def equity_interest_pledge_too_high(securities: str, time_serial: tuple,
data_hub: DataHubEntry,
database: DatabaseEntry,
context: AnalysisContext,
**kwargs) -> AnalysisResult:
nop(database, context, kwargs)
query_fields = ['质押次数', '无限售股质押数量', '限售股份质押数量', '总股本', '质押比例']
if not data_hub.get_data_center().check_readable_name(query_fields):
return AnalysisResult(securities, None,
AnalysisResult.SCORE_NOT_APPLIED, '无法识别的字段名',
'无法识别的字段名')
df = data_hub.get_data_center().query('Stockholder.PledgeStatus',
securities, (years_ago(2), now()),
fields=query_fields +
['stock_identity', 'due_date'],
readable=True)
if df is None or len(df) == 0:
return AnalysisResult(securities, None,
AnalysisResult.SCORE_NOT_APPLIED, '没有数据', '没有数据')
df = df.sort_values('due_date', ascending=False)
score = 100
reason = []
previous_pledge_times = 0
for index, row in df.iterrows():
due_date = row['due_date']
pledge_rate = row['质押比例']
pledge_times = row['质押次数']
if pledge_times != previous_pledge_times:
if pledge_rate > 50.0:
score = 0
if pledge_rate > 20.0:
score = 60
reason.append('%s: 质押比例：%.2f%%' %
(str(due_date.date()), pledge_rate))
previous_pledge_times = pledge_times
if len(reason) == 0:
brief = '近4年没有超过20%%的质押记录'
reason = brief
else:
brief = '近4年有%s次超过20%%的质押记录' % len(reason)
reason = brief
return AnalysisResult(securities, None, score, reason, brief)
def analysis_investigation(securities: str, time_serial: tuple, data_hub: DataHubEntry,
database: DatabaseEntry, context: AnalysisContext, **kwargs) -> AnalysisResult:
nop(time_serial, database, )
if context.cache.get('investigation', None) is None:
context.cache['investigation'] = data_hub.get_data_center().query('Market.Investigation')
df = context.cache.get('investigation', None)
error_report = check_gen_report_when_data_missing(df, securities, 'Market.Investigation',
['stock_identity', 'investigate_date',
'investigate_topic', 'investigate_reason'])
if error_report is not None:
return error_report
df_slice = df[df['stock_identity'] == securities]
df_slice_in_4_years = df_slice[df_slice['investigate_date'] > years_ago(5)]
score = 100
reason = []
for index, row in df_slice_in_4_years.iterrows():
score = 0
investigate_date = row['investigate_date']
investigate_topic = row['investigate_topic']
investigate_reason = row['investigate_reason']
reason.append('%s: <<%s>> -- %s' % (date2text(investigate_date), investigate_topic, investigate_reason))
if len(reason) == 0:
reason.append('近四年无立案调查记录')
return AnalysisResult(securities, None, score, reason, AnalysisResult.WEIGHT_ONE_VOTE_VETO)
def analysis_location_limitation(securities: str, time_serial: tuple,
data_hub: DataHubEntry,
database: DatabaseEntry,
context: AnalysisContext,
**kwargs) -> [AnalysisResult]:
nop(kwargs)
nop(database)
nop(time_serial)
if context.cache.get('securities_info', None) is None:
context.cache['securities_info'] = data_hub.get_data_center().query(
'Market.SecuritiesInfo')
df = context.cache.get('securities_info', None)
df_slice = df[df['stock_identity'] == securities]
area = get_dataframe_slice_item(df_slice, 'area', 0, '')
# Add your exclude area here
exclude = area in []
# List the stock code here helps us resolving report line stock and result not matching issue
reason = '%s: 地域为%s' % (securities, str(area))
