def get_appropriate_date_momentum(
company_df: pd.DataFrame,
company,
duration: Tuple[int, int] = (0, 1),
verbosity: int = 1) -> Tuple[datetime.datetime, float]:
"""
Return appropriate date which is present in data record.
Parameters
----------
company_df : pd.DataFrame
Company dataframe
duration : Tuple[year,month], optional
Desired duration to go back to retrive record, by default (0,1)
verbosity : int, optional
Level of detail logging,1=< Deatil, 0=Less detail , by default 1
Returns
-------
Tuple[datetime.datetime,float]
Date,Close value on date retrived
Raises
------
ValueError
If desired old is older than first record
"""
current_date = company_df.iloc[-1].Date
desired_date = current_date - \
dateutil.relativedelta.relativedelta(
years=duration[0], months=duration[1])
if desired_date < company_df.iloc[0].Date:
logger.error(
f"Given desired date {desired_date.strftime('%d-%m-%Y')} is older than first recorded date {company_df.iloc[0].Date.strftime('%d-%m-%Y')}"
)
raise ValueError
dd_copy = desired_date
if verbosity > 0:
logger.debug(
f"Your desired date for monthly return for {company} is {desired_date.strftime('%d-%m-%Y')}"
)
if len(company_df.loc[company_df['Date'] == desired_date]) != 0:
desired_close = company_df.loc[company_df['Date'] == desired_date]
else:
for i in range(1, 100):
if len(company_df.loc[company_df['Date'] == desired_date]) == 0:
desired_date = desired_date - \
dateutil.relativedelta.relativedelta(days=i)
desired_close = company_df.loc[company_df['Date'] ==
desired_date]
break
if verbosity > 0:
logger.warning(
f"Desired date: {dd_copy.strftime('%d-%m-%Y')} not found going for next possible date: {desired_date.strftime('%d-%m-%Y')}"
)
return desired_date, desired_close.iloc[-1].Close
def get_appropriate_date_ema(
company_df: pd.DataFrame,
desired_date: datetime.datetime,
verbosity: int = 1) -> Tuple[datetime.datetime, float]:
"""
Return appropriate date which is present in data record.
Parameters
----------
company_df : pd.DataFrame
Company dataframe
duration : datetime.datetime
Desired date cut-off to calculate ema
verbosity : int, optional
Level of detail logging, by default 1
Returns
-------
Tuple[datetime.datetime,float]
Date,Close value on date retrived
Raises
------
ValueError
If desired old is older than first record
"""
if desired_date < company_df.index[0]:
logger.error(
f"Given desired date {desired_date.strftime('%d-%m-%Y')} is older than first recorded date {company_df.index[0].strftime('%d-%m-%Y')}"
)
if verbosity > 0:
logger.debug(
f"Your desired EMA cut-off date is {desired_date.strftime('%d-%m-%Y')}"
)
for day_idx in range(1, 100):
if desired_date not in company_df.index:
date = desired_date - \
dateutil.relativedelta.relativedelta(days=day_idx)
else:
date = desired_date
if date in company_df.index:
break
if verbosity > 0 and desired_date != date:
logger.warning(
f"Desired date: {desired_date.strftime('%d-%m-%Y')} not found going for next possible date: {date.strftime('%d-%m-%Y')}"
)
return date
def volume_n_days_indicator(
self,
duration: int = 90,
save: bool = True,
export_path: str = ".",
verbosity: int = 1,
) -> Optional[pd.DataFrame]:
"""Mean Volume Indicator based on desired days
Args:
duration (int, optional): Total days from current date to retrive data. Defaults to 90.
save (bool, optional): Save to hard disk. Defaults to True.
export_path (str, optional): Path to save, to be used only if 'save' is true. Defaults to ".".
verbosity (int, optional): Level of detail logging,1=< Deatil, 0=Less detail. Defaults to 1.
