簡述

這里通過加大了對沖beta的比例，使得整個策略算是更進(jìn)一步了。

效果

代碼

# 導(dǎo)入函數(shù)庫 import statsmodels.api as sm from statsmodels import regression import numpy as np import pandas as pd import time from datetime import date import scipy.stats as stats# 一鍵回測說明： # 百度聚寬-》注冊賬號-》我的策略里面創(chuàng)建策略-》復(fù)制代碼到里面 # 右邊回測 開始時間：2017-1-1 終止時間：今天 資金：10000000# 初始化函數(shù)，設(shè)定基準(zhǔn)等等 def initialize(context):g.tc= 2 # 調(diào)倉頻率# 下面是框架固定部分，不需要修改 g.N = 20 # 持倉數(shù)目g.t = 0 # 記錄運行的天數(shù)g.weight_list = [1] # 因子的權(quán)重參數(shù)log.set_level('order', 'error')set_option('use_real_price', True) # 用真實價格交易set_slippage(FixedSlippage(0)) # 將滑點設(shè)置為0set_commission(PerTrade(buy_cost=0.0000, sell_cost=0.000, min_cost=0)) # 手續(xù)費設(shè)置為0# set_benchmark('000905.XSHG') #中證500為業(yè)績基準(zhǔn)set_benchmark('000300.XSHG') # 滬深300為業(yè)績基準(zhǔn)# 選股范圍為全市場選股：上證+深證股票# g.stockrange= get_index_stocks('000001.XSHG')+get_index_stocks('399106.XSHE')g.stockrange = get_index_stocks('000300.XSHG')init_cash_ = context.portfolio.starting_cash/3g.init_cash = init_cash_# 操作期貨set_subportfolios([SubPortfolioConfig(cash=init_cash_, type='stock'),SubPortfolioConfig(cash=init_cash_ * 2 , type='futures')])g.futures_rate = 0.01 # 100倍的杠桿set_option('futures_margin_rate', g.futures_rate)# 每當(dāng)g.calDays == 0的時候就更新一下初始值g.orignalValue = 0g.TotalDays = 30g.calDays = 0g.levels = [0.01, 0.032, 0.055]# -20 -8 -1 0 1 8 20# 全買 買2/3 買1/3 賣1/6 買1/6 ...g.betaName = 'IF9999.CCFX'## 每根日線運行一次 def handle_data(context, data):if g.t == 0:# 設(shè)置可行股票池：用set_feasible_stocks函數(shù)剔除當(dāng)前或者計算樣本期間停牌的股票g.all_stocks = pickStock(context, g.stockrange)previousStr = str(context.previous_date)# adjustedBeta# adjustedBeta(context)stock_sort = get_all_cleaned_factor_ranked(g.all_stocks, g.weight_list, previousStr)# 調(diào)倉rebalance_position(context, stock_sort)# order_stock_sell(context,data,stock_sort)# order_stock_buy(context,data,stock_sort) # g.t+=1 # speed upg.t = (g.t + 1) % g.tcdef get_all_cleaned_factor_ranked(stocks, weight_list, previousStr):# value = my_apha1(stocks)value = my_alpha2(stocks)value = (-value).argsort()[:g.N]return list(map(lambda x: stocks[x], value))''' factor1 '''def single_alpha1(stock):df = get_price(stock, count=26, fields=['close'])returns = (df[1:] - df[:-2]) / df[:-2] # 25 linesvalue = [0] * 5for i in range(5):if np.array(returns).tolist()[-1 - i] < 0:value[i] = returns.ix[-20 - i:-1 - i].std() ** 2else:value[i] = df['close'][-1 - i] ** 2return 5 - np.array(value).argmax()def my_apha1(stocks):value = list(map(single_alpha1, stocks))MAX = max(value)MIN = min(value)value = (np.array(value) - MIN) / (MAX - MIN)return value''' END factor1 '''''' factor2 '''def single_alpha2(stock):df = get_price(stock, count=8, fields=['open', 'close', 'volume'])volume = np.log(df['volume'].tolist())delta_volume = volume[2:] - volume[:-2]close = df['close']open_ = df['open']rate = (close - open_) / open_rate = np.array(rate)[2:]return delta_volume, ratedef my_alpha2(stocks):values = list(map(single_alpha2, stocks))T_values = list(zip(*values))T_values = list(map(doubleListRank, T_values))T_values = list(map(list, zip(*T_values)))values = np.array(map(Neg_correlation, T_values))return valuesdef Neg_correlation(adlist):return -stats.pearsonr(np.array(adlist[0]), np.array(adlist[1]))[0]def doubleListRank(ddlist):ddlist6 = list(zip(*ddlist))ddlist6 = list(map(rank, ddlist6))return list(map(list, zip(*ddlist6)))def rank(alist):MAX = max(alist)MIN = min(alist)return (np.array(alist) - MIN) / (MAX - MIN)# (-1 * correlation(rank(delta(log(volume), 2)), rank(((close - open)/ open)), 6)) # 6天以來的相關(guān)系數(shù)() 兩個數(shù)值分別是 # rank(delta(log(volume), 2))''' END factor2 '''''' 初步篩選股票 '''# pick stocks def pickStock(context, stocks):# universe = set_feasible_stocks(stocks)universe = filter_specials(stocks, context)# 過濾上市時間小于60天的股票for stock in universe:days_public = (context.current_dt.date() - get_security_info(stock).start_date).daysif days_public < 60:universe.remove(stock)g.lenth = len(universe)return universe# 過濾停牌的股票 def set_feasible_stocks(stock_list):current_data = get_current_data()stock_list = [stock for stock in stock_list if not current_data[stock].paused] # 不考慮停盤的股票return stock_list# 將多因子的dataframe進(jìn)行排序，并且將有空值的行去掉 def rank_stock(all_factor, weight_list):C = len(all_factor.columns)ranked = all_factor.iloc[:, 0].rank() * weight_list[0]if C > 1:for j in range(1, C):ranked = all_factor.iloc[:, j].rank() * weight_list[j] + rankedranked = pd.DataFrame(ranked)ranked.columns = ['rank']one_sort = ranked.sort('rank', ascending=g.ascending)stock_sort = one_sort.index[:g.N]return stock_sort#過濾退市，停牌，STdef filter_specials(stock_list,context):curr_data = get_current_data()stock_list = [stock for stock in stock_list if \(not curr_data[stock].paused) # 未停牌and (not curr_data[stock].is_st) # 非STand ('ST' not in curr_data[stock].name)and ('*' not in curr_data[stock].name)and ('退' not in curr_data[stock].name)and (curr_data[stock].low_limit < curr_data[stock].day_open < curr_data[stock].high_limit) ]return stock_list''' END 初步篩選股票 '''''' 調(diào)倉 '''def rebalance_position(context, stocks_list):current_holding = context.subportfolios[0].positions.keys()stocks_to_sell = list(set(current_holding) - set(stocks_list))# 賣出bulk_orders(stocks_to_sell, 0)total_value = context.subportfolios[0].total_valuerebalance_beta(context)# 買入 bulk_orders(stocks_list, total_value / len(stocks_list))def rebalance_beta(context):if g.calDays == 0:g.orignalValue = context.subportfolios[0].total_valueelse:total_value = context.subportfolios[0].total_valuerate = (total_value - g.orignalValue) / g.orignalValuemoney = context.subportfolios[1].transferable_cash / g.futures_rateif rate >= 0 and rate < g.levels[0]:bete_order(money / 6.0)elif rate <= -g.levels[0] and rate >= -g.levels[1]:bete_order(money / 3)elif rate < -g.levels[1] and rate >= -g.levels[2]:bete_order(money * 2.0 / 3)elif rate < -g.levels[2]:bete_order(money)elif rate < 0 and rate >= -g.levels[0]:bete_order(-money / 6.0)elif rate > g.levels[0] and rate <= g.levels[1]:bete_order(-money / 3)elif rate > g.levels[1] and rate <= g.levels[2]:bete_order(-money * 2.0 / 3)elif rate > g.levels[2]:bete_order(-money)print (context.subportfolios[0].total_value - g.init_cash) / g.init_cashg.calDays = (g.calDays + 1) % g.TotalDays# 批量買賣股票 def bulk_orders(stocks_list, target_value):for i in stocks_list:order_target_value(i, target_value, pindex=0)def bete_order(money):amount = int( money / get_current_data()[g.betaName].last_price)order_target(g.betaName, amount , side='long', pindex=1) # 做空''' END 調(diào)倉 '''''' 調(diào)倉初版 '''##獲得賣出信號，并執(zhí)行賣出操作 # 輸入：context, data，已排序股票列表stock_sort-list類型 # 輸出：none def order_stock_sell(context, data, stock_sort):# 對于不需要持倉的股票，全倉賣出for stock in context.portfolio.positions:# 除去排名前g.N個股票（選股！）if stock not in stock_sort:stock_sell = stockorder_target_value(stock_sell, 0)# 獲得買入信號，并執(zhí)行買入操作 # 輸入：context, data，已排序股票列表stock_sort-list類型 # 輸出：none def order_stock_buy(context, data, stock_sort):# 對于需要持倉的股票，按分配到的份額買入for stock in stock_sort:stock_buy = stockorder_target_value(stock_buy, g.everyStock)''' END 調(diào)倉初版 ''' 《新程序員》：云原生和全面數(shù)字化實踐50位技術(shù)專家共同創(chuàng)作，文字、視頻、音頻交互閱讀

總結(jié)

以上是生活随笔為你收集整理的【量化投资】策略七（聚宽）的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯，歡迎將生活随笔推薦給好友。