return AnalysisResult(securities, None, not exclude, reason,
'排除' if exclude else '正常')
def analysis_current_and_quick_ratio(securities: str, time_serial: tuple, data_hub: DataHubEntry,
database: DatabaseEntry, context: AnalysisContext, **kwargs) -> [AnalysisResult]:
nop(database, context, kwargs)
df = data_hub.get_data_center().query_from_factor(
'Factor.Finance', securities, time_serial, fields=['流动比率', '速动比率'], readable=True)
if df is None or len(df) == 0:
return AnalysisResult(securities, None, AnalysisResult.SCORE_NOT_APPLIED, '')
# Annual report
df = df[df['period'].dt.month == 12]
results = []
for index, row in df.iterrows():
score = 100
reason = []
period = row['period']
if row['流动比率'] < 2.0:
score -= 50
reason.append('%s: 流动比率为%.2f < 2.0' % (str(period), row['流动比率']))
else:
reason.append('%s: 流动比率为%.2f - 合格' % (str(period), row['流动比率']))
if row['速动比率'] < 1.0:
score -= 50
reason.append('%s: 速动比率为%.2f < 1.0' % (str(period), row['速动比率']))
else:
reason.append('%s: 速动比率为%.2f - 合格' % (str(period), row['速动比率']))
results.append(AnalysisResult(securities, period, score, reason))
return results
def analysis_inquiry(securities: str, time_serial: tuple, data_hub: DataHubEntry,
database: DatabaseEntry, context: AnalysisContext, **kwargs) -> AnalysisResult:
nop(time_serial, database, )
df = data_hub.get_data_center().query('Market.Enquiries', securities)
if df is None or len(df) == 0:
return AnalysisResult(securities, None, AnalysisResult.SCORE_PASS, '四年内无问询记录（也可能是数据缺失）')
error_report = check_gen_report_when_data_missing(df, securities, 'Market.Enquiries',
['stock_identity', 'enquiry_date', 'enquiry_topic'])
if error_report is not None:
return error_report
df_slice = df[df['stock_identity'] == securities]
df_slice_in_4_years = df_slice[df_slice['enquiry_date'] > years_ago(5)]
score = 100
reason = []
for index, row in df_slice_in_4_years.iterrows():
enquiry_date = row['enquiry_date']
enquiry_topic = row['enquiry_topic']
enquiry_title = row['enquiry_title']
if '问询函' in enquiry_topic or '关注函' in enquiry_topic:
score = 59
reason.append('%s: <<%s>> -- %s' % (date2text(enquiry_date), enquiry_topic, enquiry_title))
if len(reason) == 0:
reason.append('近四年无敏感问询')
return AnalysisResult(securities, None, score, reason)
def analysis_finance_report_sign(securities: str, time_serial: tuple, data_hub: DataHubEntry,
database: DatabaseEntry, context: AnalysisContext, **kwargs) -> [AnalysisResult]:
nop(database, kwargs)
if context.cache.get('finance_audit', None) is None:
context.cache['finance_audit'] = data_hub.get_data_center().query('Finance.Audit', None, time_serial)
df = context.cache.get('finance_audit', None)
error_report = check_gen_report_when_data_missing(df, securities, 'Finance.Audit',
['stock_identity', 'period', 'conclusion'])
if error_report is not None:
return error_report
df_slice = df[df['stock_identity'] == securities]
# df_slice_in_4_years = df_slice[df_slice['period'] > years_ago(5)]
results = []
for index, row in df_slice.iterrows():
reason = []
period = row['period']
conclusion = row['conclusion']
if pd.isnull(conclusion):
score = AnalysisResult.SCORE_NOT_APPLIED
reason.append(date2text(period) + ' : No sign data.')
