年初至今4倍，极致的Day Trading，56.8%胜率

策略作者: 水滴

```python
# 风险及免责提示：该策略由聚宽用户在聚宽社区分享，仅供学习交流使用。
# 原文一般包含策略说明，如有疑问请到原文和作者交流讨论。
# 原文网址：https://www.joinquant.com/post/38940
# 标题：年初至今4倍，极致的Day Trading，56.8%胜率
# 作者：Dr.QYQ

'''
优化说明:
    1.使用修正标准分
        rsrs_score的算法有：
            仅斜率slope，效果一般；
            仅标准分zscore，效果不错；
            修正标准分 = zscore * r2，效果最佳;
            右偏标准分 = 修正标准分 * slope，效果不错。
    2.将原策略的每次持有两只etf改成只买最优的一个，收益显著提高
    3.将每周调仓换成每日调仓，收益显著提高
    4.因为交易etf，所以手续费设为万分之三，印花税设为零，未设置滑点
    5.修改股票池中候选etf，删除银行，红利等收益较弱品种，增加纳指etf以增加不同国家市场间轮动的可能性
    6.根据研报，默认参数介已设定为最优
    7.加入防未来函数
    8.增加择时与选股模块的打印日志，方便观察每笔操作依据
'''

# 导入函数库
from jqdata import *
from jqlib.technical_analysis import *
import numpy as np
from datetime import datetime, timedelta
import pandas as pd

# 初始化函数

def initialize(context):
    # 设定沪深300作为基准
    set_benchmark('000300.XSHG')
    # 用真实价格交易
    set_option('use_real_price', True)
    # 打开防未来函数
    set_option("avoid_future_data", True)
    # 将滑点设置为0
    set_slippage(FixedSlippage(0.001))
    # 设置交易成本万分之三
    # set_order_cost(OrderCost(open_tax=0, close_tax=0, open_commission=0.0003, close_commission=0.0003, close_today_commission=0, min_commission=5),
    #               type='fund')
    # 股票类每笔交易时的手续费是：买入时无佣金，卖出时佣金万分之1.5，印花税0.1%, 每笔交易佣金最低扣5块钱
    set_order_cost(OrderCost(close_tax=0.001,
                   close_commission=0.00015, min_commission=5), type='stock')
    # 过滤order中低于error级别的日志
    log.set_level('order', 'error')
    # 初始化各类全局变量

# 动量轮动参数
    g.stock_num = 5  # 筛选的标的支数。
    g.stock_tobuy = 1  # 需要购买的股票数
    g.momentum_day = 29  # 最新动量参考最近momentum_day的
    g.num_days = 5  # 计算分数变化
    # rsrs择时参数
    g.ref_stock = '000300.XSHG'  # 用ref_stock做择时计算的基础数据
    g.N = 14  # 计算最新斜率slope，拟合度r2参考最近N天
    g.M = 600  # 计算最新标准分zscore，rsrs_score参考最近M天
    g.score_threshold = 0.7  # rsrs标准分指标阈值
    # 个股择时参数
    g.sec_data_num = 5  # 个股数据点数
    # g.take_profit = 0.12 # 移动止盈
    # ma择时参数
    g.mean_day = 7  # 计算ref_stock结束ma收盘价，参考最近mean_day
    # 计算初始ma收盘价，参考(mean_day + mean_diff_day)天前，窗口为mean_diff_day的一段时间
    g.mean_diff_day = 10
    g.slope_series = initial_slope_series()[:-1]  # 除去回测第一天的slope，避免运行时重复加入
    # 设置交易时间，每天运行
    run_daily(my_trade, time='09:31', reference_security='000300.XSHG')
    run_daily(check_lose, time='14:50', reference_security='000300.XSHG')
    # run_daily(check_profit, time='10:00')
    run_daily(print_trade_info, time='15:05', reference_security='000300.XSHG')

# 0-0 选取股票池

def get_stock_pool():
    # preday = str(date.today() - timedelta(1)) # get previous date
    # 从多个热门概念中选出市值在50亿以上,500亿以下的标的。
    concept_names = list(set([
        "虚拟现实",
        "元宇宙",
        "锂电池",
        "集成电路",
        "国产软件",
        "MiniLED",
        "智能穿戴",
        "智能电网",
        "智能医疗",
        "风电",
        "核电",
        "电力物联网",
        "电力改革",
        "量子通信",
        "互联网+",
        "光伏",
        "工业4.0",
        "特高压",
        "氟化工",
        "煤化工",
        "稀土永磁",
        "白酒",
        "煤炭",
        "钴",
        "盐湖提锂",
        "磷化工",
        "草甘膦",
        "航运",
        "第三代半导体",
        "太阳能",
        "柔性屏",
        "芯片",