Returns:
All Volume based indicator
Example:
```python
from stock_analysis.indicator import Indicator
ind = Indicator('./data/company_list.yaml')
vol = ind.volume_n_days_indicator(150)
```
"""
with parallel_backend(n_jobs=-1, backend="multiprocessing"):
result = Parallel()(
delayed(self.unit_vol_indicator_n_days)(company, duration)
for company in self.data["company"])
vol_ind_df = pd.DataFrame(result)
if verbosity > 0:
logger.debug(f"Here are sample 5 company\n{vol_ind_df.head()}")
if save is True:
new_folder(export_path)
vol_ind_df.to_csv(
f"{export_path}/VolumeIndicator90Days_detailed_{now_strting}.csv",
index=False,
)
if verbosity > 0:
logger.debug(
f"Save at {export_path}/VolumeIndicator90Days_detailed_{now_strting}.csv"
)
else:
return vol_ind_df
def _ema_indicator_n3(self,
ema_canditate: Tuple[int, int] = (5, 13, 26),
cutoff_date: Union[str, datetime.datetime] = 'today',
verbosity: int = 1) -> pd.DataFrame:
with multiprocessing.Pool(multiprocessing.cpu_count() - 1) as pool:
result = pool.starmap(
self._parallel_ema_indicator_n3,
[(company, ema_canditate, cutoff_date, verbosity)
for company in self.data['company']])
ema_indicator_df = pd.DataFrame(result)
ema_indicator_df.dropna(inplace=True)
if verbosity > 0:
logger.debug(
f"Here are sample 5 company\n{ema_indicator_df.head()}")
return ema_indicator_df
def _ema_indicator_n3(
self,
ema_canditate: Tuple[int, int, int] = (5, 13, 26),
cutoff_date: Union[str, datetime.datetime] = "today",
verbosity: int = 1,
) -> pd.DataFrame:
with parallel_backend(n_jobs=-1, backend="multiprocessing"):
result = Parallel()(delayed(self.unit_ema_indicator_n3)(
company, ema_canditate, cutoff_date, verbosity)
for company in self.data["company"])
ema_indicator_df = pd.DataFrame(result)
ema_indicator_df.dropna(inplace=True)
if verbosity > 0:
logger.debug(
f"Here are sample 5 company\n{ema_indicator_df.head()}")
return ema_indicator_df
def unit_momentum(self, company: str, start, end, verbosity: int = 1):
logger.info(f"Retriving data for {company}")
try:
company_df = DataRetrive.single_company_specific(
company_name=f"{company}.NS", start_date=start, end_date=end
)
company_df.reset_index(inplace=True)
ar_yearly = annualized_rate_of_return(
end_date=company_df.iloc[-1].Close,
start_date=company_df.iloc[0].Close,
duration=1,
) # (company_df.iloc[-30,0] - company_df.iloc[0,0]).days/365)
ar_monthly = annualized_rate_of_return(
end_date=company_df.iloc[-1].Close,
start_date=get_appropriate_date_momentum(
company_df, company, verbosity=verbosity
)[1],
duration=(company_df.iloc[-1, 0] - company_df.iloc[-30, 0]).days / 30,
)
monthly_start_date = get_appropriate_date_momentum(
company_df, company, verbosity=0
)[0].strftime("%d-%m-%Y")
except (IndexError, KeyError, ValueError):
if verbosity > 0:
logger.debug(f"Data is not available for: {company}")
company_df = pd.DataFrame(
{"Date": [datetime.datetime(1000, 1, 1)] * 30, "Close": [pd.NA] * 30}
)
ar_yearly, ar_monthly, monthly_start_date = pd.NA, pd.NA, pd.NA
return {
"company": company,
"yearly_start_date": company_df.iloc[0].Date.strftime("%d-%m-%Y"),
"yearly_start_date_close": company_df.iloc[0].Close,
"yearly_end_date": company_df.iloc[-1].Date.strftime("%d-%m-%Y"),
"yearly_end_date_close": company_df.iloc[-1].Close,
"return_yearly": ar_yearly,
"monthly_start_date": monthly_start_date,
"monthly_start_date_close": company_df.iloc[-30].Close,
"monthly_end_date": company_df.iloc[-1].Date.strftime("%d-%m-%Y"),
"monthly_end_date_close": company_df.iloc[-1].Close,
"return_monthly": ar_monthly,
}
def volume_indicator_n_days(self,
duration: int = 90,
save: bool = True,
export_path: str = '.',
verbosity: int = 1) -> pd.DataFrame:
"""Mean Volume Indicator based on desired days
Parameters
----------
duration : int, optional
Total days from current date to retrive data, by default 90
save : bool, optional
Save to hard disk, by default True
export_path : str, optional
Path to save, to be used only if 'save' is true, by default '.'