elif conclusion != '标准无保留意见':
score = AnalysisResult.SCORE_FAIL
reason.append(date2text(period) + ' : ' + conclusion)
else:
score = AnalysisResult.SCORE_PASS
results.append(AnalysisResult(securities, period, score, reason, AnalysisResult.WEIGHT_ONE_VOTE_VETO))
return results
def analysis_finance_report_sign(securities: str, data_hub: DataHubEntry,
database: DatabaseEntry, context: AnalysisContext) -> AnalysisResult:
nop(database)
if context.cache.get('finance_audit', None) is None:
context.cache['finance_audit'] = data_hub.get_data_center().query('Finance.Audit')
df = context.cache.get('finance_audit', None)
error_report = check_gen_report_when_data_missing(df, securities, 'Finance.Audit',
['stock_identity', 'period', 'conclusion'])
if error_report is not None:
return error_report
df_slice = df[df['stock_identity'] == securities]
df_slice_in_4_years = df_slice[df_slice['period'] > years_ago(5)]
score = 100
reason = []
for index, row in df_slice_in_4_years.iterrows():
period = row['period']
conclusion = row['conclusion']
if conclusion != '标准无保留意见':
score = 0
reason.append(date2text(period) + ' : ' + conclusion)
if len(reason) == 0:
reason.append('近四年均为标准无保留意见')
return AnalysisResult(securities, score, reason, AnalysisResult.WEIGHT_ONE_VOTE_VETO)
def analysis_dispersed_ownership(securities: str, time_serial: tuple,
data_hub: DataHubEntry,
database: DatabaseEntry,
context: AnalysisContext,
**kwargs) -> AnalysisResult:
nop(database, context, kwargs)
df = data_hub.get_data_center().query(
'Stockholder.Statistics',
securities,
(years_ago(3), now()),
)
if df is None or len(df) == 0:
return AnalysisResult(securities, None,
AnalysisResult.SCORE_NOT_APPLIED, '没有数据')
df = df[df['period'].dt.month == 12]
df = df.sort_values('period', ascending=False)
score = 100
reason = []
applied = False
for index, row in df.iterrows():
period = row['period']
stockholder_top10 = row['stockholder_top10']
stockholder_top10_nt = row['stockholder_top10_nt']
if not isinstance(stockholder_top10, (list, tuple)):
continue
if len(stockholder_top10) != 10:
continue
largest_ratio = 0.0
biggest_holder = ''
for stockholder_data in stockholder_top10:
if 'hold_ratio' not in stockholder_data.keys(
) or 'holder_name' not in stockholder_data.keys():
break
applied = True
hold_ratio = stockholder_data.get('hold_ratio')
holder_name = stockholder_data.get('holder_name')
if hold_ratio > largest_ratio:
largest_ratio = hold_ratio
biggest_holder = holder_name
if largest_ratio == 0.0:
return AnalysisResult(securities, None,
AnalysisResult.SCORE_NOT_APPLIED,
'缺少必要数据，请确保数据包含tushare pro数据源')
if largest_ratio < 0.1:
score = 0
reason.append('%s: 最大股东 %s 持股比例为%.2f%%，小于10%%' %
(str(period), biggest_holder, largest_ratio * 100))
else:
reason.append('%s: 最大股东 %s 持股比例为%.2f%%' %
(str(period), biggest_holder, largest_ratio * 100))
if len(reason) == 0:
reason.append('没有数据')
return AnalysisResult(securities, None, score, reason) if applied else \
AnalysisResult(securities, None, AnalysisResult.SCORE_NOT_APPLIED, reason)
