp.pop(symbol, None) is not None: self.Liquidate(symbol, 'Removed from universe') for security in changes.AddedSecurities: if security.Symbol not in self._momp: self._momp[security.Symbol] = MomentumPercent(self._lookback) # Warm up the indicator with history price if it is not ready added_symbols = [k for k, v in self._momp.items() if not v.IsReady] history = self.History(added_symbols, 1 + self._lookback, Resolution.DAILY) history = history.close.unstack(level=0) for symbol in added_symbols: ticker = symbol.ID.ToString() if ticker in history: for time, value in history[ticker].dropna().items(): item = IndicatorDataPoint(symbol, time, value) self._momp[symbol].Update(item) def optimize_portfolio(self, selected_symbols): short_lookback = self.p_short_lookback returns = self.History(selected_symbols, short_lookback, Resolution.DAILY)['close'].unstack(level=0).pct_change().dropna() n_assets = len(selected_symbols) n_portfolios = self.p_n_portfolios results = np.zeros((3, n_portfolios)) weights_record = [] np.random.seed(self.p_rand_seed) for i in range(n_portfolios): weights = np.random.random(n_assets) weights /= np.sum(weights) portfolio_return = np.sum(returns.mean() * weights) * short_lookback portfolio_stddev = np.sqrt(np.dot(weights.T, np.dot(returns.cov() * short_lookback, weights))) downside_stddev = np.sqrt(np.mean(np.minimum(0, returns).apply(lambda x: x**2, axis=0).dot(weights))) sortino_ratio = portfolio_return / downside_stddev results[0,i] = portfolio_return results[1,i] = portfolio_stddev results[2,i] = sortino_ratio weights_record.append(weights) best_sortino_idx = np.argmax(results[2]) return weights_record[best_sortino_idx] def adjust_portfolio(self): current_symbols = set(self.Portfolio.Keys) target_symbols = set(self.target_weights.keys()) # Liquidate removed symbols removed_symbols = current_symbols - target_symbols for symbol in removed_symbols: self.Liquidate(symbol) # Adjust holdings for selected symbols for symbol, target_weight in self.target_weights.items(): current_weight = self.Portfolio[symbol].Quantity / self.Portfolio.TotalPortfolioValue if symbol in self.Portfolio else 0 adjusted_weight = current_weight * (1 - self.adjustment_step) + target_weight * self.adjustment_step self.SetHoldings(symbol, adjusted_weight) def get_next_adjustment_date(self, current_date, initial=False): if self.p_adjustment_frequency == 'weekly': return current_date + timedelta(days=7) elif self.p_adjustment_frequency == 'bi-weekly': return current_date + timedelta(days=14) elif self.p_adjustment_frequency == 'monthly': if initial: next_month = current_date.replace(day=1) + timedelta(days=32) return next_month.replace(day=1) next_month = current_date.replace(day=1) + timedelta(days=32) return next_month.replace(day=1) else: raise ValueError(f"Unsupported adjustment frequency: {self.p_adjustment_frequency}")