        "新能源",
        "智能音箱",
        "苹果",
        "特斯拉",
        "宁德时代",
        "碳中和",
        "军工",
        "军民融合",
        "海工装备",
        "超级电容",
        "区块链",
        "边缘计算",
        "云计算",
        "数字货币",
        "人工智能",
        "汽车电子",
        "无人驾驶",
        "车联网",
        "网约车",
        "充电桩",
        "冷链物流",
        "OLED",
        "大飞机",
        "大数据",
        "燃料电池",
        "医疗器械",
        "生物疫苗",
        "生物医药",
        "辅助生殖",
        "健康中国",
        "基因测序",
        "超级真菌",
        "节能环保",
        "装配式建筑",
        "乡村振兴",
        "建筑节能",
        "文化传媒",
        "电子竞技",
        "网络游戏",
        "数据中心",
        "高端装备",
        '三胎',
        '养老',
        "稀缺资源",
        "稀土永磁",
        "新材料",
        "绿色电力"
    ]))

all_concepts = get_concepts()
    concept_codes = []
    for name in concept_names:
        #print(f'concept is:{name}')
        code = all_concepts[all_concepts['name'] == name].index[0]
        concept_codes.append(code)

all_concept_stocks = []

for concept in concept_codes:
        all_concept_stocks += get_concept_stocks(concept)

q = query(valuation.code).filter(valuation.market_cap >= 30,
                                     valuation.market_cap <= 1000, valuation.code.in_(all_concept_stocks))
    stock_df = get_fundamentals(q)
    stock_pool = [code for code in stock_df['code']]
    # 移除创业板和科创板标的
    stock_pool = [code for code in stock_pool if not (
        code.startswith('30') or code.startswith('688'))]
    stock_pool = filter_st_stock(stock_pool)  # 去除st
    stock_pool = filter_paused_stock(stock_pool)  # 去除停牌
    return stock_pool

# 1-1 选股模块-动量因子轮动
# 基于股票年化收益和判定系数打分,并按照分数从大到小排名

def get_rank(stock_pool, context):
    '''get rank score for stocks in stock pool'''
    send_info = []
    stock_dict_list = []
    for stock in stock_pool:
        score_list = []
        pre_dt = context.current_dt - timedelta(1)
        data = get_price(
            stock,
            end_date=context.current_dt,
            count=100,  # 多取几天以防数据不够
            frequency="120m",
            fields=["close"],
            skip_paused=True,
        )  # 最新的在最下面
        security_info = get_security_info(stock)
        stock_name = security_info.display_name
        # print(f'stock name {stock_name}')
        # 对于次新股票，可能没有数据，所以要drop NA
        data = data.dropna()
        # 收盘价
        y = data["log"] = np.log(data.close)
        # print(f'{len(y)} data points')
        # 分析的数据个数（天）
        x = data["num"] = np.arange(data.log.size)
        # 拟合 1 次多项式
        # y = kx + b, slope 为斜率 k，intercept 为截距 b
        # slope, intercept = np.polyfit(x, y, 1)
        # 直接连接首尾点计算斜率
        if len(y) < g.momentum_day + g.num_days:
            print("次新股，用所有数据")
            slope = (y.iloc[-1] - y.iloc[0]) / g.momentum_day   # 最新的在最上面
            # print(f'slope: {slope}\n')
            # 拟合出截距
            try:
                _, intercept = np.polyfit(x,y,1)
            except ValueError:
                print('Can not fit intercept, use first y value instead')
                intercept = y.iloc[0]
            
            # intercept = y.iloc[0]