def analysis_repurchase(securities: str, time_serial: tuple,
data_hub: DataHubEntry, database: DatabaseEntry,
context: AnalysisContext, **kwargs) -> AnalysisResult:
nop(time_serial, database, context, kwargs)
df = data_hub.get_data_center().query('Stockholder.Repurchase', securities,
(years_ago(1), now()))
if df is None or len(df) == 0:
return AnalysisResult(securities, None, AnalysisResult.SCORE_JUST,
'前后一年内没有回购数据', '无回购数据')
# df = df.where(df.notnull(), None)
volume = 0
reasons = []
for index, row in df.iterrows():
proc = row['proc']
if proc != '股东大会通过':
# For multiple calculate, just count pass
continue
ann_date = row['ann_date']
# TS_ISSUE: Some fields may miss
end_date = row['end_date'] if 'end_date' in df.columns else None
volume = row['vol'] if 'vol' in df.columns else None
low_limit = row['low_limit'] if 'low_limit' in df.columns else None
high_limit = row['high_limit'] if 'high_limit' in df.columns else None
end_date_text = ('截止%s' % end_date.date()) if \
isinstance(end_date, datetime.datetime) and end_date is not pd.NaT else ''
if not pd.isnull(low_limit) and not pd.isnull(high_limit):
if low_limit == high_limit:
price_text = '将以%s元的价格' % low_limit
else:
price_text = '将以%s - %s元的价格' % (low_limit, high_limit)
elif not pd.isnull(low_limit):
price_text = '将以最低%s元的价格' % low_limit
elif not pd.isnull(high_limit):
price_text = '将以最高%s元的价格' % high_limit
else:
price_text = ''
if not pd.isnull(volume):
volume += int(volume)
volume_text = ('%s股' % volume)
else:
volume_text = ''
reasons.append(
'%s: 股东大会通过，%s%s回购%s股票' %
(ann_date.date(), end_date_text, price_text, volume_text))
brief = '期间计划回购%s股' % volume
return AnalysisResult(securities, None, AnalysisResult.SCORE_PASS, reasons, brief) if len(reasons) > 0 else \
AnalysisResult(securities, None, AnalysisResult.SCORE_JUST, '前后一年内没有回购数据', '近一年无数据')
def analysis_roe_roa(securities: str, time_serial: tuple,
data_hub: DataHubEntry, database: DatabaseEntry,
context: AnalysisContext, **kwargs) -> [AnalysisResult]:
nop(database, context, kwargs)
df = data_hub.get_data_center().query_from_factor(
'Factor.Finance',
securities,
time_serial,
fields=['总资产收益率', '净资产收益率'],
readable=True)
if df is None or len(df) == 0:
return AnalysisResult(securities, None,
AnalysisResult.SCORE_NOT_APPLIED, '')
# Annual report
df = df[df['period'].dt.month == 12]
results = []
for index, row in df.iterrows():
score = 100
reason = []
period = row['period']
if row['总资产收益率'] < 0.05:
score -= 50
reason.append('%s: 总资产收益率为%s%% - 过低' %
(str(period), format_pct(row['总资产收益率'])))
elif row['总资产收益率'] > 0.15:
score -= 50
reason.append('%s: 总资产收益率为%s%% - 偏高，需要引起注意' %
(str(period), format_pct(row['总资产收益率'])))
else:
# Theory: 7.5% - 13%, but we use 5% - 15% for wider tolerance.
reason.append('%s: 总资产收益率为%s%% - 合理' %
(str(period), format_pct(row['总资产收益率'])))
pass
if row['净资产收益率'] < 0.15:
score -= 50
reason.append('%s: 净资产收益率为%s%% - 偏低' %
(str(period), format_pct(row['净资产收益率'])))
elif row['净资产收益率'] > 0.40:
score = 0
reason.append('%s: 净资产收益率为%s%% - 过高，可能是造假或偶然因素' %
(str(period), format_pct(row['净资产收益率'])))
else:
