def test_portfolio_without_value():
port = Portfolio(date='2018-03-19',
cash=1000.21,
positions=dict(GAZP=682, VSMO=145, TTLK=123))
date = pd.to_datetime('2018-03-19').date()
assert port.date == date
assert f'Дата - {date}' in port.__str__()
assert port.value[PORTFOLIO] == 3_699_111.41
assert port.weight['VSMO'] == pytest.approx(0.691071)
def case_optimizer():
pos = dict(
AKRN=679,
BANEP=392,
CHMF=173,
GMKN=139,
LKOH=123,
LSNGP=59,
LSRG=1341,
MSRS=38,
MSTT=2181,
MTSS=1264,
MVID=141,
PMSBP=2715,
RTKMP=1674,
SNGSP=263,
TTLK=234,
UPRO=1272,
VSMO=101,
)
port = Portfolio(date="2018-07-24", cash=102_262, positions=pos)
save_max_trade = optimizer.MAX_TRADE
save_cut_off = portfolio.VOLUME_CUT_OFF
save_dividends_metrics = metrics.DividendsMetrics
save_returns_metrics = metrics.ReturnsMetrics
optimizer.MAX_TRADE = 0.006
portfolio.VOLUME_CUT_OFF = 2.9 * optimizer.MAX_TRADE
metrics.DividendsMetrics = BaseDividendsMetrics
metrics.ReturnsMetrics = BaseReturnsMetrics
opt = Optimizer(port)
yield opt
optimizer.MAX_TRADE = save_max_trade
portfolio.VOLUME_CUT_OFF = save_cut_off
optimizer.DividendsMetrics = save_dividends_metrics
optimizer.ReturnsMetrics = save_returns_metrics
def make_test_portfolio():
save_class = metrics.ReturnsMetrics
metrics.ReturnsMetrics = BaseReturnsMetrics
port = Portfolio(date='2018-03-19',
cash=1000.21,
positions=dict(GAZP=682, VSMO=145, TTLK=123))
yield port
metrics.ReturnsMetrics = save_class
def trading():
"""Распечатка всей аналитики, необходимой для управления портфелем"""
port = Portfolio(date=DATE, cash=CASH, positions=POSITIONS)
optimizer = Optimizer(port)
print(optimizer.portfolio)
print(optimizer.dividends_metrics)
print(optimizer.returns_metrics)
print(optimizer)
def test_yields_vs_dividends_metrics_predicted_false():
positions = dict(GMKN=128,
MSRS=0,
SNGSP=0)
port = Portfolio(date='2018-08-30',
cash=7_764,
positions=positions)
metrics_yields = BaseDividendsMetrics(port).yields
tickers = tuple(key for key in sorted(positions))
itr = DividendsCasesIterator(tickers, pd.Timestamp('2018-08-30'), Freq.yearly)
df = itr._real_dividends_yields(pd.Timestamp('2018-08-30'), labels=False)
assert np.allclose(metrics_yields, df.iloc[:, 1:].T)
def test_yields_vs_dividends_metrics():
positions = dict(MSRS=128,
PRTK=0,
MVID=0)
port = Portfolio(date='2017-07-31',
cash=7_764,
positions=positions)
metrics_yields = BaseDividendsMetrics(port).yields
tickers = tuple(key for key in sorted(positions))
itr = DividendsCasesIterator(tickers, pd.Timestamp('2018-07-31'), Freq.yearly)
df = itr._real_dividends_yields(pd.Timestamp('2018-07-31'))
assert np.allclose(metrics_yields, df.iloc[:, 1:-1].T)
def test_iter():
iterator = DividendsCasesIterator(tuple(['AKRN', 'MTSS', 'PMSBP']), pd.Timestamp('2016-01-05'), Freq.yearly)
df = next(iter(iterator))
assert len(df) == 3
assert df.iat[0, 0] == 'AKRN'
assert df.iat[1, 0] == 'MTSS'
assert df.iat[2, 0] == 'PMSBP'
positions = dict(AKRN=128,
PMSBP=0,
MTSS=0)
port = Portfolio(date='2015-01-05',
cash=7_764,
positions=positions)
metrics_yields = BaseDividendsMetrics(port).yields
assert np.allclose(metrics_yields.T, df.iloc[:, 1:-1])
def test_cases_vs_dividends_metrics():
positions = dict(MSTT=128,
MVID=0,
ALRS=0)
port = Portfolio(date='2017-06-28',
cash=7_764,
positions=positions)
metrics_yields = BaseDividendsMetrics(port).yields
tickers = tuple(key for key in sorted(positions))
df = DividendsCasesIterator(tickers, pd.Timestamp('2018-07-31'), Freq.yearly).cases(pd.Timestamp('2018-06-28'))
assert len(df) == 3
assert df.iat[0, 0] == 'ALRS'
assert df.iat[1, 0] == 'MSTT'
assert df.iat[2, 0] == 'MVID'
assert np.allclose(metrics_yields.T, df.iloc[:, 1:-1])
assert not df['y'].isnull().any()
def test_cases_vs_dividends_metrics_predicted_false():
positions = dict(PRTK=128,
AKRN=0,
LKOH=0)
port = Portfolio(date='2018-07-31',
cash=7_764,
positions=positions)
metrics_yields = BaseDividendsMetrics(port).yields
tickers = tuple(key for key in sorted(positions))
