python 钟摆理论的简单实现——完美躲过股灾和精准抄底

      针对之前神一般的钟摆理论，根据实际经验做出一些优化和分析，欢迎大家一起来讨论。

核心指标的优化：

       （1）根据格雷厄姆的成长价值公式进行估值，并且根据A股的实际情况或者市场情绪给予一定溢价：8%。 价值=当期(正常)利润×(7.5 + 两倍的预期年增长率)，

        （2）回测采用30分钟线的数据 回测时间：2011-02-01到2014-12-31 （没有任何大牛市） 起始资金100万 期末资金300万 总收益:200%

         回测详情的简单介绍：入选的股票基本上是二线蓝筹股和一下现金流充足的上市公司。总共涉及53个不同的股票。平均持股时间5天左右，胜率60%左右。交易流动性良好。（仅存在停牌造成的影响） 对投资结果的感悟：在没有牛市的情况下，投资必须谨慎乐观。市场总是充满机会又无处不是陷阱。普通散户必须在量化了风险和收益之后才能果断进场。没事可以天天盯盘，培养自己的盘感。但是！！！交易必须三思而后行。不求多，但求准；无需对各种利空，利好的消息草木皆兵，但务必戒贪；不必相信权威，但切忌不能心存侥幸。股市有涨有跌，淡定持股，开心享受福利收益。恒心加一颗平常心，三年照样获得2倍的资产增值。 

import numpy as np

start = '2011-02-01'                       
end = '2014-12-31'                         
benchmark = 'HS300'                      
commission = Commission(buycost=0.0008, sellcost=0.0018)  
universe = set_universe('HS300',date=end)


capital_base = 1000000                      
freq = 'm'                                 
refresh_rate = 30

max_percent_of_a_stock = 0.9 

def initialize(account):                   
    pass

def handle_data(account):                  
    global max_percent_of_a_stock
    buylist = []
    selist = []
    current_date = account.current_date
    current_date = current_date.strftime('%Y%m%d')

    overflow = 0.08 

    spread_rate = dict(spreadRateByIntrinsicValue(account, overflow=overflow, precedingDate=True))
            
    referencePortfolioValue = account.referencePortfolioValue

    
    stock_set_for_duotou = []
    stock_set_for_duotou.extend(account.avail_secpos.keys())
    stock_set_for_duotou.extend(spread_rate.keys())
    stock_set_for_duotou = list(set(stock_set_for_duotou))
    
    duotou_5_10_Map = duotou_5_10(current_date, stock_set_for_duotou, precedingDate=True)
    isButtom_Map = isButtom(current_date, stock_set_for_duotou, precedingDate=True, downPercent=0.3)
    
    for stock in account.avail_secpos.keys():
        if stock not in spread_rate and not duotou_5_10_Map.get(stock, False):
            selist.append(stock)
            
    for stock in selist:
        sell_value = account.referencePrice[stock]*account.valid_secpos[stock]
        order_to(stock, 0)
            
    for stock in spread_rate.keys():
        if stock not in account.valid_secpos:
            buylist.append(stock)
    
    for stock in buylist:
        
        if duotou_5_10_Map.get(stock, False) or isButtom_Map.get(stock, False):
            buy_value = min(referencePortfolioValue*max_percent_of_a_stock, account.cash/len(buylist))
            if buy_value/referencePortfolioValue >= 0.0001:
                order_pct(stock, buy_value/referencePortfolioValue)


def preceding_date(date):
    cal = DataAPI.TradeCalGet(exchangeCD=u"XSHG",beginDate='20110101',endDate=date,field=['calendarDate','isOpen'],pandas="1")
    cal = cal[cal['isOpen']==1]
    date = cal['calendarDate'].values[-2].replace('-','')
    return date

def duotou_5_10(date, stockList, precedingDate=True):
    if precedingDate:
        date = preceding_date(date)
    duotou = {}
    if stockList is None or len(stockList) == 0:
        return duotou
    kLine = DataAPI.MktStockFactorsOneDayGet(tradeDate=date,secID=stockList,field=['secID','MA5','MA10'],pandas="1")
    kLine = kLine.dropna()
    for stock, ma5, ma10 in zip(kLine['secID'].values, kLine['MA5'].values, kLine['MA10'].values):
        if ma5 > ma10:
            duotou[stock] = True
        else:
            duotou[stock] = False
    return duotou

def spreadRateByIntrinsicValue(account, overflow=0.0, precedingDate=True):
    stock_list = account.universe
    current_date = account.current_date
    date = current_date.strftime('%Y%m%d')
    if precedingDate:
        date = preceding_date(date)
    eq_EPS_EGRO = DataAPI.MktStockFactorsOneDayGet(tradeDate=date,secID=stock_list,field=['secID','EPS','EGRO'],pandas="1")
    eq_EPS_EGRO['Value'] = eq_EPS_EGRO['EPS']*(7.5+2*eq_EPS_EGRO['EGRO']/5)
    eq_EPS_EGRO = eq_EPS_EGRO.dropna()
    spread_rate = []
    for stock, intrinsic_value in zip(eq_EPS_EGRO['secID'].values, eq_EPS_EGRO['Value'].values):
        intrinsic_value = intrinsic_value*(1+overflow)
        reference_price = account.referencePrice[stock]
        if reference_price > 0 and reference_price < intrinsic_value:
            spread_rate.append((stock, (intrinsic_value-reference_price)/reference_price))
    return sorted(spread_rate, key=lambda k: k[-1], reverse=True)

def isButtom(date, stockList, precedingDate=True, downPercent=0.2):
    cal = DataAPI.TradeCalGet(exchangeCD=u"XSHG",beginDate='20110101',endDate=date,field=u"prevTradeDate",pandas="1")
    daysAhead = cal['prevTradeDate'].values[-20].replace('-','')
    if precedingDate:        
        date = cal['prevTradeDate'].values[-1].replace('-','')
    rs = {}
    if stockList is None or len(stockList) == 0:
        return rs
    dayInfo = DataAPI.MktEqudAdjGet(secID=stockList, beginDate=daysAhead, endDate=date ,field=['secID', 'openPrice', 'closePrice', 'preClosePrice'],pandas="1")
    dayInfo.dropna()
    for stock in stockList:
        stockDayInfo = dayInfo[dayInfo['secID']==stock]
        closePrices = stockDayInfo['closePrice'].values
        ma5 = np.mean(closePrices[-5:])
        ma10 = np.mean(closePrices[-10:])
        closePrice = closePrices[-1]
        maxClosePrice = np.max(closePrices)
        openPrice = stockDayInfo['openPrice'].values[-1]
        preClosePrice = stockDayInfo['preClosePrice'].values[-1]
        if (maxClosePrice-closePrice)/maxClosePrice > downPercent and closePrice < ma5 and ma5 < ma10 and (closePrice > openPrice or abs(closePrice-openPrice)/openPrice < 0.02) and abs(closePrice-preClosePrice)/preClosePrice<0.07:
            rs[stock] = True
        else:
            rs[stock] = False
    return rs

使用的平台：优矿量化交易系统