策略化#13,#15,#16,#44,#83号因子

#015/alpha#015.py

@@ -0,0 +1,139 @@
+# -*- coding:utf-8 -*-
+
+from CloudQuant import MiniSimulator
+import numpy as np
+import pandas as pd
+
+username = 'Harvey_Sun'
+password = 'P948894dgmcsy'
+Strategy_Name = 'WQ_alpha#015_20_100'
+
+INIT_CAP = 100000000
+START_DATE = '20130101'
+END_DATE = '20161231'
+Fee_Rate = 0.001
+program_path = 'C:/cStrategy/'
+buy_number = 100
+period = 20
+s_or_b = False  # true表示选因子较小的股票，false表示选因子较大的股票
+
+
+def correlation(x, y):
+    upside = ((x - x.mean()) * (y - y.mean())).sum(skipna=False)
+    dnside = x.std() * y.std()
+    result = upside / dnside
+    return result
+
+
+def initial(sdk):
+    sdk.prepareData(['LZ_GPA_QUOTE_THIGH', 'LZ_GPA_QUOTE_TVOLUME',
+                     'LZ_GPA_CMFTR_CUM_FACTOR', 'LZ_GPA_SLCIND_STOP_FLAG'])
+    stock_pool = []
+    sdk.setGlobal('stock_pool', stock_pool)
+    step = period
+    sdk.setGlobal('step', step)
+
+
+def init_per_day(sdk):
+    step = sdk.getGlobal('step')
+    if step == period:
+        stock_list = sdk.getStockList()
+        high = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_THIGH')[-5:], columns=stock_list)
+        volume = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_TVOLUME')[-5:], columns=stock_list)
+        adj_f = pd.DataFrame(sdk.getFieldData('LZ_GPA_CMFTR_CUM_FACTOR')[-5:], columns=stock_list)
+        stop = pd.DataFrame(sdk.getFieldData('LZ_GPA_SLCIND_STOP_FLAG')[-6:], columns=stock_list)
+        volume.replace(0, np.nan, inplace=True)
+        adj_high = high * adj_f
+        adj_volume = volume / adj_f
+        # 计算昨天的factor
+        high_rank = adj_high.rank(axis=1, pct=True)
+        volume_rank = adj_volume.rank(axis=1, pct=True)
+        factor = pd.DataFrame(index=stock_list)
+        for i in [0, 1, 2]:
+            temp = correlation(high_rank[i:i+3], volume_rank[i:i+3])
+            factor[i] = temp
+        factor = -1 * factor.T.sum(axis=0, skipna=False)
+        # 选出排序靠前的股票
+        factor[stop.notnull().any()] = np.nan  # 剔除停牌的股票
+        factor.sort_values(ascending=s_or_b, inplace=True)
+        stock_pool = factor.index[:buy_number]
+        # 选出当天应该调仓的股票
+        position = sdk.getPositions()
+        position_dict = dict([i.code, i.optPosition] for i in position)
+        stock_to_buy = set(stock_pool) - set(position_dict.keys())
+        stock_to_sell = set(position_dict.keys()) - set(stock_pool)
+        sdk.setGlobal('position_dict', position_dict)
+        sdk.setGlobal('stock_to_buy', stock_to_buy)
+        sdk.setGlobal('stock_to_sell', stock_to_sell)
+    else:
+        pass
+
+
+def strategy(sdk):
+    step = sdk.getGlobal('step')
+    if step == period:
+        step = 1
+        today = sdk.getNowDate()
+        sdk.sdklog(today,'====================')
+
+        position_dict = sdk.getGlobal('position_dict')
+        stock_to_buy = sdk.getGlobal('stock_to_buy')
+        stock_to_sell = sdk.getGlobal('stock_to_sell')
+        quotes = sdk.getQuotes(list(stock_to_buy | stock_to_sell))
+
+        sell_orders = []
+        for stock in stock_to_sell:
+            if stock in quotes.keys():
+                price = quotes[stock].open
+                volume = position_dict[stock]
+                order = [stock, price, volume, -1]
+                sell_orders.append(order)
+        if sell_orders:
+            sdk.makeOrders(sell_orders)
+            sdk.sdklog('===sell orders===')
+            sdk.sdklog(np.array(sell_orders))
+        available_cash = sdk.getAccountInfo().availableCash
+        available_cash_one_stock = available_cash / len(stock_to_buy)
+        buy_orders = []
+        for stock in stock_to_buy:
+            if stock in quotes.keys():
+                price = quotes[stock].open
+                volume = int(available_cash_one_stock / (price * 100)) * 100
+                if volume > 0:
+                    order = [stock, price, volume, 1]
+                    buy_orders.append(order)
+        if buy_orders:
+            sdk.makeOrders(buy_orders)
+            sdk.sdklog('===buy orders===')
+            sdk.sdklog(np.array(buy_orders))
+    else:
+        step += 1
+    sdk.setGlobal('step', step)
+
+
+config = {
+    'username': username,
+    'password': password,
+    'initCapital': INIT_CAP,
+    'startDate': START_DATE,
+    'endDate': END_DATE,
+    'strategy': strategy,
+    'initial': initial,
+    'preparePerDay': init_per_day,
+    'feeRate': Fee_Rate,
+    'strategyName': Strategy_Name,
+    'logfile': '%s.log' % Strategy_Name,
+    'rootpath': program_path,
+    'executeMode': 'D',
+    'feeLimit': 5,
+    'cycle': 10,
+    'dealByVolume': True,
+    'allowForTodayFactors': ['LZ_GPA_SLCIND_STOP_FLAG']
+}
+
+if __name__ == "__main__":
+    # 在线运行所需代码
+    import os
+
+    config['strategyID'] = os.path.splitext(os.path.split(__file__)[1])[0]
+    MiniSimulator(**config).run()