verbosity : int, optional
Level of detail logging,1=< Deatil, 0=Less detail , by default 1
Returns
-------
pd.DataFrame
All Volume based indicator
"""
with multiprocessing.Pool(multiprocessing.cpu_count() - 1) as pool:
result = pool.starmap(self._parallel_vol_indicator_n_days,
[(company, duration)
for company in self.data['company']])
vol_ind_df = pd.DataFrame(result)
if verbosity > 0:
logger.debug(f"Here are sample 5 company\n{vol_ind_df.head()}")
if save is True:
vol_ind_df.to_csv(
f"{export_path}/VolumeIndicator90Days_detailed_{now_strting}.csv",
index=False)
if verbosity > 0:
logger.debug(
f"Save at {export_path}/VolumeIndicator90Days_detailed_{now_strting}.csv"
)
else:
return vol_ind_df
def ema_indicator(
self,
ema_canditate: Tuple[int, int] = (50, 200),
cutoff_date: Union[str, datetime.datetime] = "today",
save: bool = True,
export_path: str = ".",
verbosity: int = 1,
) -> Optional[pd.DataFrame]:
"""Exponential moving average based on desired two period (or no of days)
Args:
ema_canditate (Tuple[int, int], optional): Two number used two calculate EMA. Defaults to (50, 200).
cutoff_date (Union[str, datetime.datetime], optional): Desired date till which to calculate ema. Defaults to "today".
save (bool, optional): Save to hard disk. Defaults to True.
export_path (str, optional): Path to save, to be used only if 'save' is true. Defaults to ".".
verbosity (int, optional): Level of detail logging,1=< Detail, 0=Less detail. Defaults to 1.
Returns:
EMA and indicators based on it
Example:
```python
from stock_analysis.indicator import Indicator
ind = Indicator('./data/company_list.yaml')
ema = ind.ema_indicator((50,200), '01/06/2020')
```
"""
with parallel_backend(n_jobs=-1, backend="multiprocessing"):
result = Parallel()(delayed(self.unit_ema_indicator)(
company, ema_canditate, cutoff_date, verbosity)
for company in self.data["company"])
ema_indicator_df = pd.DataFrame(result)
ema_indicator_df.dropna(inplace=True)
ema_indicator_df["percentage_diff"] = ema_indicator_df.apply(
lambda x: percentage_diff(
x[f"ema{str(ema_canditate[0])}"],
x[f"ema{str(ema_canditate[1])}"],
return_absolute=True,
),
axis=1,
)
ema_indicator_df["outcome"] = ema_indicator_df.apply(
lambda x: outcome_analysis(x["percentage_diff"]), axis=1)
ema_indicator_df = ema_indicator_df[[
"company",
"ema_date",
f"ema{str(ema_canditate[0])}",
f"ema{str(ema_canditate[1])}",
"percentage_diff",
"outcome",
"action",
]]
if verbosity > 0:
logger.debug(
f"Here are sample 5 company\n{ema_indicator_df.head()}")
if save is True:
new_folder(export_path)
ema_indicator_df.to_csv(
f"{export_path}/ema_indicator{str(ema_canditate[0])}-{str(ema_canditate[1])}_{len(self.data['company'])}company_{now_strting}.csv",
index=False,
)
if verbosity > 0:
logger.debug(
f"Exported at {export_path}/ema_indicator{str(ema_canditate[0])}-{str(ema_canditate[1])}_{len(self.data['company'])}company_{now_strting}.csv"
)
else:
return ema_indicator_df
def ema_crossover_detail_indicator(
self,
ema_canditate: Tuple[int, int, int] = (5, 13, 26),
save: bool = True,
export_path: str = ".",
verbosity: int = 1,
) -> Optional[pd.DataFrame]:
"""Exponential moving average for crossover triple period technique
Args:
ema_canditate (Tuple[int, int, int], optional): Three Period (or days) to calculate EMA. Defaults to (5, 13, 26).
save (bool, optional): Save to hard disk. Defaults to True.
export_path (str, optional): Path to save, to be used only if 'save' is true. Defaults to ".".
verbosity (int, optional): Level of detail logging,1=< Deatil, 0=Less detail. Defaults to 1.
