WARNING! The data type and the competition type are mismatch.

noka'sworld

I had tried to follow the documentation. and develop a really easy strategy to test out the function of the platform. it simply do a long position when volume in this hour>volume in previous hour but get the following error

import xarray as xr

import qnt.ta as qnta
import qnt.backtester as qnbt
import qnt.data as qndata



def load_hourdata(period):
    return qndata.crypto.load_data(tail = 365 * 5)



def strategy(data):
    crypto_data = qndata.crypto.load_data( tail = 365 * 5)
    BTC_vol = crypto_data.sel(field = 'vol').sel(asset = 'BTC')
    BTC_past_vol= qnta.shift(BTC_vol, periods=1)
    
    weights  = xr.where(BTC_vol > BTC_past_vol, 1, -1) # 1 - long position (**buy**), -1 - short position (**sell**)
    return weights



weights = qnbt.backtest(
    competition_type= "crypto_daily_long_short",
    load_data= load_data,
    lookback_period= 365*4,
    start_date= "2014-01-01",
    strategy= strategy,
    analyze= True,
    build_plots= True
)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-19-dcf17b33384f> in <module>
     29     strategy= strategy,
     30     analyze= True,
---> 31     build_plots= True
     32 )

~/book/qnt/backtester.py in backtest(competition_type, strategy, load_data, lookback_period, test_period, start_date, end_date, window, step, analyze, build_plots, collect_all_states)
    293     log_info("Load data for cleanup...")
    294     data = qndata.load_data_by_type(competition_type, assets=result.asset.values.tolist(), tail=60)
--> 295     result = qnout.clean(result, data)
    296     result.name = competition_type
    297     log_info("Write result...")

~/book/qnt/output.py in clean(output, data, kind, debug)
     66         # uniq asset fix
     67         val,idx = np.unique(output.asset, return_index=True)
---> 68         output = output.isel(asset=idx)
     69 
     70         if single_day:

/usr/local/lib/python3.7/site-packages/xarray/core/dataarray.py in isel(self, indexers, drop, missing_dims, **indexers_kwargs)
   1145         # lists, or zero or one-dimensional np.ndarray's
   1146 
-> 1147         variable = self._variable.isel(indexers, missing_dims=missing_dims)
   1148 
   1149         coords = {}

/usr/local/lib/python3.7/site-packages/xarray/core/variable.py in isel(self, indexers, missing_dims, **indexers_kwargs)
   1130         indexers = either_dict_or_kwargs(indexers, indexers_kwargs, "isel")
   1131 
-> 1132         indexers = drop_dims_from_indexers(indexers, self.dims, missing_dims)
   1133 
   1134         key = tuple(indexers.get(dim, slice(None)) for dim in self.dims)

/usr/local/lib/python3.7/site-packages/xarray/core/utils.py in drop_dims_from_indexers(indexers, dims, missing_dims)
    838         if invalid:
    839             raise ValueError(
--> 840                 f"Dimensions {invalid} do not exist. Expected one or more of {dims}"
    841             )
    842 

ValueError: Dimensions {'asset'} do not exist. Expected one or more of ('time',)

support

@noka-sworld Hello, we should definitely improve the documentation. When you use the "load_crypto" API you are loading data on an hourly basis. This is a functionality we provided some time ago when introducing cryptos (7 cryptos only).

Later we extended the crypto universe to more assets, and provided a universe containing the top 10 in terms of market capitalization at each point in time, and reverted to daily data.

The API to be used is:

import qnt.data as qndata

crypto_data = qndata.cryptodaily.load_data(tail = 365 * 5)

It will load crypto data on a daily basis and the code will not bug.