def get_recently_trade_date(dt=datetime.date.today()):
date_path = ConfigUtils.get_stock("STOCKS_DATE")
print(date_path)
if date_path and os.path.exists(date_path):
trade_dates = pd.read_csv(ConfigUtils.get_stock("STOCKS_DATE"),
header=0)
trade_date_dict = trade_dates.set_index(
"calendar_date")['is_trading_day'].to_dict()
tmp_dt = str(dt)
if tmp_dt in trade_date_dict:
if trade_date_dict[tmp_dt] == 1:
return tmp_dt
else:
dt_num = 1
dt_pass = str(dt - datetime.timedelta(days=dt_num))
while dt_pass in trade_date_dict and trade_date_dict[
dt_pass] == 0:
dt_num += 1
dt_pass = str(dt - datetime.timedelta(days=dt_num))
if dt_pass in trade_date_dict:
return dt_pass
print("Date Is Not Exist !!!, Reload Trade Dates. ")
else:
print("Date Is Not Exist, Reloading Trade Dates. ")
init_trade_date()
print("Date Loading Finish. ")
return None
def __init__(self, trn, trn_label, val, val_label, name):
self.trn = trn
self.val = val
self.trn_label = trn_label
self.val_label = val_label
self.path = os.path.join(ConfigUtils.get_model("MODELPATH"), name)
if not os.path.exists(ConfigUtils.get_model("MODELPATH")):
os.makedirs(ConfigUtils.get_model("MODELPATH"))
self.param = {
'num_leaves': 60,
'n_estimatores': 3000,
'min_data_in_leaf': 30,
'objective': 'multiclass',
'num_class': 21,
'lambda_l1': 0.1,
'lambda_l2': 0.2,
'max_depth': 5,
'learning_rate': 0.01,
"min_child_samples": 20,
"boosting": "gbdt",
# "feature_fraction": 0.45,
"bagging_freq": 1,
"bagging_fraction": 0.8,
"bagging_seed": 11,
"nthread": 30,
'metric': 'multi_logloss',
"random_state": 1111,
"verbosity": -1
}
def period_trades():
pth = os.path.join(ConfigUtils.get_stock("DATA_DIR"))
m_pth = os.path.join(ConfigUtils.get_stock("DATA_M_DIR"))
if not os.path.exists(m_pth):
os.mkdir(m_pth)
for fname in os.listdir(pth):
try:
df = pd.read_csv(os.path.join(pth, fname))
df['date'] = pd.to_datetime(df['date'])
df = df.set_index('date')
df = df.sort_index(ascending=True)
df_period = df.to_period('M')
grouped = df_period.groupby('date')
results = pd.DataFrame(columns=[
'date', 'code', 'open', 'preclose', 'close', 'high', 'low',
'volume', 'turn', 'amount', 'pctChg'
])
for name, group in grouped:
code = group.iloc[0]['code']
open = group.iloc[0]['open']
preclose = group.iloc[0]['preclose']
close = group.iloc[-1]['close']
high = group['high'].max()
low = group['low'].min()
volume = group['volume'].sum()
turn = group['turn'].sum()
amount = group['amount'].sum()
if pd.isna(preclose):
pctChg = (close - open) / preclose
else:
pctChg = (close - preclose) / preclose
# print(name, code, preclose, open, close, high, low, volume, turn, amount, pctChg)
series = pd.Series(
{
'date': name,
'code': code,
'open': open,
'preclose': preclose,
'close': close,
'high': high,
'low': low,
'volume': volume,
'turn': turn,
'amount': amount,
'pctChg': pctChg
},
name=name)
results = results.append(series)
results.reset_index(drop=True)
results.to_csv(os.path.join(m_pth, fname), index=False)
print("{} Month K done, {}".format(
fname,
datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')))
except pd.errors.EmptyDataError:
print("Empty file:", fname)
except KeyError:
print("KeyError file:", fname)
def prepare():
dirs = [ConfigUtils.get_stock("DATA_DIR"), ConfigUtils.get_stock("DB_DIR")]
for dir in dirs:
if os.path.exists(dir):
clean_files()
return
else:
os.makedirs(dir)