            # (e ^ slope) ^ 250 - 1
            annualized_returns = math.pow(math.exp(slope), 250) - 1
            r_squared = 1 - (
                        sum((y - (slope * x + intercept)) ** 2) /
                        ((g.momentum_day - 1) * np.var(y, ddof=1))
                    )
            score = annualized_returns * np.abs(r_squared)
                    # print(f'score: {score}\n')
            score_list.append(score)
        else:
            slope = [(y.iloc[-1-D] - y.iloc[-g.momentum_day-D]) /
                     g.momentum_day for D in range(g.num_days)]  # 最新的在最上面
            # print(f'slope: {slope}\n')
            # intercept = [y.iloc[-g.momentum_day-D] for D in range(g.num_days)]
            # (e ^ slope) ^ 250 - 1
            for i in range(g.num_days):
                annualized_returns = math.pow(math.exp(slope[i]), 250) - 1
                if i == 0:  # 如果i=0，则前n天数据为df.iloc[-n::]
                    _, intercept = np.polyfit(x[-g.momentum_day-i::], y.iloc[-g.momentum_day-i::],1)
                    r_squared = 1 - (
                        sum((y.iloc[-g.momentum_day-i::] - (slope[i] * x[-g.momentum_day-i::] + intercept)) ** 2) /
                        ((g.momentum_day - 1) * np.var(y[-g.momentum_day-i::], ddof=1))
                    )
                    score = annualized_returns * np.abs(r_squared)
                    # print(f'score: {score}\n')
                    score_list.append(score)
                else:
                    _, intercept = np.polyfit(x[-g.momentum_day-i:-i], y.iloc[-g.momentum_day-i:-i],1)
                    r_squared = 1 - (
                        sum((y.iloc[-g.momentum_day-i:-i] - (slope[i] * x[-g.momentum_day-i:-i] + intercept)) ** 2) /
                        ((g.momentum_day - 1) * np.var(y[-g.momentum_day-i:-i], ddof=1))
                    )
                    score = annualized_returns * np.abs(r_squared)
                    # print(f'score: {score}\n')
                    score_list.append(score)
        stock_dict_tmp = {stock_name: score_list}
        stock_dict_list.append(stock_dict_tmp)
    # merge list of dictionaries into one dictionary
    stock_dict = {k: v for d in stock_dict_list for k, v in d.items()}
    # create pandas dataframe from dict
    stock_df = pd.DataFrame.from_dict(stock_dict, orient='index')
    stock_df['code'] = stock_pool
    # sort by latest score
    stock_df = stock_df.sort_values(by=[0], ascending=False)
    # get top num stocks
    stock_top_names = stock_df.index.values[:g.stock_num]
    stock_top_codes = stock_df['code'].values[:g.stock_num]
    # get all stock names and codes
    stock_names = stock_df.index.values
    stock_codes = stock_df['code'].values
    # print results
    print("#" * 30 + "候选" + "#" * 30)
    for name, code in zip(stock_top_names, stock_top_codes):
        print("{}({}):{}".format(name, code,
          stock_df[0][name]))
    print("#" * 64)
    return stock_top_names,stock_top_codes, stock_df

def rank_stock_change(df):
    '''rank_stock_plot
    line plot the score variation for *num_days* for each rank stock.