# Theory: 15% - 39%.
reason.append('%s: 净资产收益率为%s%% - 合理' %
(str(period), format_pct(row['净资产收益率'])))
results.append(AnalysisResult(securities, period, score, reason))
return results
def analysis_stock_unlock(securities: str, time_serial: tuple,
data_hub: DataHubEntry, database: DatabaseEntry,
context: AnalysisContext,
**kwargs) -> AnalysisResult:
nop(time_serial, database, context, kwargs)
no_data_result = AnalysisResult(securities, None,
AnalysisResult.SCORE_PASS,
'前三个月或后半年内没有解禁数据', '无解禁数据')
df: pd.DataFrame = data_hub.get_data_center().query(
'Stockholder.StockUnlock', securities, (years_ago(2), now()))
if df is None or df.empty:
return no_data_result
df = df[df['unlock_date'].notna()]
df['unlock_date'] = df['unlock_date'].apply(text_auto_time)
mask = (df['unlock_date'] > days_ago(90)) & (df['unlock_date'] <=
days_after(180))
df = df.loc[mask]
if df is None or df.empty:
return no_data_result
df_group = df.groupby('unlock_date')
reasons = []
float_share_sum = 0
for g, df in df_group:
unlock_date = g
float_share = sum(df['float_share'])
float_ratio = sum(df['float_ratio'])
float_share_sum += float_share
reasons.append('%s: 解禁%s股，占总股份%s%%' %
(unlock_date.date(), float_share, float_ratio))
# for index, row in df.iterrows():
# float_date = row['float_date']
# float_share = row['float_share']
# float_ratio = row['float_ratio']
#
# # Maybe have not converted to datetime but keeping str
# # Maybe there're a lot of unlock in one day
#
# if not isinstance(float_date, datetime.datetime):
# float_date = text_auto_time(float_date)
# if days_ago(90) < float_date < days_after(180):
# reasons.append('%s: 解禁%s股，占总股份%s%%' % (float_date.date(), float_share, float_ratio))
brief = '共解禁%s股' % float_share_sum
return AnalysisResult(securities, None, AnalysisResult.SCORE_FAIL, reasons, brief) if len(reasons) > 0 else \
AnalysisResult(securities, None, AnalysisResult.SCORE_PASS, '前三个月或后半年内没有解禁数据', '无解禁数据')
def analysis_location_limitation(securities: str, data_hub: DataHubEntry,
database: DatabaseEntry,
context: AnalysisContext) -> AnalysisResult:
nop(database)
if context.cache.get('securities_info', None) is None:
context.cache['securities_info'] = data_hub.get_data_center().query(
'Market.SecuritiesInfo')
df = context.cache.get('securities_info', None)
df_slice = df[df['stock_identity'] == securities]
area = get_dataframe_slice_item(df_slice, 'area', 0, '')
exclude = area in ['黑龙江', '辽宁', '吉林']
reason = securities + '地域为' + str(area)
return AnalysisResult(securities, not exclude, reason)
def analysis_less_than_3_years(securities: str, data_hub: DataHubEntry,
database: DatabaseEntry,
context: AnalysisContext) -> AnalysisResult:
nop(database)
if context.cache.get('securities_info', None) is None:
context.cache['securities_info'] = data_hub.get_data_center().query(
'Market.SecuritiesInfo')
df = context.cache.get('securities_info', None)
df_slice = df[df['stock_identity'] == securities]
listing_date = get_dataframe_slice_item(df_slice, 'listing_date', 0, now())
exclude = now().year - listing_date.year < 3
reason = '上市日期' + str(listing_date) + ('小于三年' if exclude else '大于三年')
return AnalysisResult(securities, not exclude, reason)
def analysis_exclude_industries(securities: str, data_hub: DataHubEntry,
database: DatabaseEntry,
context: AnalysisContext) -> AnalysisResult:
nop(database)
if context.cache.get('securities_info', None) is None:
context.cache['securities_info'] = data_hub.get_data_center().query(
'Market.SecuritiesInfo')
df = context.cache.get('securities_info', None)
df_slice = df[df['stock_identity'] == securities]
industry = get_dataframe_slice_item(df_slice, 'industry', 0, '')
exclude = industry in ['种植业', '渔业', '林业', '畜禽养殖', '农业综合']
reason = '所在行业[' + str(industry) + (']属于农林牧渔' if exclude else ']不属于农林牧渔')
return AnalysisResult(securities, not exclude, reason)
def analysis_less_than_3_years(securities: str, time_serial: tuple,
data_hub: DataHubEntry, database: DatabaseEntry,
context: AnalysisContext,
**kwargs) -> [AnalysisResult]:
nop(kwargs)
nop(database)
nop(time_serial)
if context.cache.get('securities_info', None) is None:
context.cache['securities_info'] = data_hub.get_data_center().query(
'Market.SecuritiesInfo')
df = context.cache.get('securities_info', None)
df_slice = df[df['stock_identity'] == securities]
listing_date = get_dataframe_slice_item(df_slice, 'listing_date', 0, now())
less_than_3_years = now().year - listing_date.year < 3
reason = '上市日期 %s (%s)' % (str(listing_date),
('小于三年' if less_than_3_years else '大于三年'))
return AnalysisResult(securities, None, not less_than_3_years, reason)
def analysis_black_list(securities: str, time_serial: tuple,
data_hub: DataHubEntry, database: DatabaseEntry,
context: AnalysisContext,
**kwargs) -> [AnalysisResult]:
nop(kwargs)
nop(database)
nop(time_serial)
if context.cache.get('black_table', None) is None:
black_list_module = data_hub.get_data_extra('black_list')
if black_list_module is not None:
context.cache[
'black_table'] = black_list_module.get_black_list_data()
black_table = context.cache.get('black_table', None)
if isinstance(black_table, pd.DataFrame) and not black_table.empty:
df_slice = black_table[black_table['security'] == securities]
in_black_list = len(df_slice) > 0
reason = get_dataframe_slice_item(df_slice, 'content', 0,
'') if in_black_list else '不在黑名单中'
return AnalysisResult(securities, None, not in_black_list, reason)
else:
return AnalysisResult(securities, None,
AnalysisResult.SCORE_NOT_APPLIED, '载入黑名单模块失败')
def analysis_increase_decrease(securities: str, time_serial: tuple,
data_hub: DataHubEntry, database: DatabaseEntry,
context: AnalysisContext,
**kwargs) -> AnalysisResult:
nop(time_serial, database, context, kwargs)
no_data_result = AnalysisResult(securities, None,
AnalysisResult.SCORE_NOT_APPLIED,
'前后一年内没有增减持数据', '近一年无数据')
df = data_hub.get_data_center().query('Stockholder.ReductionIncrease',
securities, (years_ago(2), now()))
if df is None or len(df) == 0:
return no_data_result
volume = 0
reasons = []
for index, row in df.iterrows():
stock_holder = row['stock_holder']
holder_type = row['holder_type']
holder_type = {
'G': '高管',
'P': '个人',
'C': '公司',
}.get(holder_type, '')
increase_or_decrease = row['in_de']
change_vol = row['change_vol'] if 'change_vol' in df.columns else '?'
change_ratio = row[
'change_ratio'] if 'change_ratio' in df.columns else '?'
avg_price = row['avg_price'] if 'avg_price' in df.columns else '?'
begin_date = text_auto_time(
row['begin_date']) if 'begin_date' in df.columns else None
close_date = text_auto_time(
row['close_date']) if 'close_date' in df.columns else None
if (begin_date is not None and days_ago(365) < begin_date < days_after(365)) or \
(close_date is not None and days_ago(365) < close_date < days_after(365)):
if increase_or_decrease == 'IN':
volume += change_vol
operation = '增持'
elif increase_or_decrease == 'DE':
volume -= change_vol
operation = '减持'
else:
operation = ''
if operation != '':
reasons.append(
'%s - %s: %s[%s]以平均价格%s元%s%s股，占流通股%s%%' %
(begin_date.date() if begin_date is not None else '?',
close_date.date() if close_date is not None else '?',
holder_type, stock_holder, avg_price, operation,
change_vol, change_ratio))
if len(reasons) == 0:
return no_data_result
if volume > 0:
conclusion = '此期间净增持%s股' % volume
elif volume < 0:
conclusion = '此期间净减持%s股' % -volume
else:
conclusion = '此期间增持减持持平'
reasons.append(conclusion)
final_score = AnalysisResult.SCORE_FAIL if volume < 0 else AnalysisResult.SCORE_PASS
return AnalysisResult(securities, None, final_score, reasons,
conclusion.replace('此', ''))