def clean_files():
for the_file in os.listdir(ConfigUtils.get_stock("DATA_DIR")):
file_path = os.path.join(ConfigUtils.get_stock("DATA_DIR"), the_file)
try:
if os.path.isfile(file_path):
os.unlink(file_path)
except Exception as e:
print(e)
def persist(strategy, results):
if not os.path.exists(ConfigUtils.get_stock("OUTPUT")):
os.mkdir(ConfigUtils.get_stock("OUTPUT"))
with open(ConfigUtils.get_stock("OUTPUT") + "/" + strategy + ".txt",
'w') as wf:
for e in results:
wf.write(e[0] + "-" + e[1] + "-" + str(e[2]) + '\n')
def get_trade_dates(start='2010-01-01', end='2020-12-30'):
date_path = ConfigUtils.get_stock("STOCKS_DATE")
print(date_path)
trade_dates = pd.read_csv(ConfigUtils.get_stock("STOCKS_DATE"), header=0)
trade_dates = trade_dates[(trade_dates['calendar_date'] >= start)
& (trade_dates['calendar_date'] <= end) &
(trade_dates['is_trading_day'] == 1)]
# print(trade_dates['calendar_date'].values)
return trade_dates['calendar_date'].values
def read_data(code_name, root=ConfigUtils.get_stock("DATA_DIR")):
code = code_name[0]
name = code_name[1]
df = None
file_name = str(code) + '_' + str(name) + '.csv'
file_path = root + "/" + file_name
if os.path.exists(file_path):
try:
df = pd.read_csv(file_path)
except pd.errors.EmptyDataError as e:
df = None
pass
if df is not None and not df.empty:
# print(df.keys())
df["open"] = df['open'].astype(float)
df["high"] = df["high"].astype(float)
df["low"] = df["low"].astype(float)
df["close"] = df["close"].astype(float)
df["preclose"] = df["preclose"].astype(float)
df["volume"] = df["volume"].astype(float)
df["amount"] = df["amount"].astype(float)
df["turn"] = df["turn"].astype(float)
df["pctChg"] = df["pctChg"].astype(float)
return df
return None
def reptile_dynamic_wz():
wz_path = os.path.join(ConfigUtils.get_stock("DATA"), "wz_top_20.csv")
url = 'http://www.waizichigu.com/Handler/Handler6.ashx'
rs = post_wzcg(None, url=url)
if rs:
pprint.pprint(rs)
df = pd.DataFrame(rs)
df.to_csv(wz_path, index=None, mode='w+')
def reptile_by_date(
trade_date=date.today() + timedelta(-1),
size=50,
):
if not os.path.exists(ConfigUtils.get_stock("WZ_DIR")):
os.mkdir(ConfigUtils.get_stock("WZ_DIR"))
wz_path = os.path.join(ConfigUtils.get_stock("WZ_DIR"),
"wz_{}.csv".format(str(trade_date)))
total = get_foreign_num(trade_date)
print("total:", total)
if total > 0:
df = pd.DataFrame()
for i in range(0, int((total + size - 1) / size)):
outJson = post_wzcg(trade_date, page=(i + 1), rows=size)
print(outJson)
if outJson and "rows" in outJson:
row = pd.DataFrame(outJson['rows'])
df = df.append(row, ignore_index=True)
time.sleep(1)
df.to_csv(wz_path, index=None, mode='w+')
def parse_position_change(code='600600'):
if not os.path.exists(ConfigUtils.get_stock("POS_DIR")):
os.mkdir(ConfigUtils.get_stock("POS_DIR"))
pos_path = os.path.join(ConfigUtils.get_stock("POS_DIR"),
"position_{}.csv".format(code))
html = reptile_position(code=code)
html = (html.replace('<br>', '')).replace('<br/>', '')
soup = BeautifulSoup(html, 'lxml') # html.parser是解析器，也可是lxml
# print(soup.prettify())
inner = soup.select('table[class="tb0td1"]')
if len(inner) > 0:
tr_rows = inner[0].select('tr[height]')
tb_head = tr_rows[0].select('td')
heads = [ct.text for ct in tb_head]
# print(heads)
content = []
for row in tr_rows[1:]:
cols = row.select('td')
tmp_rows = []
for ct in cols:
[s.extract() for s in ct(['a', 'br'])]
tmp_rows.append(ct.text.replace('\n', '').replace('\r', ''))
# print(tmp_rows)
content.append(tmp_rows)
with open(pos_path, mode="ab+") as csvfile:
csvfile.write(codecs.BOM_UTF8)
with open(pos_path, 'a+', newline='', encoding='utf-8') as csvfile:
writer = csv.writer(csvfile)
# 先写入columns_name
writer.writerow(heads)
# 写入多行用writerows
writer.writerows(content)
def scan_stocks():
pd_names = pd.read_csv(ConfigUtils.get_stock("STOCK_NAME"))
# 数据库获取沪、深、中小板
zxb_stocks = [s.code for s in get_mark_stocks(mark='zxb')]
hsb_stocks = [s.code for s in get_mark_stocks(mark='hsb')]
ssb_stocks = [s.code for s in get_mark_stocks(mark='ssb')]
candidate_stocks = zxb_stocks + hsb_stocks + ssb_stocks
print("total stock size:", len(candidate_stocks))
records = get_records(dt='2020-12-15', direction='limit', period='d1')
print("涨停数量:", len(records))
for index, row in tqdm(pd_names.iterrows()):
code = row['code']
name = row['code_name']
if stock_utils.is_jiucaiban(code):
continue
if code[3:] not in candidate_stocks:
# print(name, code)
continue
code_name = (code, name)
df = stock_utils.read_data(code_name)
df.reset_index(drop=True, inplace=True)
if len(df) < 60:
continue
df = df.tail(n=5)
df.reset_index(drop=True, inplace=True)
# 涨停分析
close = df.iloc[-1]['close']
chg = df.iloc[-1]['pctChg']
volume = df.iloc[-1]['volume']
volume_rate = (volume - df.iloc[-2]['volume']) / df.iloc[-2]['volume']
chgs = [c for c in df['pctChg']] # 涨幅序列
vols = [v / df.iloc[0]['volume'] for v in df['volume']] # 量比序列
feat = standardization(chgs[:-1]) + standardization(vols[:-1])
feat = standardization(chgs[:-1])
# print(feat)
tmp_count = 0
for record in records:
tmp_chg = json.loads(record.extra)['chgs'][:-1]
tmp_vol = json.loads(record.extra)['vols'][:-1]
tmp_feat = standardization(tmp_chg) + standardization(tmp_vol)
tmp_feat = standardization(tmp_chg)
dist = cosine_dist(feat, tmp_feat)
if dist > 0.9:
# print(name, record.name, chg, "sim:", dist)
tmp_count += 1
if tmp_count > 5:
print(code, name, chg)
def init_trade_date():
# 登陆系统 ####
lg = bs.login()
# 显示登陆返回信息
print('login respond error_code:' + lg.error_code)
print('login respond error_msg:' + lg.error_msg)
# 获取交易日信息 ####
st = ConfigUtils.get_stock("START_DATE")
et = ConfigUtils.get_stock("END_DATE")
print(st, et)
rs = bs.query_trade_dates(start_date=st, end_date=et)
print('query_trade_dates respond error_code:' + rs.error_code)
print('query_trade_dates respond error_msg:' + rs.error_msg)
# 打印结果集 ####
data_list = []
while (rs.error_code == '0') & rs.next():
# 获取一条记录，将记录合并在一起
data_list.append(rs.get_row_data())
result = pd.DataFrame(data_list, columns=rs.fields)
result.to_csv(ConfigUtils.get_stock("STOCKS_DATE"), index=False)
# 结果集输出到csv文件 ####import ujson
bs.logout()
def get_stocks(config=None):
if config:
data = xlrd.open_workbook(config)
table = data.sheets()[0]
rows_count = table.nrows
codes = table.col_values(0)[1:rows_count - 1]
names = table.col_values(1)[1:rows_count - 1]
return list(zip(codes, names))
else:
data_files = os.listdir(ConfigUtils.get_stock('DATA_DIR'))
stocks = []
for file in data_files:
code_name = file.split(".")[0]
code = code_name.split("-")[0]
name = code_name.split("-")[1]
appender = (code, name)
stocks.append(appender)
return stocks
def get_all_stock_names():
# 登陆系统 ####
lg = bs.login()
# 显示登陆返回信息
print('login respond error_code:' + lg.error_code + ', error_msg:' +
lg.error_msg)
dt = stock_utils.get_recently_trade_date()
dt = '2020-08-03'
k_rs = bs.query_all_stock(day=dt)
print(k_rs)
data_list = []
while (k_rs.error_code == '0') & k_rs.next():
# 获取一条记录，将记录合并在一起
data_list.append(k_rs.get_row_data())
result = pd.DataFrame(data_list, columns=k_rs.fields)
print(result.tail())
result.to_csv(ConfigUtils.get_stock("STOCK_NAME"), index=False)
print("init all stock names")
bs.logout()
def get_all_stock_industries():
lg = bs.login()
print('login respond error_code:' + lg.error_code)
print('login respond error_msg:' + lg.error_msg)
# 获取行业分类数据
rs = bs.query_stock_industry(date='2020-08-01')
# rs = bs.query_stock_basic(code_name="浦发银行")
print('query_stock_industry error_code:' + rs.error_code)
print('query_stock_industry respond error_msg:' + rs.error_msg)
# 打印结果集
industry_list = []
while (rs.error_code == '0') & rs.next():
# 获取一条记录，将记录合并在一起
industry_list.append(rs.get_row_data())
result = pd.DataFrame(industry_list, columns=rs.fields)
# 结果集输出到csv文件
result.to_csv(ConfigUtils.get_stock("STOCK_INDUSTRY"), index=False)
print(result)
# 登出系统
bs.logout()
def hs300_stocks():
# 登陆系统
lg = bs.login()
# 显示登陆返回信息
print('login respond error_code:' + lg.error_code)
print('login respond error_msg:' + lg.error_msg)
# 获取证券基本资料
rs = bs.query_hs300_stocks()
print('query_hs300 error_code:' + rs.error_code)
print('query_hs300 error_msg:' + rs.error_msg)
# 打印结果集
hs300_stocks = []
while (rs.error_code == '0') & rs.next():
# 获取一条记录，将记录合并在一起
hs300_stocks.append(rs.get_row_data())
result = pd.DataFrame(hs300_stocks, columns=rs.fields)
# 结果集输出到csv文件
# result.to_csv("D:/hs300_stocks.csv", encoding="gbk", index=False)
result.to_csv(ConfigUtils.get_stock("HS_300_STOCK_NAME"), index=False)
print(result.tail())
print("init all hs300 stock names")
def update_all_trades():
try:
et = stock_utils.get_recently_trade_date()
st = ConfigUtils.get_stock("START_DATE")
print(st, et)
# 登陆系统 ####
lg = bs.login()
# 显示登陆返回信息
print('login respond error_code:' + lg.error_code + ', error_msg:' +
lg.error_msg)
print("stock—name path:", ConfigUtils.get_stock("STOCK_NAME"))
pd_names = pd.read_csv(ConfigUtils.get_stock("STOCK_NAME"))
for index, row in pd_names.iterrows():
code = row['code']
name = row['code_name']
k_rs = bs.query_history_k_data_plus(
code,
ConfigUtils.get_stock("STOCK_FIELDS"),
start_date=st,
end_date=et)
data_list = []
while (k_rs.error_code == '0') & k_rs.next():
# 获取一条记录，将记录合并在一起
data_list.append(k_rs.get_row_data())
result = pd.DataFrame(data_list, columns=k_rs.fields)
print(result.tail())
if not os.path.exists(ConfigUtils.get_stock("DATA_DIR")):
os.makedirs(ConfigUtils.get_stock("DATA_DIR"))
result.to_csv(os.path.join(ConfigUtils.get_stock("DATA_DIR"),
str(code) + "_" + str(name) + ".csv"),
index=False)
print("Downloading :" + code + " , name :" + name + ", " +
datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
bs.logout()
except IOError as e:
print("Update Data Error ", e)
def gen_records(day_ago=0):
records = list()
pd_names = pd.read_csv(ConfigUtils.get_stock("STOCK_NAME"))
# 数据库获取沪、深、中小板
zxb_stocks = [s.code for s in get_mark_stocks(mark='zxb')]
hsb_stocks = [s.code for s in get_mark_stocks(mark='hsb')]
ssb_stocks = [s.code for s in get_mark_stocks(mark='ssb')]
candidate_stocks = zxb_stocks + hsb_stocks + ssb_stocks
print("total stock size:", len(candidate_stocks))
for index, row in tqdm(pd_names.iterrows()):
code = row['code']
name = row['code_name']
if stock_utils.is_jiucaiban(code):
continue
if code[3:] not in candidate_stocks:
continue
code_name = (code, name)
df = stock_utils.read_data(code_name)
df = df.head(n=len(df) - day_ago) # n天前 新高/低入口
df.reset_index(drop=True, inplace=True)
if len(df) < 60:
continue
for period in [500, 120, 60, 20, 10, 5]:
df = df.tail(n=period)
df.reset_index(drop=True, inplace=True)
idxMax = df['close'].idxmax(axis=0)
idxMin = df['close'].idxmin(axis=0)
# print(idxMax, idxMin)
dateMax = df.iloc[idxMax]['date']
dateMin = df.iloc[idxMin]['date']
date = df.iloc[-1]['date']
volume = df.iloc[-1]['volume']
record_date = datetime.datetime.strptime(date, '%Y-%m-%d').date()
d_period_inc = (df.iloc[-1]['close'] -
df.iloc[0]['close']) / df.iloc[0]['close']
volume_avg = df['volume'].sum() / period
amount_avg = df['amount'].sum() / period
period_inc_avg = d_period_inc / period
extraJson = {
'volume_avg': volume_avg,
'amount_avg': amount_avg,
'inc_avg': period_inc_avg
}
flag, direction = False, None
r = Record(name, code, record_date, df.iloc[-1]['close'],
"d{}".format(period), volume, d_period_inc)
r.set_extra(json.dumps(extraJson))
if df.iloc[-1]['date'] == dateMax:
flag, direction = True, 'up'
if df.iloc[-1]['date'] == dateMin and 'ST' not in name:
flag, direction = True, 'down'
if flag:
r.set_direction(direction)
records.append(r)
# 涨停分析
close = df.iloc[-1]['close']
chg = df.iloc[-1]['pctChg']
volume = df.iloc[-1]['volume']
volume_rate = (volume - df.iloc[-2]['volume']) / df.iloc[-2]['volume']
if float(chg) >= 9:
chgs = [c for c in df['pctChg']] # 涨幅序列
vols = [v / df.iloc[0]['volume'] for v in df['volume']] # 量比序列
extraJson = {
'volume_rate': volume_rate,
'chgs': chgs,
'vols': vols
}
r = Record(name, code, record_date, close, "d1", volume, chg)
r.set_direction(direction='limit')
r.set_extra(json.dumps(extraJson))
records.append(r)
return records
def generate_feature():
start_dt = '2019-01-01'
date_info = pd.read_csv(ConfigUtils.get_stock("STOCKS_DATE"),
encoding='utf-8')
company_info = pd.read_csv(ConfigUtils.get_stock("HS_300_STOCK_NAME"),
encoding='utf-8')
# 时序处理
dt = datetime.date.today()
tmp_list = sorted([
row['calendar_date']
for idx, row in date_info.iterrows() if row['calendar_date'] < str(dt)
and row['calendar_date'] >= start_dt and row['is_trading_day'] == 1
],
reverse=True)
date_map = dict(zip(tmp_list, range(len(tmp_list))))
# 读取股票交易信息
remove_stock = []
tmp_list = []
for i, row in tqdm.tqdm(company_info.iterrows()):
code, name = row["code"], row["code_name"]
path = os.path.join(ConfigUtils.get_stock("DATA_DIR"),
code + "_" + name + ".csv")
if not os.path.exists(path):
continue
tmp_df = pd.read_csv(path)
tmp_df = tmp_df[tmp_df.date >= start_dt]
if filter_stock(tmp_df, code, name):
remove_stock.append(code)
continue
tmp_df = tmp_df.sort_values('date', ascending=True).reset_index()
tmp_list.append(tmp_df)
stock_info = pd.concat(tmp_list)
ts_code_map = dict(
zip(stock_info['code'].unique(), range(stock_info['code'].nunique())))
stock_info = stock_info.reset_index()
stock_info['ts_code_id'] = stock_info['code'].map(ts_code_map)
stock_info.drop('index', axis=1, inplace=True)
stock_info['trade_date_id'] = stock_info['date'].map(date_map)
stock_info['ts_date_id'] = (