Returns:
Results is based on crossover ema and detailed metrics
Example:
```python
from stock_analysis.indicator import Indicator
ind = Indicator('./data/company_list.yaml')
ema = ind.ema_crossover_detail_indicator((5,10,020), '01/06/2020')
```
"""
logger.info("Performing EMA Indicator Task")
ema_short = self._ema_indicator_n3(ema_canditate=ema_canditate,
verbosity=verbosity)
logger.info("Extarcting detail company quote data")
batch_company_quote = pd.DataFrame()
with parallel_backend(n_jobs=-1, backend="multiprocessing"):
company_quote = Parallel()(
delayed(self.unit_quote_retrive)(company)
for company in ema_short["company"])
for single_company_quote in company_quote:
if isinstance(single_company_quote, pd.DataFrame):
batch_company_quote = batch_company_quote.append(
single_company_quote)
batch_company_quote = batch_company_quote.reset_index().rename(
columns={"index": "company"})
batch_company_quote = batch_company_quote[[
"company",
"longName",
"price",
"regularMarketVolume",
"marketCap",
"bookValue",
"priceToBook",
"averageDailyVolume3Month",
"averageDailyVolume10Day",
"fiftyTwoWeekLowChange",
"fiftyTwoWeekLowChangePercent",
"fiftyTwoWeekRange",
"fiftyTwoWeekHighChange",
"fiftyTwoWeekHighChangePercent",
"fiftyTwoWeekLow",
"fiftyTwoWeekHigh",
]]
batch_company_quote["company"] = batch_company_quote[
"company"].str.replace(".NS", "")
ema_quote = ema_short.merge(batch_company_quote,
on="company",
validate="1:1")
if verbosity > 0:
logger.debug(f"Here are sample 5 company\n{ema_quote.head()}")
if save is not False:
ema_quote.to_csv(
f"{export_path}/ema_crossover_detail_indicator{str(ema_canditate[0])}-{str(ema_canditate[1])}-{str(ema_canditate[2])}_{len(self.data['company'])}company_{now_strting}.csv",
index=False,
)
if verbosity > 0:
logger.debug(
f"Exported at {export_path}/ema_crossover_detail_indicator{str(ema_canditate[0])}-{str(ema_canditate[1])}-{str(ema_canditate[2])}_{len(self.data['company'])}company_{now_strting}.csv"
)
else:
return ema_quote
def ema_indicator(self,
ema_canditate: Tuple[int, int] = (50, 200),
cutoff_date: Union[str, datetime.datetime] = 'today',
save: bool = True,
export_path: str = '.',
verbosity: int = 1) -> pd.DataFrame:
"""Exponential moving average based on desired two period (or no of days)
Parameters
----------
ema_canditate : Tuple[int, int], optional
[description], by default (50, 200)
cutoff_date : Union[str,datetime.datetime], optional
Desired date till which to calculate ema. 'today' for current day,
eg 01/01/2020 for any other date, by default 'today'
save : bool, optional
Save to hard disk, by default True
export_path : str, optional
Path to save, to be used only if 'save' is true, by default '.'
verbosity : int, optional
Level of detail logging,1=< Deatil, 0=Less detail , by default 1
Returns
-------
-> pd.DataFrame
EMA and indicators based on it
"""
with multiprocessing.Pool(multiprocessing.cpu_count() - 1) as pool:
result = pool.starmap(
self._parallel_ema_indicator,
[(company, ema_canditate, cutoff_date, verbosity)
for company in self.data['company']])
ema_indicator_df = pd.DataFrame(result)
ema_indicator_df.dropna(inplace=True)
ema_indicator_df['percentage_diff'] = ema_indicator_df.apply(
lambda x: percentage_diff_analysis(
x[f'ema{str(ema_canditate[0])}'], x[
f'ema{str(ema_canditate[1])}']),
axis=1)
ema_indicator_df['outcome'] = ema_indicator_df.apply(
lambda x: outcome_analysis(x['percentage_diff']), axis=1)
ema_indicator_df = ema_indicator_df[[
'company', 'ema_date', f'ema{str(ema_canditate[0])}',
f'ema{str(ema_canditate[1])}', 'percentage_diff', 'outcome',
'action'
]]
if verbosity > 0:
logger.debug(
f"Here are sample 5 company\n{ema_indicator_df.head()}")
if save is True:
ema_indicator_df.to_csv(
f"{export_path}/ema_indicator{str(ema_canditate[0])}-{str(ema_canditate[1])}_{len(self.data['company'])}company_{now_strting}.csv",
index=False)
if verbosity > 0:
logger.debug(
f"Exported at {export_path}/ema_indicator{str(ema_canditate[0])}-{str(ema_canditate[1])}_{len(self.data['company'])}company_{now_strting}.csv"
)
else:
return ema_indicator_df
def ema_crossover_indicator_detail(self,
ema_canditate: Tuple[int, int,
int] = (5, 13, 26),
save: bool = True,
export_path: str = '.',
verbosity: int = 1) -> pd.DataFrame:
"""Exponential moving average for crossover triple period technique
Parameters
----------
ema_canditate : Tuple[int, int, int], optional
Three Period (or days) to calculate EMA, by default (5,13,26)
save : bool, optional
Save to hard disk, by default True
export_path : str, optional
Path to save, to be used only if 'save' is true, by default '.'