    '''
    df.drop(['code'],axis=1, inplace=True)
    stock_df = df.head(g.stock_num) # 只取头部
    rank_stock_dif = stock_df.diff(
        axis=1, periods=-1).dropna(axis=1)
    return rank_stock_dif

# 2-1 择时模块-计算线性回归统计值
# 对输入的自变量每日最低价x(series)和因变量每日最高价y(series)建立OLS回归模型,返回元组(截距,斜率,拟合度)
def get_ols(x, y):
    slope, intercept = np.polyfit(x, y, 1)
    r2 = 1 - (sum((y - (slope * x + intercept))**2) /
              ((len(y) - 1) * np.var(y, ddof=1)))
    return (intercept, slope, r2)

# 2-2 择时模块-设定初始斜率序列
# 通过前M日最高最低价的线性回归计算初始的斜率,返回斜率的列表

def initial_slope_series():
    data = attribute_history(g.ref_stock, g.N + g.M, '1d', ['high', 'low'])
    return [get_ols(data.low[i:i+g.N], data.high[i:i+g.N])[1] for i in range(g.M)]

# 2-3 择时模块-计算标准分
# 通过斜率列表计算并返回截至回测结束日的最新标准分

def get_zscore(slope_series):
    mean = np.mean(slope_series)
    std = np.std(slope_series)
    return (slope_series[-1] - mean) / std

# 2-4 择时模块-计算综合信号
# 1.获得rsrs与MA信号,rsrs信号算法参考优化说明，MA信号为一段时间两个端点的MA数值比较大小
# 2.信号同时为True时返回买入信号，同为False时返回卖出信号，其余情况返回持仓不变信号
# 3.改进：加入个股的卖点判据

def get_timing_signal(stock, rank_stock_diff, context):
    '''
    计算大盘信号: RSRS + MA
    '''
    # 计算MA信号
    close_data = attribute_history(
        g.ref_stock, g.mean_day + g.mean_diff_day, '1d', ['close'])
    today_MA = close_data.close[g.mean_diff_day:].mean()
    before_MA = close_data.close[:-g.mean_diff_day].mean()
    # 计算rsrs信号
    high_low_data = attribute_history(g.ref_stock, g.N, '1d', ['high', 'low'])
    intercept, slope, r2 = get_ols(high_low_data.low, high_low_data.high)
    g.slope_series.append(slope)
    rsrs_score = get_zscore(g.slope_series[-g.M:]) * r2  # 修正标准分
    print(
        f'today_MA is {today_MA}, before_MA is {before_MA}, rsrs score is {rsrs_score}')
    '''
    个股择时：
    1. MA5买卖点
    2. 3日斜率
    3. 移动止盈
    4. 效果不如不要。。。。
    5. 试试MCAD
    6. 试试KDJ
    '''
    # 计算个股x日斜率
    # close_data_sec = attribute_history(stock, g.sec_data_num, '1d', ['close'])
    # current_price = attribute_history(
    #    stock, 1, '1m', ['close'])  # get current stock price
    # MA_sec = close_data_sec.close.mean()
    # print(f'现价:{current_price.close[-1]}, MA{g.sec_data_num}: {MA_sec}')
    # 计算2日斜率
    # close_data_sec_ = attribute_history(stock, 2, '1d', ['close'])
    # y = close_data_sec_['log'] = np.log(close_data_sec_.close)
    # x = close_data_sec_['num'] = np.arange(close_data_sec_.log.size)
    # slope_sec,_ = np.polyfit(x,y,1)
    # print(f'Slope < 0: {slope_sec<0}')
    # 移动止盈
    # stock_data = attribute_history(stock,g.sec_data_num,'1d',['close','high'])
    # stock_price = attribute_history(stock, 1, '1m', 'close')
    # highest = stock_data.close.max()
    # profit = highest*(1-g.take_profit) # 移动止盈线
    # MACD
    # dif, dea, macd = MACD(stock, check_date=context.current_dt,
    #                      SHORT=12, LONG=29, MID=7, unit='1d')
    # KDJ
    # K, D, J = KDJ(stock, check_date=context.current_dt,
    #              unit='1d', N=9, M1=3, M2=3)
    # 连续num_days分数变化
    # stock_name = get_security_info(stock).display_name
    # print(f'今日自选股是{stock_name}')
    stock_dif = rank_stock_diff.loc[stock]
    # 如果连续num_days日下降即sig=num_days，卖出
    sig = np.sum((stock_dif < 0).astype(int))
    print(f'连续下降{sig}日')
    #综合判断所有信号:大盘信号 + 个股信号
    if sig < 2:  # rsrs_score > g.score_threshold and today_MA > before_MA and sig < 2 :
        print('BUY')
        return "BUY"
    # (rsrs_score < -g.score_threshold and today_MA < before_MA) or sig >= 2:
    elif sig >= 2:
        print('SELL')
        return "SELL"
    else:
        print('KEEP')
        return "KEEP"

# 3-1 过滤模块-过滤停牌股票
# 输入选股列表，返回剔除停牌股票后的列表
def filter_paused_stock(stock_list):
    current_data = get_current_data()
    return [stock for stock in stock_list if not current_data[stock].paused]

# 3-2 过滤模块-过滤ST及其他具有退市标签的股票
# 输入选股列表，返回剔除ST及其他具有退市标签股票后的列表

def filter_st_stock(stock_list):
    current_data = get_current_data()
    return [stock for stock in stock_list
            if not current_data[stock].is_st]