#016/alpha#016.py

@@ -0,0 +1,136 @@
+# -*- coding:utf-8 -*-
+
+from CloudQuant import MiniSimulator
+import numpy as np
+import pandas as pd
+
+username = 'Harvey_Sun'
+password = 'P948894dgmcsy'
+Strategy_Name = 'WQ_alpha#016_10_100'
+
+INIT_CAP = 100000000
+START_DATE = '20130101'
+END_DATE = '20161231'
+Fee_Rate = 0.001
+program_path = 'C:/cStrategy/'
+buy_number = 100
+period = 10
+s_or_b = False  # true表示选因子较小的股票，false表示选因子较大的股票
+
+
+def correlation(x, y):
+    upside = ((x - x.mean()) * (y - y.mean())).sum(skipna=False)
+    dnside = x.std() * y.std()
+    result = upside / dnside
+    return result
+
+
+def initial(sdk):
+    sdk.prepareData(['LZ_GPA_QUOTE_THIGH', 'LZ_GPA_QUOTE_TVOLUME',
+                     'LZ_GPA_CMFTR_CUM_FACTOR', 'LZ_GPA_SLCIND_STOP_FLAG'])
+    stock_pool = []
+    sdk.setGlobal('stock_pool', stock_pool)
+    step = period
+    sdk.setGlobal('step', step)
+
+
+def init_per_day(sdk):
+    step = sdk.getGlobal('step')
+    if step == period:
+        stock_list = sdk.getStockList()
+        high = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_THIGH')[-5:], columns=stock_list)
+        volume = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_TVOLUME')[-5:], columns=stock_list)
+        adj_f = pd.DataFrame(sdk.getFieldData('LZ_GPA_CMFTR_CUM_FACTOR')[-5:], columns=stock_list)
+        stop = pd.DataFrame(sdk.getFieldData('LZ_GPA_SLCIND_STOP_FLAG')[-6:], columns=stock_list)
+        volume.replace(0, np.nan, inplace=True)
+        adj_high = high * adj_f
+        adj_volume = volume / adj_f
+        # 计算昨天的factor
+        high_rank = adj_high.rank(axis=1, pct=True)
+        volume_rank = adj_volume.rank(axis=1, pct=True)
+        factor = -1 * ((high_rank - high_rank.mean()) * (volume_rank - volume_rank.mean())).sum(skipna=False).rank(
+            pct=True)
+        # 选出排序靠前的股票
+        factor[stop.notnull().any()] = np.nan  # 剔除停牌的股票
+        factor.sort_values(ascending=s_or_b, inplace=True)
+        stock_pool = factor.index[:buy_number]
+        # 选出当天应该调仓的股票
+        position = sdk.getPositions()
+        position_dict = dict([i.code, i.optPosition] for i in position)
+        stock_to_buy = set(stock_pool) - set(position_dict.keys())
+        stock_to_sell = set(position_dict.keys()) - set(stock_pool)
+        sdk.setGlobal('position_dict', position_dict)
+        sdk.setGlobal('stock_to_buy', stock_to_buy)
+        sdk.setGlobal('stock_to_sell', stock_to_sell)
+    else:
+        pass
+
+
+def strategy(sdk):
+    step = sdk.getGlobal('step')
+    if step == period:
+        step = 1
+        today = sdk.getNowDate()
+        sdk.sdklog(today,'====================')
+
+        position_dict = sdk.getGlobal('position_dict')
+        stock_to_buy = sdk.getGlobal('stock_to_buy')
+        stock_to_sell = sdk.getGlobal('stock_to_sell')
+        quotes = sdk.getQuotes(list(stock_to_buy | stock_to_sell))
+
+        sell_orders = []
+        for stock in stock_to_sell:
+            if stock in quotes.keys():
+                price = quotes[stock].open
+                volume = position_dict[stock]
+                order = [stock, price, volume, -1]
+                sell_orders.append(order)
+        if sell_orders:
+            sdk.makeOrders(sell_orders)
+            sdk.sdklog('===sell orders===')
+            sdk.sdklog(np.array(sell_orders))
+        available_cash = sdk.getAccountInfo().availableCash
+        available_cash_one_stock = available_cash / len(stock_to_buy)
+        buy_orders = []
+        for stock in stock_to_buy:
+            if stock in quotes.keys():
+                price = quotes[stock].open
+                volume = int(available_cash_one_stock / (price * 100)) * 100
+                if volume > 0:
+                    order = [stock, price, volume, 1]
+                    buy_orders.append(order)
+        if buy_orders:
+            sdk.makeOrders(buy_orders)
+            sdk.sdklog('===buy orders===')
+            sdk.sdklog(np.array(buy_orders))
+    else:
+        step += 1
+    sdk.setGlobal('step', step)
+
+
+config = {
+    'username': username,
+    'password': password,
+    'initCapital': INIT_CAP,
+    'startDate': START_DATE,
+    'endDate': END_DATE,
+    'strategy': strategy,
+    'initial': initial,
+    'preparePerDay': init_per_day,
+    'feeRate': Fee_Rate,
+    'strategyName': Strategy_Name,
+    'logfile': '%s.log' % Strategy_Name,
+    'rootpath': program_path,
+    'executeMode': 'D',
+    'feeLimit': 5,
+    'cycle': 10,
+    'dealByVolume': True,
+    'allowForTodayFactors': ['LZ_GPA_SLCIND_STOP_FLAG']
+}
+
+if __name__ == "__main__":
+    # 在线运行所需代码
+    import os
+
+    config['strategyID'] = os.path.splitext(os.path.split(__file__)[1])[0]
+    MiniSimulator(**config).run()

#044/alpha#044.py

@@ -6,7 +6,7 @@ import pandas as pd
 
 username = 'Harvey_Sun'
 password = 'P948894dgmcsy'
-Strategy_Name = 'WQ_alpha#044_10_100'
+Strategy_Name = 'WQ_alpha#044_20_100'
 
 INIT_CAP = 100000000
 START_DATE = '20130101'
@@ -14,8 +14,15 @@ END_DATE = '20161231'
 Fee_Rate = 0.001
 program_path = 'C:/cStrategy/'
 buy_number = 100
-period = 10
-lowest = False  # 是否选因子值最小的股票，False为选取因子最大的股票
+period = 20
+s_or_b = False  # true表示选因子较小的股票，false表示选因子较大的股票
+
+
+def correlation(x, y):
+    upside = ((x - x.mean()) * (y - y.mean())).sum(skipna=False)
+    dnside = x.std() * y.std()
+    result = upside / dnside
+    return result
 
 
 def initial(sdk):
@@ -35,19 +42,17 @@ def init_per_day(sdk):
         volume = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_TVOLUME')[-5:], columns=stock_list)
         adj_f = pd.DataFrame(sdk.getFieldData('LZ_GPA_CMFTR_CUM_FACTOR')[-5:], columns=stock_list)
         stop = pd.DataFrame(sdk.getFieldData('LZ_GPA_SLCIND_STOP_FLAG')[-6:], columns=stock_list)
+        volume.replace(0, np.nan, inplace=True)
         adj_high = high * adj_f
         adj_volume = volume / adj_f
-
         # 计算昨天的factor
         volume_rank = adj_volume.rank(axis=1, pct=True)
-        factor = {}
-        for i in adj_high.columns:
-            factor[i] = pd.concat([adj_high[i], volume_rank[i]], axis=1).corr().iloc[0,1]
-        factor = pd.Series(factor)
+        factor = -1 * correlation(adj_high, volume_rank)
+        # 选出排序靠前的股票
         factor[stop.notnull().any()] = np.nan  # 剔除停牌的股票
-        factor.sort_values(ascending=lowest, inplace=True)
-        # 选取股票
-        stock_pool = factor.index[:buy_number]  # 选出的股票
+        factor.sort_values(ascending=s_or_b, inplace=True)
+        stock_pool = factor.index[:buy_number]
+        # 选出当天应该调仓的股票
         position = sdk.getPositions()
         position_dict = dict([i.code, i.optPosition] for i in position)
         stock_to_buy = set(stock_pool) - set(position_dict.keys())
@@ -62,7 +67,7 @@ def init_per_day(sdk):
 def strategy(sdk):
     step = sdk.getGlobal('step')
     if step == period:
-        step = 0
+        step = 1
         today = sdk.getNowDate()
         sdk.sdklog(today,'====================')
 

#083/alpha#083.py

@@ -0,0 +1,145 @@
+# -*- coding:utf-8 -*-
+
+from CloudQuant import MiniSimulator
+import numpy as np
+import pandas as pd
+
+username = 'Harvey_Sun'
+password = 'P948894dgmcsy'
+Strategy_Name = 'WQ_alpha#083_10_100'
+
+INIT_CAP = 100000000
+START_DATE = '20130101'
+END_DATE = '20161231'
+Fee_Rate = 0.001
+program_path = 'C:/cStrategy/'
+buy_number = 100
+period = 10
+s_or_b = False  # true表示选因子较小的股票，false表示选因子较大的股票
+
+
+def correlation(x, y):
+    upside = ((x - x.mean()) * (y - y.mean())).sum(skipna=False)
+    dnside = x.std() * y.std()
+    result = upside / dnside
+    return result
+
+
+def initial(sdk):
+    sdk.prepareData(['LZ_GPA_QUOTE_THIGH', 'LZ_GPA_QUOTE_TVOLUME',
+                     'LZ_GPA_QUOTE_TLOW', 'LZ_GPA_QUOTE_TCLOSE', 'LZ_GPA_QUOTE_TVALUE',
+                     'LZ_GPA_CMFTR_CUM_FACTOR', 'LZ_GPA_SLCIND_STOP_FLAG'])
+    stock_pool = []
+    sdk.setGlobal('stock_pool', stock_pool)
+    step = period
+    sdk.setGlobal('step', step)
+
+
+def init_per_day(sdk):
+    step = sdk.getGlobal('step')
+    if step == period:
+        stock_list = sdk.getStockList()
+        high = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_THIGH')[-7:], columns=stock_list)
+        low = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_TLOW')[-7:], columns=stock_list)
+        close = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_TCLOSE')[-7:], columns=stock_list)
+        volume = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_TVOLUME')[-7:], columns=stock_list)
+        value = pd.DataFrame(sdk.getFieldData('LZ_GPA_QUOTE_TVALUE')[-7:], columns=stock_list)
+        adj_f = pd.DataFrame(sdk.getFieldData('LZ_GPA_CMFTR_CUM_FACTOR')[-7:], columns=stock_list)
+        stop = pd.DataFrame(sdk.getFieldData('LZ_GPA_SLCIND_STOP_FLAG')[-8:], columns=stock_list)
+        volume.replace(0, np.nan, inplace=True)
+        adj_high = high * adj_f
+        adj_low = low * adj_f
+        adj_close = close * adj_f
+        adj_volume = volume / adj_f
+        # 计算昨天的factor
+        temp = (adj_high - adj_low) / adj_close.rolling(window=5).mean()
+        volume_rank = adj_volume.rank(axis=1, pct=True)
+        vwap = value / adj_volume
+        upside = temp.diff(2).rank(axis=1, pct=True).iloc[-1] * volume_rank.iloc[-1]
+        dnside = (temp / (vwap - adj_close)).iloc[-1]
+        factor = upside / dnside
+        factor[adj_close.isnull().any()] = np.nan
+        # 选出排序靠前的股票
+        factor[stop.notnull().any()] = np.nan  # 剔除停牌的股票
+        factor.sort_values(ascending=s_or_b, inplace=True)
+        stock_pool = factor.index[:buy_number]
+        # 选出当天应该调仓的股票
+        position = sdk.getPositions()
+        position_dict = dict([i.code, i.optPosition] for i in position)
+        stock_to_buy = set(stock_pool) - set(position_dict.keys())
+        stock_to_sell = set(position_dict.keys()) - set(stock_pool)
+        sdk.setGlobal('position_dict', position_dict)
+        sdk.setGlobal('stock_to_buy', stock_to_buy)
+        sdk.setGlobal('stock_to_sell', stock_to_sell)
+    else:
+        pass
+
+
+def strategy(sdk):
+    step = sdk.getGlobal('step')
+    if step == period:
+        step = 1
+        today = sdk.getNowDate()
+        sdk.sdklog(today, '====================')
+
+        position_dict = sdk.getGlobal('position_dict')
+        stock_to_buy = sdk.getGlobal('stock_to_buy')
+        stock_to_sell = sdk.getGlobal('stock_to_sell')
+        quotes = sdk.getQuotes(list(stock_to_buy | stock_to_sell))
+
+        sell_orders = []
+        for stock in stock_to_sell:
+            if stock in quotes.keys():
+                price = quotes[stock].open
+                volume = position_dict[stock]
+                order = [stock, price, volume, -1]
+                sell_orders.append(order)
+        if sell_orders:
+            sdk.makeOrders(sell_orders)
+            sdk.sdklog('===sell orders===')
+            sdk.sdklog(np.array(sell_orders))
+        available_cash = sdk.getAccountInfo().availableCash
+        available_cash_one_stock = available_cash / len(stock_to_buy)
+        buy_orders = []
+        for stock in stock_to_buy:
+            if stock in quotes.keys():
+                price = quotes[stock].open
+                volume = int(available_cash_one_stock / (price * 100)) * 100
+                if volume > 0:
+                    order = [stock, price, volume, 1]
+                    buy_orders.append(order)
+        if buy_orders:
+            sdk.makeOrders(buy_orders)
+            sdk.sdklog('===buy orders===')
+            sdk.sdklog(np.array(buy_orders))
+    else:
+        step += 1
+    sdk.setGlobal('step', step)
+
+
+config = {
+    'username': username,
+    'password': password,
+    'initCapital': INIT_CAP,
+    'startDate': START_DATE,
+    'endDate': END_DATE,
+    'strategy': strategy,
+    'initial': initial,
+    'preparePerDay': init_per_day,
+    'feeRate': Fee_Rate,
+    'strategyName': Strategy_Name,
+    'logfile': '%s.log' % Strategy_Name,
+    'rootpath': program_path,
+    'executeMode': 'D',
+    'feeLimit': 5,
+    'cycle': 10,
+    'dealByVolume': True,
+    'allowForTodayFactors': ['LZ_GPA_SLCIND_STOP_FLAG']
+}
+
+if __name__ == "__main__":
+    # 在线运行所需代码
+    import os
+
+    config['strategyID'] = os.path.splitext(os.path.split(__file__)[1])[0]
+    MiniSimulator(**config).run()

.gitignore

@@ -1,2 +1,3 @@
 *.csv
-*.log
+*.log
+*.png

WQ_Alphas_Analysis.py

@@ -110,7 +110,6 @@ def factor_cal(data, factor_id, no_zdt=False, no_st=False, no_new=False, save=Fa
         volume[rolling_null(close, 63)] = np.nan
         cap[rolling_null(close, 63)] = np.nan
     
-    
     adj_OPEN = OPEN * price_adj_f
     adj_close = close * price_adj_f
     adj_high = high * price_adj_f
@@ -119,7 +118,6 @@ def factor_cal(data, factor_id, no_zdt=False, no_st=False, no_new=False, save=Fa
     returns = adj_close.pct_change()
     vwap = value / adj_volume
     
-
     print '开始计算'
     if factor_id == '001':
         returns[returns < 0] = returns.rolling(window=20).std(skipna=False)
@@ -232,7 +230,7 @@ def factor_cal(data, factor_id, no_zdt=False, no_st=False, no_new=False, save=Fa
         factor[rolling_null(close, 10)] = np.nan
         factor[rolling_stop(stop, 10)] = np.nan
     elif factor_id == '019':
-        factor = -1 * np.sign(adj_close.diff(7)) * (1 + returns.rolling(window=250).sum())
+        factor = -1 * np.sign(adj_close.diff(7)) * ((1 + returns.rolling(window=250).sum()).rank(axis=1, pct=True) + 1)
         factor[rolling_null(close, 250)] = np.nan
         factor[rolling_stop(stop, 8)] = np.nan
     elif factor_id == '020':
@@ -1139,7 +1137,9 @@ data = {'high': high,
         'price_adj_f': price_adj_f}
 
 
-factor = factor_cal(data, '004', no_zdt=False, no_st=False, no_new=False)
+factor = factor_cal(data, '019', no_zdt=False, no_st=False, no_new=True, save=True)
+save_factor(factor, '015', no_zdt=True)
+
 factor_006 = factor_cal(data, '006', no_zdt=False)
 factors = pd.concat([factor.stack(), factor_006.stack()], axis=1)
 days = factors.index.levels[0]
@@ -1179,7 +1179,7 @@ a['group_mean'].mean().plot()
 a['group_mean']['Factor Average'].plot(figsize=(14, 7))
 a['group_return'].columns.name
 
-a = _analysis(stt, data, 'close-close', stock_number=50, ascending=True, period=10)
+a = _analysis(stt, data, 'close-close', stock_number=10, ascending=False, period=10)
 a['mean_return'].mean()
 a['factor_mean'].min()
 (a['mean_return'] + 1).cumprod().plot(figsize=(14, 7))