verbosity : int, optional
Level of detail logging,1=< Deatil, 0=Less detail , by default 1
Returns
-------
pd.DataFrame
"""
logger.info("Performing EMA Indicator Task")
ema_short = self._ema_indicator_n3(ema_canditate=ema_canditate,
verbosity=verbosity)
logger.info("Extarcting detail company quote data")
batch_company_quote = pd.DataFrame()
with multiprocessing.Pool(multiprocessing.cpu_count() - 1) as pool:
company_quote = pool.map(self._parallel_quote_retrive,
ema_short['company'])
for single_company_quote in company_quote:
batch_company_quote = batch_company_quote.append(
single_company_quote)
batch_company_quote = batch_company_quote.reset_index().rename(
columns={'index': 'company'})
batch_company_quote = batch_company_quote[[
'company', 'longName', 'price', 'regularMarketVolume', 'marketCap',
'bookValue', 'priceToBook', 'averageDailyVolume3Month',
'averageDailyVolume10Day', 'fiftyTwoWeekLowChange',
'fiftyTwoWeekLowChangePercent', 'fiftyTwoWeekRange',
'fiftyTwoWeekHighChange', 'fiftyTwoWeekHighChangePercent',
'fiftyTwoWeekLow', 'fiftyTwoWeekHigh'
]]
batch_company_quote['company'] = batch_company_quote[
'company'].str.replace('.NS', '')
ema_quote = ema_short.merge(batch_company_quote,
on='company',
validate='1:1')
if verbosity > 0:
logger.debug(f"Here are sample 5 company\n{ema_quote.head()}")
if save is not False:
ema_quote.to_csv(
f"{export_path}/ema_indicator_detail{str(ema_canditate[0])}-{str(ema_canditate[1])}_{len(self.data['company'])}company_{now_strting}.csv",
index=False)
if verbosity > 0:
logger.debug(
f"Exported at {export_path}/ema_indicator_detail{str(ema_canditate[0])}-{str(ema_canditate[1])}_{len(self.data['company'])}company_{now_strting}.csv"
)
else:
return ema_quote
def relative_momentum(
self,
end_date: str = "today",
top_company_count: int = 20,
save: bool = True,
export_path: str = ".",
verbosity: int = 1,
) -> Optional[pd.DataFrame]:
"""The strategy is used to identity stocks which had 'good performance'
based on desired 'return' duration
Args:
end_date (str, optional): End date of of stock record to retrive. Must be in format: dd/mm/yyyy. Defaults to 'today'.
top_company_count (int, optional): No of top company to retrieve based on Annualized return. Defaults to 20.
save (bool, optional): Wether to export to disk. Defaults to True.
export_path (str, optional): Path to export csv.To be used only if 'save' is True. Defaults to '.'.
verbosity (int, optional): Level of detail logging,1=< Deatil, 0=Less detail. Defaults to 1.