# 3-3 过滤模块-过滤涨停的股票
# 输入选股列表，返回剔除未持有且已涨停股票后的列表

def filter_limitup_stock(context, stock_list):
    last_prices = history(1, unit='1m', field='close',
                          security_list=stock_list)
    current_data = get_current_data()
    # 已存在于持仓的股票即使涨停也不过滤，避免此股票再次可买，但因被过滤而导致选择别的股票
    return [stock for stock in stock_list if stock in context.portfolio.positions.keys()
            or last_prices[stock][-1] < current_data[stock].high_limit]

# 3-4 过滤模块-过滤跌停的股票
# 输入股票列表，返回剔除已跌停股票后的列表

def filter_limitdown_stock(context, stock_list):
    last_prices = history(1, unit='1m', field='close',
                          security_list=stock_list)
    current_data = get_current_data()
    return [stock for stock in stock_list if stock in context.portfolio.positions.keys()
            or last_prices[stock][-1] > current_data[stock].low_limit]

# 4-1 交易模块-自定义下单
# 报单成功返回报单(不代表一定会成交),否则返回None,应用于
def order_target_value_(security, value):
    if value == 0:
        log.debug("Selling out %s" % (security))
    else:
        log.debug("Order %s to value %f" % (security, value))
    # 如果股票停牌，创建报单会失败，order_target_value 返回None
    # 如果股票涨跌停，创建报单会成功，order_target_value 返回Order，但是报单会取消
    # 部成部撤的报单，聚宽状态是已撤，此时成交量>0，可通过成交量判断是否有成交
    return order_target_value(security, value)

# 4-2 交易模块-开仓
# 买入指定价值的证券,报单成功并成交(包括全部成交或部分成交,此时成交量大于0)返回True,报单失败或者报单成功但被取消(此时成交量等于0),返回False

def open_position(security, value):
    order = order_target_value_(security, value)
    if order != None and order.filled > 0:
        return True
    return False

# 4-3 交易模块-平仓
# 卖出指定持仓,报单成功并全部成交返回True，报单失败或者报单成功但被取消(此时成交量等于0),或者报单非全部成交,返回False

def close_position(position):
    security = position.security
    if position.total_amount != 0:
        order = order_target_value_(security, 0)  # 可能会因停牌失败
    else:
        print(f'目前没有持有{get_security_info(security).display_name}')
        return position
    if order != None:
        if order.status == OrderStatus.held and order.filled == order.amount:
            return True
    return False

# 4-4 交易模块-调仓
# 当择时信号为买入时开始调仓，输入过滤模块处理后的股票列表，执行交易模块中的开平仓操作

def adjust_position(context, buy_stock, stock_position):
    # 根据股票数量分仓
    # 此处只根据可用金额平均分配购买，不能保证每个仓位平均分配
    position_count = len(context.portfolio.positions)
    if buy_stock not in context.portfolio.positions:
        if g.stock_tobuy > position_count:
            value = context.portfolio.cash / (g.stock_tobuy - position_count)
            if context.portfolio.positions[buy_stock].total_amount == 0:
                open_position(buy_stock, value)
            else:
                stock = list(context.portfolio.positions.keys())[
                    stock_position]
                log.info("[%s]已不在应买入列表中" % (stock))
                position = context.portfolio.positions[stock]
                close_position(position)
                # recount the securities after selling out old ones
                position_count = len(context.portfolio.positions)
                value = context.portfolio.cash / (g.stock_tobuy - position_count)
                if context.portfolio.positions[buy_stock].total_amount == 0:
                    open_position(buy_stock, value)
                
    else:
        log.info("[%s]已经持有无需重复买入" % (buy_stock))

# 4-5 交易模块-择时交易
# 结合择时模块综合信号进行交易

def my_trade(context):
    # if context.current_dt.minute != 35:
    #    return
    # 以下的代码每小时跑一次
    # 获取选股列表并过滤掉:st,st*,退市,涨停,跌停,停牌
    stock_pool = get_stock_pool()
    checkout_names, check_out_list, rank_stock = get_rank(stock_pool, context)
    print(check_out_list)
    print(f'check_out_list is {check_out_list}')
    rank_stock_diff = rank_stock_change(rank_stock)
    # print(f'stock_df is:\n {g.stock_df}')
    # check_out_list = filter_st_stock(check_out_list)
    # check_out_list = filter_limitup_stock(context, check_out_list)
    check_out_list = filter_limitdown_stock(context, check_out_list)
    check_out_list = filter_paused_stock(check_out_list)

if not check_out_list:  # empoty list is False
        print(f'Stock is limit up or limit down.')