10000 + stock_info['ts_code_id']) * 10000 + stock_info['trade_date_id']
# 特征工程
col = ['close', 'open', 'high', 'low']
feature_col = []
for tmp_col in col:
stock_info[tmp_col + '_' +
'transform'] = (stock_info[tmp_col] - stock_info['preclose']
) / stock_info['preclose']
feature_col.append(tmp_col + '_' + 'transform')
print('stock_info 占据内存约: {:.2f} GB'.format(
sys.getsizeof(stock_info) / (1024**3)))
# 提取前5天收盘价与今天收盘价的盈亏比, 增加10, 20, 30, 40, 50, 60, 120, 180
for i in [
0, 1, 2, 3, 4, 9, 19, 29, 39, 49, 59, 79, 99, 119, 149, 179, 199,
249
]:
tmp_df = pd.DataFrame(stock_info,
columns=['ts_date_id', 'close'],
dtype='float32')
tmp_df = tmp_df.rename(
columns={'close': 'close_shift_{}'.format(i + 1)})
feature_col.append('close_shift_{}'.format(i + 1))
tmp_df['ts_date_id'] = tmp_df['ts_date_id'] + i + 1
stock_info = pd.merge(stock_info, tmp_df, how='left', on='ts_date_id')
stock_info.drop_duplicates(subset=['ts_date_id'],
keep='last',
inplace=True)
stock_info.drop('level_0', axis=1, inplace=True)
for i in [
0, 1, 2, 3, 4, 9, 19, 29, 39, 49, 59, 79, 99, 119, 149, 179, 199,
249
]:
stock_info['close_shift_{}'.format(i + 1)] = (
stock_info['close'] - stock_info['close_shift_{}'.format(i + 1)]
) / stock_info['close_shift_{}'.format(i + 1)]
# print(stock_info)
# stock_info.dropna(inplace=True)
# 标签制作
# make_label 未来2天的涨幅
use_col = []
for i in range(3):
tmp_df = stock_info[['ts_date_id', 'high', 'low']]
tmp_df = tmp_df.rename(
columns={
'high': 'high_shift_{}'.format(i + 1),
'low': 'low_shift_{}'.format(i + 1)
})
use_col.append('high_shift_{}'.format(i + 1))
use_col.append('low_shift_{}'.format(i + 1))
tmp_df['ts_date_id'] = tmp_df['ts_date_id'] - i - 1
stock_info = stock_info.merge(tmp_df, how='left', on='ts_date_id')
# stock_info.dropna(inplace=True)
for i in range(3):
stock_info['high_shift_{}'.format(i + 1)] = (stock_info['high_shift_{}'.format(i + 1)] - stock_info['close']) / \
stock_info['close']
stock_info['low_shift_{}'.format(i + 1)] = (stock_info['low_shift_{}'.format(i + 1)] - stock_info['close']) / \
stock_info['close']
tmp_array = stock_info[use_col].values
max_increse = np.max(tmp_array, axis=1)
min_increse = np.min(tmp_array, axis=1)
stock_info['label_max'] = max_increse
stock_info['label_min'] = min_increse
stock_info['change'] = (stock_info['high'] -
stock_info['low']) / stock_info['preclose']
# stock_info['label_final'] = (stock_info['label_max'] > 0.06) & (stock_info['label_min'] > -0.03)
stock_info['label_final'] = (stock_info['label_max'] -
stock_info['label_min']) * 100
stock_info = stock_info.dropna(subset=['label_final'])
def label_fun(x):
if x >= 20:
return 20
else:
return x
stock_info['label_final'] = stock_info['label_final'].apply(
lambda x: label_fun(x))
stock_info['label_final'] = stock_info['label_final'].apply(
lambda x: int(x))
print("正负样本：", stock_info['label_final'].value_counts())
# print(stock_info[stock_info.date == '2020-08-21'])
# print(stock_info[stock_info.label_final == 1])
# stock_info = stock_info.reset_index()
stock_info = stock_info.reset_index()
stock_info.drop('index', axis=1, inplace=True)
stock_info.to_csv(ConfigUtils.get_stock("STOCKS_DATESET"), index=False)
def reptile_dc():
pd_names = pd.read_csv(ConfigUtils.get_stock("STOCK_NAME"))
codes = set([row['code'][3:] for index, row in pd_names.iterrows()])
for code in codes:
parse_position_change(code)
def get_m_candidates():
pd_names = pd.read_csv(ConfigUtils.get_stock("STOCK_NAME"))
# 数据库获取沪、深、中小板
zxb_stocks = [s.code for s in get_mark_stocks(mark='zxb')]
hsb_stocks = [s.code for s in get_mark_stocks(mark='hsb')]
ssb_stocks = [s.code for s in get_mark_stocks(mark='ssb')]
candidate_stocks = zxb_stocks + hsb_stocks + ssb_stocks
print("total stock size:", len(candidate_stocks))
t_date = datetime.date.today().strftime('%Y-%m')
break_stocks = list()
high_stocks = list()
for index, row in tqdm(pd_names.iterrows()):
code = row['code']
name = row['code_name']
if stock_utils.is_jiucaiban(code):
continue
if code[3:] not in candidate_stocks:
continue
code_name = (code, name)
df = stock_utils.read_data(code_name,
root=ConfigUtils.get_stock("DATA_M_DIR"))
if df is None or len(df) < 2:
continue
df = df.tail(n=12) # n天前 新高/低入口
df.reset_index(drop=True, inplace=True)
close_df = df.sort_values(by='close', ascending=False)
volume_df = df.sort_values(by='volume', ascending=False)
amount_df = df.sort_values(by='amount', ascending=False)
pctChg_df = df.sort_values(by='pctChg', ascending=False)
idxCloseMax0, idxCloseMax1 = close_df.index[0], close_df.index[1]
idxVolMax0, idxVolMax1 = volume_df.index[0], volume_df.index[1]
idxAmountMax0, idxAmountMax1 = amount_df.index[0], amount_df.index[1]
idxChgMax0, idxChgMax1 = pctChg_df.index[0], pctChg_df.index[1]
volMax0, volMax1 = df.iloc[idxVolMax0]['volume'], df.iloc[idxVolMax1][
'volume']
pctMax0, pctMax1 = df.iloc[idxChgMax0]['pctChg'], df.iloc[idxChgMax1][
'pctChg']
closeMax0, closeMax1 = df.iloc[idxCloseMax0]['close'], df.iloc[
idxCloseMax1]['close']
date0, date1 = df.iloc[idxCloseMax0]['date'], df.iloc[idxCloseMax1][
'date']
pch0, pch1 = df.iloc[idxCloseMax0]['pctChg'], df.iloc[idxCloseMax1][
'pctChg']
amount0, amount1 = df.iloc[idxCloseMax0]['amount'], df.iloc[
idxCloseMax1]['amount']
# 经过3个月洗盘, 最近股价将赶超新高
if t_date == date1 and (idxCloseMax1 - idxCloseMax0) > 2:
# print(code_name, volMax1, date0, date1, idxVolMax0, idxChgMax0, idxCloseMax0, closeMax0)
r = Record(name, code, t_date, closeMax1, 'm', amount1, pch1)
break_stocks.append(r)
# 持续新高
if t_date == date0:
# print(code_name, dateMax0, dateMax1, idxVolMax0, idxChgMax0, idxCloseMax0, closeMax0)
r = Record(name, code, t_date, closeMax0, 'm', amount0, pch0)
high_stocks.append(r)
break_stocks = sorted(break_stocks, key=lambda x: x.volume, reverse=True)
high_stocks = sorted(high_stocks, key=lambda x: x.volume, reverse=True)
# print(break_stocks[:30])
# print(high_stocks[:30])
break_stocks = [r for r in break_stocks if r.volume < 5e11]
high_stocks = [r for r in high_stocks if r.volume < 90081936793]
return break_stocks, high_stocks
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# @Time : 2020/11/29 5:09 下午
# @Author : zhangzhen12
# @Site :
# @File : wzcc_dynamic.py
# @Software: PyCharm
import codecs
import csv
import os
import pandas as pd
import tqdm
from common.config import ConfigUtils
wz_path = ConfigUtils.get_stock("WZ_DIR")
names = list(os.listdir(wz_path))
dates = sorted([name[3:-4] for name in names])
dates = dates[-100:]
stock_percent_dict = {}
stock_amount_dict = {}
stocks = set()
for name in tqdm.tqdm(names):
date = name[3:-4]
df = pd.read_csv(os.path.join(wz_path, name))
# print(list(df.iterrows())[0])
for index, row in df.iterrows():
stock_name = row['name']