Returns:
Record based on monthly and yearly calculation
Example:
```python
from stock_analysis import MomentumStrategy
sa = MomentumStrategy('./data/company_list.yaml')
ms = sa.relative_momentum(end_date='01/06/2020')
```
"""
if end_date == "today":
end = datetime.datetime.now()
else:
end = datetime.datetime.strptime(end_date, "%d/%m/%Y").date()
start = end - dateutil.relativedelta.relativedelta(years=1)
with parallel_backend(n_jobs=-1, backend="multiprocessing"):
result = Parallel()(
delayed(self.unit_momentum)(company, start, end, verbosity)
for company in self.data["company"])
momentum_df = pd.DataFrame(result)
momentum_df.dropna(inplace=True)
momentum_df.sort_values(by=["return_yearly"],
ascending=False,
inplace=True)
momentum_df.reset_index(inplace=True, drop=True)
if verbosity > 0:
logger.debug(
f"Sample output:\n{momentum_df.head(top_company_count)}")
if save is True:
new_folder(export_path)
momentum_df.head(top_company_count).to_csv(
f"{export_path}/momentum_result_{end.strftime('%d-%m-%Y')}_top_{top_company_count}.csv",
index=False,
)
if verbosity > 0:
logger.debug(
f"Saved at {export_path}/momentum_result_{end.strftime('%d-%m-%Y')}_top_{top_company_count}.csv"
)
else:
return momentum_df.head(top_company_count)
def relative_momentum_with_ema(
self,
end_date: str = "today",
top_company_count: int = 20,
ema_canditate: Tuple[int, int] = (50, 200),
save: bool = True,
export_path: str = ".",
verbosity: int = 1,
) -> Optional[pd.DataFrame]:
"""The strategy is used to identity stocks with 'good performance'
based on desired 'return' duration and 'exponential moving avg'.
Args:
end_date (str, optional): End date of of stock record to retrive. Must be in format: dd/mm/yyyy. Defaults to 'today'.
top_company_count (int, optional): No of top company to retrieve based on Annualized return. Defaults to 20.
ema_canditate (Tuple[int, int], optional): Period (or days) to calculate EMA. Defaults to (50, 200).
save (bool, optional): Wether to export to disk. Defaults to True.
export_path (str, optional): Path to export csv.To be used only if 'save' is True. Defaults to '.'.
verbosity (int, optional): Level of detail logging,1=< Deatil, 0=Less detail. Defaults to 1.
Returns:
Record based on monthly and yearly calculation and EMA calculation
Example:
```python
from stock_analysis import MomentumStrategy
sa = MomentumStrategy('./data/company_list.yaml')
mes = sa.relative_momentum_with_ema('01/06/2020', 30)
```
"""
logger.info("Performing Momentum Strategy task")
momentum_df = self.relative_momentum(
end_date=end_date,
top_company_count=top_company_count,
save=False,
verbosity=verbosity,
)
momentum_df.reset_index(drop=True, inplace=True)
ind = Indicator(company_name=momentum_df.loc[:, "company"])
logger.info(
f"Performing EMA task on top {top_company_count} company till {end_date}"
)
if end_date == "today":
cutoff_date = end_date
save_date = datetime.datetime.now().strftime("%d-%m-%Y")
else:
save_date = end_date.replace("/", "-")
cutoff_date = datetime.datetime.strptime(end_date, "%d/%m/%Y")
assert isinstance(cutoff_date,
datetime.datetime), "Incorrect date type"
ema_df = ind.ema_indicator(
ema_canditate=ema_canditate,
cutoff_date=cutoff_date,
save=False,
verbosity=verbosity,
)
momentum_ema_df = momentum_df.merge(ema_df,
on="company",
validate="1:1")
if save is True:
new_folder(export_path)
momentum_ema_df.reset_index(drop=True, inplace=True)
momentum_ema_df.to_csv(
f"{export_path}/momentum_ema{ema_canditate[0]}-{ema_canditate[1]}_{save_date}_top_{top_company_count}.csv",
index=False,
)
logger.debug(
f"Saved at {export_path}/momentum_ema{ema_canditate[0]}-{ema_canditate[1]}_{save_date}_top_{top_company_count}.csv"
)
if verbosity > 0:
logger.debug(f"Sample output:\n{momentum_ema_df.head()}")
else:
return momentum_ema_df
def momentum_strategy(self,
end_date: str = 'today',
top_company_count: int = 20,
save: bool = True,
export_path: str = '.',
verbosity: int = 1) -> pd.DataFrame:
"""
The strategy is used to identity stocks which had 'good performance'
based on desired 'return' duration
eg
>>>from stock_analysis import UnitStrategy
>>>sa = UnitStrategy('./data/company_list.yaml')
>>>sa.momentum_strategy(end_date='01/06/2020')
Parameters
----------
end_date : str, optional
End date of of stock record to retrive.
