在bigquant的主頁有這麼一個代碼,使用的是機器學習,止盈止損大盤風控。從回測來看效果不錯,但實際上很不一定。
粘貼在這裏,方便以後學習。
學習特徵列表包括下面的幾項,多數是技術面,只有一個基本面
return_5
return_10
return_20
avg_amount_0/avg_amount_5
avg_amount_5/avg_amount_20
rank_avg_amount_0/rank_avg_amount_5
rank_avg_amount_5/rank_avg_amount_10
rank_return_0
rank_return_5
rank_return_10
rank_return_0/rank_return_5
rank_return_5/rank_return_10
pe_ttm_0
# 本代碼由可視化策略環境自動生成 2019年5月6日 20:12
# 本代碼單元只能在可視化模式下編輯。您也可以拷貝代碼,粘貼到新建的代碼單元或者策略,然後修改。
# 回測引擎:每日數據處理函數,每天執行一次
def m4_handle_data_bigquant_run(context, data):
#獲取當日日期
today_date = data.current_dt.strftime('%Y-%m-%d')
#大盤風控模塊,讀取風控數據
benckmark_risk=context.benckmark_risk.ix[today_date].values[0]
#當risk爲1時,市場有風險,全部平倉,不再執行其它操作
if benckmark_risk > 0:
position_all = context.portfolio.positions.keys()
for i in position_all:
context.order_target(i, 0)
print(today_date,'大盤風控止損觸發,全倉賣出')
return
# 按日期過濾得到今日的預測數據
ranker_prediction = context.ranker_prediction[
context.ranker_prediction.date == data.current_dt.strftime('%Y-%m-%d')]
# 1. 資金分配
# 平均持倉時間是hold_days,每日都將買入股票,每日預期使用 1/hold_days 的資金
# 實際操作中,會存在一定的買入誤差,所以在前hold_days天,等量使用資金;之後,儘量使用剩餘資金(這裏設置最多用等量的1.5倍)
is_staging = context.trading_day_index < context.options['hold_days'] # 是否在建倉期間(前 hold_days 天)
cash_avg = context.portfolio.portfolio_value / context.options['hold_days']
cash_for_buy = min(context.portfolio.cash, (1 if is_staging else 1.5) * cash_avg)
cash_for_sell = cash_avg - (context.portfolio.cash - cash_for_buy)
positions = {e.symbol: p.amount * p.last_sale_price
for e, p in context.perf_tracker.position_tracker.positions.items()}
#---------------------------START:止贏止損模塊(含建倉期)--------------------
# 新建當日止贏止損股票列表是爲了handle_data 策略邏輯部分不再對該股票進行判斷
current_stopwin_stock=[]
current_stoploss_stock = []
today_date = data.current_dt.strftime('%Y-%m-%d')
positions_stop={e.symbol:p.cost_basis
for e,p in context.portfolio.positions.items()}
if len(positions_stop)>0:
for i in positions_stop.keys():
stock_cost=positions_stop[i]
stock_market_price=data.current(context.symbol(i),'price')
# 賺3元且爲可交易狀態就止盈
if stock_market_price-stock_cost-1>3 and data.can_trade(context.symbol(i)) and not context.has_unfinished_sell_order(i):
context.order_target_percent(context.symbol(i),0)
current_stopwin_stock.append(i)
# 虧10%並且爲可交易狀態就止損
if stock_market_price/stock_cost-1 <= -0.1 and data.can_trade(context.symbol(i)) and not context.has_unfinished_sell_order(i):
context.order_target_percent(context.symbol(i),0)
current_stoploss_stock.append(i)
if len(current_stopwin_stock)>0:
print(today_date,'止盈股票列表',current_stopwin_stock)
if len(current_stoploss_stock)>0:
print(today_date,'止損股票列表',current_stoploss_stock)
#--------------------------END: 止贏止損模塊-----------------------------
#--------------------------START:持有固定天數賣出(不含建倉期)---------------
current_stopdays_stock = []
today = data.current_dt
today_date = data.current_dt.strftime('%Y-%m-%d')
# 不是建倉期(在前hold_days屬於建倉期)
if not is_staging:
equities = {e.symbol: p for e, p in context.portfolio.positions.items() if p.amount>0}
if len(equities)>0:
for i in equities:
sid = equities[i].sid # 交易標的
#如果上面的止盈止損已經賣出過了,就不要重複賣出以防止產生空單
if i in current_stopwin_stock+current_stoploss_stock:
continue
# 今天和上次交易的時間相隔hold_days就全部賣出 datetime.timedelta(context.options['hold_days'])也可以換成自己需要的天數,比如datetime.timedelta(5)
if today-equities[i].last_sale_date>=datetime.timedelta(5) and data.can_trade(context.symbol(i)) and not context.has_unfinished_sell_order(equities[i]):
context.order_target_percent(sid, 0)
current_stopdays_stock.append(i)
if len(current_stopdays_stock)>0:
print(today_date,'固定天數賣出列表',current_stopdays_stock)
#-------------------------------END:持有固定天數賣出--------------------------
# 2. 生成賣出訂單:hold_days天之後纔開始賣出;對持倉的股票,按機器學習算法預測的排序末位淘汰
if not is_staging and cash_for_sell > 0:
equities = {e.symbol: e for e, p in context.perf_tracker.position_tracker.positions.items()}
instruments = list(reversed(list(ranker_prediction.instrument[ranker_prediction.instrument.apply(
lambda x: x in equities and not context.has_unfinished_sell_order(equities[x]))])))
for instrument in instruments:
#防止多個止損條件同時滿足,出現多次賣出產生空單
if instrument not in current_stopdays_stock+current_stopwin_stock+current_stoploss_stock:
context.order_target(context.symbol(instrument), 0)
cash_for_sell -= positions[instrument]
else:
cash_for_sell -= positions[instrument]
if cash_for_sell <= 0:
break
# 3. 生成買入訂單:按機器學習算法預測的排序,買入前面的stock_count只股票
buy_cash_weights = context.stock_weights
buy_instruments_tmp = list(ranker_prediction.instrument)
#防止賣出後再次買入
buy_instruments=[k for k in buy_instruments_tmp if k not in current_stopdays_stock+current_stopwin_stock+current_stoploss_stock][:len(buy_cash_weights)]
max_cash_per_instrument = context.portfolio.portfolio_value * context.max_cash_per_instrument
for i, instrument in enumerate(buy_instruments):
cash = cash_for_buy * buy_cash_weights[i]
if cash > max_cash_per_instrument - positions.get(instrument, 0):
# 確保股票持倉量不會超過每次股票最大的佔用資金量
cash = max_cash_per_instrument - positions.get(instrument, 0)
if cash > 0:
context.order_value(context.symbol(instrument), cash)
# 回測引擎:準備數據,只執行一次
def m4_prepare_bigquant_run(context):
#在數據準備函數中一次性計算每日的大盤風控條件相比於在handle中每日計算風控條件可以提高回測速度
# 多取50天的數據便於計算均值(保證回測的第一天均值不爲Nan值),其中context.start_date和context.end_date是回測指定的起始時間和終止時間
start_date= (pd.to_datetime(context.start_date) - datetime.timedelta(days=50)).strftime('%Y-%m-%d')
benckmark_data=D.history_data(instruments=['000001.SZA'], start_date=start_date, end_date=context.end_date,fields=['close'])
#計算指數5日漲幅
benckmark_data['ret5']=benckmark_data['close']/benckmark_data['close'].shift(5)-1
#計算大盤風控條件,如果5日漲幅小於-5%則設置風險狀態risk爲1,否則爲0
benckmark_data['risk'] = np.where(benckmark_data['ret5']<-0.04,1,0)
#修改日期格式爲字符串(便於在handle中使用字符串日期索引來查看每日的風險狀態)
benckmark_data['date']=benckmark_data['date'].apply(lambda x:x.strftime('%Y-%m-%d'))
#設置日期爲索引
benckmark_data.set_index('date',inplace=True)
#把風控序列輸出給全局變量context.benckmark_risk
context.benckmark_risk=benckmark_data[['risk']]
# 回測引擎:初始化函數,只執行一次
def m4_initialize_bigquant_run(context):
# 加載預測數據
context.ranker_prediction = context.options['data'].read_df()
# 系統已經設置了默認的交易手續費和滑點,要修改手續費可使用如下函數
context.set_commission(PerOrder(buy_cost=0.0003, sell_cost=0.0013, min_cost=5))