    else:
        # check if the position still in the buying list
        for stock_pos in context.portfolio.positions:
            if stock_pos not in check_out_list[0:g.stock_tobuy]:
                log.info("旧龙头已不再买入列表，卖出")
                position = context.portfolio.positions[stock_pos]
                close_position(position)
        # print('今日自选股:{}'.format(get_security_info(check_out_list[0]).display_name))
        # 获取综合择时信号
        count = 0
        for stock_name, stock_code in zip(checkout_names[0:g.stock_tobuy], check_out_list[0:g.stock_tobuy]):
            print(f'今日自选股:{stock_name}')
            timing_signal = get_timing_signal(stock_name, rank_stock_diff, context)
            print(f'{stock_name} 今日择时信号:{timing_signal}')
            # 开始交易
            if timing_signal == 'SELL':
                position = context.portfolio.positions[stock_code]
                close_position(position)
            elif timing_signal == 'BUY' or timing_signal == 'KEEP':
                adjust_position(context, stock_code, count)
                count += 1
                # break # only buy one stock
            else:
                pass

# 4-6 交易模块-止损
# 检查持仓并进行必要的止损操作

def check_lose(context):
    for position in list(context.portfolio.positions.values()):
        securities = position.security
        cost = position.avg_cost
        price = position.price
        ret = 100*(price/cost-1)
        value = position.value
        amount = position.total_amount
        # 这里设定15%止损
        if ret <= -15:
            order_target_value(position.security, 0)
            print("！！！！！！触发止损信号: 标的={},标的价值={},浮动盈亏={}% ！！！！！！"
                  .format(securities, format(value, '.2f'), format(ret, '.2f')))
            log.info('亏死了，溜溜溜')

# 4-7 交易模块-止盈
# 根据移动止盈线止盈

def check_profit(context):
    for stock in context.portfolio.positions:
        position = context.portfolio.positions[stock]
        security = position.security
        price = attribute_history(security, 1, '1m', 'close')
        highest = attribute_history(security, g.sec_data_num, '1d', 'high')
        if price.close[-1] < highest.high.max()*(1-g.take_profit):
            print('触发止盈，卖卖卖')
            close_position(position)

# 5-1 复盘模块-打印
# 打印每日持仓信息

def print_trade_info(context):
    # 打印当天成交记录
    trades = get_trades()
    for _trade in trades.values():
        print('成交记录：'+str(_trade))
    # 打印账户信息
    for position in list(context.portfolio.positions.values()):
        securities = position.security
        cost = position.avg_cost
        price = position.price
        ret = 100*(price/cost-1)
        value = position.value
        amount = position.total_amount
        print('代码:{}'.format(securities))
        print('成本价:{}'.format(format(cost, '.2f')))
        print('现价:{}'.format(price))
        print('收益率:{}%'.format(format(ret, '.2f')))
        print('持仓(股):{}'.format(amount))
        print('市值:{}'.format(format(value, '.2f')))
    print('一天结束')
    print('———————————————————————————————————————分割线————————————————————————————————————————')

```

发布时间: 2025-02-26 20:58:18

文章分类

关于作者

水滴

注册时间: