def test_retrieving_intermittently_cached_daily(tmpdir, provider):
retriever = HistoricalRetriever(provider=provider, hist_data_dir=tmpdir,)
start_date = date(2020, 3, 3)
end_date = date(2020, 3, 3)
retriever.retrieve_bar_data(
symbol="SPY",
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
start_date = date(2020, 3, 5)
end_date = date(2020, 3, 5)
retriever.retrieve_bar_data(
symbol="SPY",
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
start_date = date(2020, 3, 2)
end_date = date(2020, 3, 6)
data = retriever.retrieve_bar_data(
symbol="SPY",
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
validate_data_range(data=data, start_date=start_date, end_date=end_date)
def test_retrieving_intermittently_cached_intraday(tmpdir, provider):
retriever = HistoricalRetriever(provider=provider, hist_data_dir=tmpdir,)
data = pd.DataFrame()
dates = generate_trading_days(
start_date=date.today() - timedelta(days=10),
end_date=date.today() - timedelta(days=1),
)
date_ranges = [
dates[1:2],
dates[-3:-2],
dates,
]
for date_range in date_ranges:
start_date = date_range[0]
end_date = date_range[-1]
data = retriever.retrieve_bar_data(
symbol="SPY",
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
validate_data_range(data=data, start_date=dates[0], end_date=dates[-1])
def test_retrieving_intermittently_cached_intraday(tmpdir, provider):
retriever = HistoricalRetriever(provider=provider, hist_data_dir=tmpdir,)
data = pd.DataFrame()
dates = generate_trading_days(
start_date=date.today() - timedelta(days=10),
end_date=date.today() - timedelta(days=1),
)
date_ranges = [
dates[1:2],
dates[-3:-2],
dates,
]
for date_range in date_ranges:
start_date = date_range[0]
end_date = date_range[-1]
contract = StockContract(symbol="SPY")
try:
data = retriever.retrieve_bar_data(
contract=contract,
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
except NotImplementedError: # todo: fix
return
validate_data_range(data=data, start_date=dates[0], end_date=dates[-1])
def __init__(self, start_date: date, end_date: date, bar_size: timedelta):
hist_retriever = HistoricalRetriever(hist_data_dir=TEST_DATA_DIR)
self._sim_data = hist_retriever.retrieve_bar_data(
symbol="SPY",
bar_size=bar_size,
start_date=start_date,
end_date=end_date,
cache_only=True,
)
self._sim_idx = 0
def test_retrieve_non_cached_daily(tmpdir, provider):
start_date = date(2020, 4, 1)
end_date = date(2020, 4, 2)
retriever = HistoricalRetriever(provider=provider, hist_data_dir=tmpdir,)
data = retriever.retrieve_bar_data(
symbol="SPY",
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
validate_data_range(data=data, start_date=start_date, end_date=end_date)
def test_retrieving_intermittently_cached_daily(tmpdir, provider):
retriever = HistoricalRetriever(provider=provider, hist_data_dir=tmpdir)
start_date = date(2020, 3, 3)
end_date = date(2020, 3, 3)
contract = StockContract(symbol="SPY")
try:
retriever.retrieve_bar_data(
contract=contract,
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
except NotImplementedError: # todo: fix
return
start_date = date(2020, 3, 5)
end_date = date(2020, 3, 5)
retriever.retrieve_bar_data(
contract=contract,
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
start_date = date(2020, 3, 2)
end_date = date(2020, 3, 6)
data = retriever.retrieve_bar_data(
contract=contract,
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
validate_data_range(data=data, start_date=start_date, end_date=end_date)
def test_retrieve_non_cached_intraday(tmpdir, provider):
start_date = date.today() - timedelta(days=7)
end_date = date.today() - timedelta(days=1)
retriever = HistoricalRetriever(provider=provider, hist_data_dir=tmpdir,)
data = retriever.retrieve_bar_data(
symbol="SPY",
start_date=start_date,
end_date=end_date,
bar_size=timedelta(minutes=1),
)
validate_data_range(data=data, start_date=start_date, end_date=end_date)
assert np.isclose(len(data), 5 * 6.5 * 60, atol=7 * 60)
def test_retriever_cached_daily():
start_date = date(2020, 4, 1)
end_date = date(2020, 4, 2)
retriever = HistoricalRetriever(hist_data_dir=TEST_DATA_DIR)
data = retriever.retrieve_bar_data(
symbol="SPY",
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
cache_only=True,
)
assert len(data) != 0
validate_data_range(data=data, start_date=start_date, end_date=end_date)
def test_retriever_cached_tick():
start_date = date(2020, 6, 17)
end_date = date(2020, 6, 19)
retriever = HistoricalRetriever(hist_data_dir=TEST_DATA_DIR)
contract = StockContract(symbol="SCHW")
data = retriever.retrieve_bar_data(
contract=contract,
start_date=start_date,
end_date=end_date,
bar_size=timedelta(0),
cache_only=True,
)
assert len(data) != 0
validate_data_range(data=data, start_date=start_date, end_date=end_date)
def test_retrieve_non_cached_daily(tmpdir, provider):
start_date = date(2020, 4, 1)
end_date = date(2020, 4, 2)
retriever = HistoricalRetriever(provider=provider, hist_data_dir=tmpdir,)
contract = StockContract(symbol="SPY")
try:
data = retriever.retrieve_bar_data(
contract=contract,
start_date=start_date,
end_date=end_date,
bar_size=timedelta(days=1),
)
except NotImplementedError:
return
validate_data_range(data=data, start_date=start_date, end_date=end_date)
def test_historical_bar_aggregator():
start_date = date(2020, 4, 6)
end_date = date(2020, 4, 7)
retriever = HistoricalRetriever(hist_data_dir=TEST_DATA_DIR)
contract = StockContract(symbol="SPY")
base_data = retriever.retrieve_bar_data(
contract=contract,
start_date=start_date,
end_date=end_date,
bar_size=timedelta(minutes=1),
cache_only=True,
)
aggregator = HistoricalAggregator(hist_data_dir=TEST_DATA_DIR)
agg_data = aggregator.aggregate_data(
contract=contract,
start_date=start_date,
end_date=end_date,
base_bar_size=timedelta(minutes=1),
target_bar_size=timedelta(minutes=5),
)
assert len(agg_data) == 156
assert agg_data.iloc[0]["open"] == base_data.iloc[0]["open"]
assert agg_data.iloc[0]["close"] == base_data.iloc[4]["close"]
assert agg_data.iloc[0]["high"] == base_data.iloc[:5]["high"].max()
assert agg_data.iloc[0]["low"] == base_data.iloc[:5]["low"].min()
assert agg_data.iloc[0]["volume"] == base_data.iloc[:5]["volume"].sum()
agg_data = aggregator.aggregate_data(
contract=contract,
start_date=start_date,
end_date=end_date,
base_bar_size=timedelta(minutes=1),
target_bar_size=timedelta(minutes=10),
)
assert len(agg_data) == 78
def test_retrieve_non_cached_intraday(tmpdir, provider):
start_date = date.today() - timedelta(days=7)
end_date = date.today() - timedelta(days=1)
retriever = HistoricalRetriever(provider=provider, hist_data_dir=tmpdir)
contract = StockContract(symbol="SPY")
try:
data = retriever.retrieve_bar_data(
contract=contract,
start_date=start_date,
end_date=end_date,
bar_size=timedelta(minutes=1),
)
except NotImplementedError:
return
validate_data_range(data=data, start_date=start_date, end_date=end_date)
assert (
np.isclose(len(data), 5 * 6.5 * 60, atol=7 * 60)
or len(data) == 4800 # for outside RTHs IB
)
from datetime import timedelta, date
from algotradepy.historical.loaders import HistoricalRetriever
from algotradepy.historical.providers import YahooProvider
if __name__ == "__main__":
provider = YahooProvider()
hr = HistoricalRetriever(provider=provider)
end_date = date.today()
hr.retrieve_bar_data(
symbol="SPY",
bar_size=timedelta(days=1),
start_date=date(2000, 1, 1),
end_date=end_date,
)