Must be in format: dd/mm/yyyy, by default 'today' for current date
top_company_count : int, optional
No of top company to retrieve based on
Annualized return, by default 20
save : int, optional
Wether to export to disk, by default True
export_path : str, optional
Path to export csv.To be used only if 'save' is True,by default'.'
verbosity : int, optional
Level of detail logging,1=< Deatil, 0=Less detail , by default 1
Returns
-------
pd.DataFrame
Record based on monthly and yearly calculation
"""
if end_date == 'today':
end = datetime.datetime.now()
else:
end = datetime.datetime.strptime(end_date, '%d/%m/%Y').date()
start = end - dateutil.relativedelta.relativedelta(years=1)
with multiprocessing.Pool(multiprocessing.cpu_count() - 1) as pool:
result = pool.starmap(self._parallel_momentum,
[(company, start, end, verbosity)
for company in self.data['company']])
momentum_df = pd.DataFrame(result)
momentum_df.dropna(inplace=True)
momentum_df.sort_values(by=['return_yearly'],
ascending=False,
inplace=True)
if verbosity > 0:
logger.debug(
f"Sample output:\n{momentum_df.head(top_company_count)}")
if save is True:
momentum_df.head(top_company_count).to_csv(
f"{export_path}/momentum_result_{end.strftime('%d-%m-%Y')}_top_{top_company_count}.csv",
index=False)
if verbosity > 0:
logger.debug(
f"Saved at {export_path}/momentum_result_{end.strftime('%d-%m-%Y')}_top_{top_company_count}.csv"
)
else:
return momentum_df.head(top_company_count)
def momentum_with_ema_strategy(self,
end_date: str = 'today',
top_company_count: int = 20,
ema_canditate: Tuple[int, int] = (50, 200),
save: bool = True,
export_path: str = '.',
verbosity: int = 1) -> pd.DataFrame:
"""The strategy is used to identity stocks with 'good performance'
based on desired 'return' duration and 'exponential moving avg'.
Parameters
----------
end_date : str, optional
End date of of stock record to retrive.
Must be in format: dd/mm/yyyy, by default 'today' for current date
top_company_count : int, optional
No of top company to retrieve based on Annualized return,
by default 20
ema_canditate : Tuple[int, int], optional
Period (or days) to calculate EMA, by default (50,200)
save : int, optional
Wether to export to disk, by default True
export_path : str, optional
Path to export csv.To be used only if 'save' is True,by default'.'
verbosity : int, optional
Level of detail logging,1=< Deatil, 0=Less detail , by default 1
Returns
-------
pd.DataFrame
Record based on monthly and yearly calculation and EMA calculation
"""
logger.info("Performing Momentum Strategy task")
momentum_df = self.momentum_strategy(
end_date=end_date,
top_company_count=top_company_count,
save=False,
verbosity=verbosity)
momentum_df.reset_index(drop=True, inplace=True)
ind = Indicator(company_name=momentum_df.loc[:, 'company'])
logger.info(
f"Performing EMA task on top {top_company_count} company till {end_date}"
)
if end_date == 'today':
cutoff_date = end_date
save_date = datetime.datetime.now().strftime('%d-%m-%Y')
else:
save_date = end_date.replace('/', '-')
cutoff_date = datetime.datetime.strptime(end_date, '%d/%m/%Y')
assert isinstance(cutoff_date,
datetime.datetime), 'Incorrect date type'
ema_df = ind.ema_indicator(ema_canditate=ema_canditate,
cutoff_date=cutoff_date,
save=False,
verbosity=verbosity)
momentum_ema_df = momentum_df.merge(ema_df,
on='company',
validate='1:1')
if save is True:
momentum_ema_df.reset_index(drop=True, inplace=True)
momentum_ema_df.to_csv(
f"{export_path}/momentum_ema{ema_canditate[0]}-{ema_canditate[1]}_{save_date}_top_{top_company_count}.csv",
index=False)
logger.debug(
f"Saved at {export_path}/momentum_ema{ema_canditate[0]}-{ema_canditate[1]}_{save_date}_top_{top_company_count}.csv"
)
if verbosity > 0:
logger.debug(f"Sample output:\n{momentum_ema_df.head()}")
else:
return momentum_ema_df