企业🤖AI Agent构建引擎,智能编排和调试,一键部署,支持私有化部署方案 广告
# Simple MACD > 来源:https://uqer.io/community/share/560a3007f9f06c597665ef61 MACD 公式算法: + 短期EMA: 短期(例如12日)的收盘价指数移动平均值(Exponential Moving Average) + 长期EMA: 长期(例如26日)的收盘价指数移动平均值(Exponential Moving Average) + DIF线: (Difference)短期EMA和长期EMA的离差值 + DEA线: (Difference Exponential Average)DIF线的M日指数平滑移动平均线 + MACD线: DIF线与DEA线的差 策略实现: + DIF从下而上穿过DEA,买进; + 相反,如DIF从上往下穿过DEA,卖出。 ## 策略中使用`talib`计算MACD ```py import pandas as pd import numpy as np import talib start = '2012-01-01' end = '2015-09-28' benchmark = 'HS300' universe = set_universe('HS300') capital_base = 1000000 refresh_rate = 5 ## 使用talib计算MACD的参数 short_win = 12 # 短期EMA平滑天数 long_win = 26 # 长期EMA平滑天数 macd_win = 20 # DEA线平滑天数 stk_num = 20 # 持仓股票数量 longest_history = 100 def initialize(account): account.universe = universe def handle_data(account): all_close_prices = account.get_attribute_history('closePrice', longest_history) long_bucket = [] short_bucket = [] for stk in account.universe: prices = all_close_prices[stk] if prices is None: continue try: # talib计算MACD macd_tmp = talib.MACD(prices, fastperiod=short_win, slowperiod=long_win, signalperiod=macd_win) DIF = macd_tmp[0] DEA = macd_tmp[1] MACD = macd_tmp[2] except: continue # 判断MACD走向 if MACD[-1] > 0 and MACD[-4] < 0: long_bucket.append(stk) elif MACD[-1] < 0 and MACD[-4] > 0: short_bucket.append(stk) hold = [] # 处理持仓中的股票 for stk in account.valid_secpos: # 在short_bucket中的,卖出 if stk in short_bucket: order_to(stk, 0) # 不在short_bucket中的,留着 else: hold.append(stk) buy_list = hold for stk in long_bucket: if stk not in hold: buy_list.append(stk) if len(buy_list) > 0: # 无论buy_list中有多少只股票,都将仓位分成stk_num份,每份买入一只股票 amount_per_stk = account.referencePortfolioValue/stk_num for stk in buy_list: amount = int(amount_per_stk/account.referencePrice[stk] / 100.0) * 100 order_to(stk, amount) ``` ![](https://box.kancloud.cn/2016-07-30_579cbb022d595.jpg)