df = DividendsCasesIterator(tickers, pd.Timestamp('2018-07-31'), Freq.yearly).cases(pd.Timestamp('2018-07-31'),
False)
assert len(df) == 3
assert df.iat[0, 0] == 'AKRN'
assert df.iat[1, 0] == 'LKOH'
assert df.iat[2, 0] == 'PRTK'
assert np.allclose(metrics_yields.T, df.iloc[:, 1:-1])
assert df['y'].isnull().all()
"""Реализует основные метрики дивидендного потока для портфеля с помощью ML-модели
ML-модели строится с помощью Gradient Boosting по данным с лагом о предыдущей дивидендной доходности и
категориальной переменной на основе тикеров. СКО прогноза вычисляется с помощью кросс-валидации
"""
@property
def _tickers_real_after_tax_mean(self):
portfolio = self._portfolio
manager = ml.dividends.manager.DividendsMLDataManager(
portfolio.positions[:-2], pd.Timestamp(portfolio.date))
return manager.value.prediction_mean
@property
def _tickers_real_after_tax_std(self):
portfolio = self._portfolio
manager = ml.dividends.manager.DividendsMLDataManager(
portfolio.positions[:-2], pd.Timestamp(portfolio.date))
return manager.value.prediction_std
if __name__ == '__main__':
from trading import POSITIONS, DATE, CASH
port = Portfolio(date=DATE, cash=CASH, positions=POSITIONS)
metrics = MLDividendsMetrics(port)
dsr = (metrics.mean -
metrics.beta * metrics.mean.iloc[-1]) / metrics.std.iloc[-1]
print(dsr.sort_values(ascending=False))
print(metrics.mean.iloc[-1] / metrics.std.iloc[-1])
def case_metrics():
positions = dict(AKRN=679, GMKN=139, LSRG=172, MSTT=2436, TTLK=234)
port = Portfolio(date='2018-05-11', cash=311_587, positions=positions)
return BaseDividendsMetrics(port)
def test_end_of_month_shape():
positions = dict(AKRN=679, GMKN=139, LSRG=172, MSTT=2436, TTLK=234)
port = Portfolio(date='2018-07-30', cash=311_587, positions=positions)
return BaseDividendsMetrics(port).real_after_tax.shape == (5, 7)
t = ((t_score * s) / (2 * m)) ** 2
"""
return (self.std[PORTFOLIO] * T_SCORE / 2 /
self.mean[PORTFOLIO])**2 * 12
@property
def std_at_draw_down(self):
"""СКО стоимости портфеля в момент оптимизации
"""
return self.std[PORTFOLIO] * (MONTH_TO_OPTIMIZE / 12)**0.5
if __name__ == '__main__':
import trading
port = Portfolio(date=trading.DATE,
cash=trading.CASH,
positions=trading.POSITIONS)
metrics = MLReturnsMetrics(port)
dsr = (metrics.mean -
metrics.beta * metrics.mean.iloc[-1]) / metrics.std.iloc[-1]
print(dsr.sort_values(ascending=False))
print(metrics.mean.iloc[-1] / metrics.std.iloc[-1])
t = 2 / (12**0.5)
mean_gradient_ = metrics.mean - metrics.mean.iloc[-1]
risk_gradient_ = metrics.std.iloc[-1] * (metrics.beta - 1)
gr = (mean_gradient_ - t * risk_gradient_) / (t * metrics.std.iloc[-1])
print(t)
print(gr.sort_values(ascending=False))
def test_ticker_with_empty_history():
port = Portfolio(date='2018-03-19',
cash=1000.21,
positions=dict(GAZP=682, VSMO=145, KUNF=123))
assert port.price['KUNF'] == 0
def _tickers_real_after_tax_mean(self):
return self.yields.mean(axis='index', skipna=False)
@property
def _tickers_real_after_tax_std(self):
return self.yields.std(axis='index', ddof=1, skipna=False)
if __name__ == '__main__':
pos = dict(AKRN=676,
BANEP=392,
CHMF=173,
GMKN=139,
LKOH=183,
LSNGP=596,
LSRG=2346,
MSRS=128,
MSTT=1823,
MTSS=1348,
MVID=141,
PMSBP=2715,
RTKMP=1674,
SNGSP=263,
TTLK=234,
UPRO=1272,
VSMO=101)
port = Portfolio(date='2018-07-27',
cash=2_482 + 7_764 + 3_416,
positions=pos)
print(BaseDividendsMetrics(port))
AFLT=0,
MSRS=699,
UPRO=1267,
PMSBP=1188 + 322 + 219,
CHMF=0,
GMKN=166 + 28,
VSMO=73,
RSTIP=87,
PHOR=0,
MRSB=0,
LKOH=123,
ENRU=319 + 148,
MVID=264 + 62)
CASH = 596_156 + 470_259 + 481_849
DATE = '2018-04-19'
TEST_PORTFOLIO = Portfolio(date=DATE, cash=CASH, positions=POSITIONS)
def test_drop_small_positions():
df = drop_small_positions(TEST_PORTFOLIO)
index = df.index
assert len(df) == MAX_TABLE_ROWS + 2
assert index[-1] == PORTFOLIO
assert df[PORTFOLIO] == pytest.approx(39226178)
assert index[-2] == OTHER
assert df[OTHER] == pytest.approx(6523354)
assert index[0] == 'RTKMP'
assert df.iloc[0] == pytest.approx(11167816)
assert index[-3] == 'MVID'
assert df.iloc[-3] == pytest.approx(1310520)