# 預測數據,通過options傳入進來,使用 read_df 函數,加載到內存 (DataFrame)
# 設置買入的股票數量,這裏買入預測股票列表排名靠前的5只
stock_count = 5
# 每隻的股票的權重,如下的權重分配會使得靠前的股票分配多一點的資金,[0.339160, 0.213986, 0.169580, ..]
#context.stock_weights = T.norm([1 / math.log(i + 2) for i in range(0, stock_count)])
#改爲等權重配置
context.stock_weights = [1 / stock_count for i in range(0, stock_count)]
# 設置每隻股票佔用的最大資金比例
context.max_cash_per_instrument = 0.2
context.options['hold_days'] = 5
m1 = M.instruments.v2(
start_date='2010-01-01',
end_date='2015-01-01',
market='CN_STOCK_A',
instrument_list='',
max_count=0
)
m2 = M.advanced_auto_labeler.v2(
instruments=m1.data,
label_expr="""# #號開始的表示註釋
# 0. 每行一個,順序執行,從第二個開始,可以使用label字段
# 1. 可用數據字段見 https://bigquant.com/docs/data_history_data.html
# 添加benchmark_前綴,可使用對應的benchmark數據
# 2. 可用操作符和函數見 `表達式引擎 <https://bigquant.com/docs/big_expr.html>`_
# 計算收益:5日收盤價(作爲賣出價格)除以明日開盤價(作爲買入價格)
shift(close, -5) / shift(open, -1)
# 極值處理:用1%和99%分位的值做clip
clip(label, all_quantile(label, 0.01), all_quantile(label, 0.99))
# 將分數映射到分類,這裏使用20個分類
all_wbins(label, 20)
# 過濾掉一字漲停的情況 (設置label爲NaN,在後續處理和訓練中會忽略NaN的label)
where(shift(high, -1) == shift(low, -1), NaN, label)
""",
start_date='',
end_date='',
benchmark='000300.SHA',
drop_na_label=True,
cast_label_int=True
)
m3 = M.input_features.v1(
features="""# #號開始的表示註釋
# 多個特徵,每行一個,可以包含基礎特徵和衍生特徵
return_5
return_10
return_20
avg_amount_0/avg_amount_5
avg_amount_5/avg_amount_20
rank_avg_amount_0/rank_avg_amount_5
rank_avg_amount_5/rank_avg_amount_10
rank_return_0
rank_return_5
rank_return_10
rank_return_0/rank_return_5
rank_return_5/rank_return_10
pe_ttm_0
"""
)
m15 = M.general_feature_extractor.v7(
instruments=m1.data,
features=m3.data,
start_date='',
end_date='',
before_start_days=0
)
m16 = M.derived_feature_extractor.v3(
input_data=m15.data,
features=m3.data,
date_col='date',
instrument_col='instrument',
drop_na=False,
remove_extra_columns=False
)
m7 = M.join.v3(
data1=m2.data,
data2=m16.data,
on='date,instrument',
how='inner',
sort=False
)
m13 = M.dropnan.v1(
input_data=m7.data
)
m6 = M.stock_ranker_train.v5(
training_ds=m13.data,
features=m3.data,
learning_algorithm='排序',
number_of_leaves=30,
minimum_docs_per_leaf=1000,
number_of_trees=20,
learning_rate=0.1,
max_bins=1023,
feature_fraction=1,
m_lazy_run=False
)
m9 = M.instruments.v2(
start_date=T.live_run_param('trading_date', '2015-01-01'),
end_date=T.live_run_param('trading_date', '2017-01-01'),
market='CN_STOCK_A',
instrument_list='',
max_count=0
)
m17 = M.general_feature_extractor.v7(
instruments=m9.data,
features=m3.data,
start_date='',
end_date='',
before_start_days=0
)
m18 = M.derived_feature_extractor.v3(
input_data=m17.data,
features=m3.data,
date_col='date',
instrument_col='instrument',
drop_na=False,
remove_extra_columns=False
)
m14 = M.dropnan.v1(
input_data=m18.data
)
m8 = M.stock_ranker_predict.v5(
model=m6.model,
data=m14.data,
m_lazy_run=False
)
m4 = M.trade.v4(
instruments=m9.data,
options_data=m8.predictions,
start_date='',
end_date='',
handle_data=m4_handle_data_bigquant_run,
prepare=m4_prepare_bigquant_run,
initialize=m4_initialize_bigquant_run,
volume_limit=0.025,
order_price_field_buy='open',
order_price_field_sell='close',
capital_base=1000000,
auto_cancel_non_tradable_orders=True,
data_frequency='daily',
price_type='後復權',
product_type='股票',
plot_charts=True,
backtest_only=False,
benchmark='000300.SHA'
)