mirror of
https://github.com/freqtrade/freqtrade.git
synced 2024-11-10 10:21:59 +00:00
42 lines
1.3 KiB
Python
42 lines
1.3 KiB
Python
# pragma pylint: disable=missing-docstring
|
|
|
|
|
|
import json
|
|
import os
|
|
from typing import Optional, List, Dict
|
|
|
|
from pandas import DataFrame
|
|
|
|
from freqtrade.analyze import populate_indicators, parse_ticker_dataframe
|
|
|
|
|
|
def load_data(ticker_interval: int = 5, pairs: Optional[List[str]] = None) -> Dict[str, List]:
|
|
"""
|
|
Loads ticker history data for the given parameters
|
|
:param ticker_interval: ticker interval in minutes
|
|
:param pairs: list of pairs
|
|
:return: dict
|
|
"""
|
|
path = os.path.abspath(os.path.dirname(__file__))
|
|
result = {}
|
|
_pairs = pairs or [
|
|
"BTC_ETH", "BTC_LTC", "BTC_ETC", "BTC_DASH", "BTC_ZEC",
|
|
"BTC_XLM", "BTC_NXT", "BTC_POWR", "BTC_ADA", "BTC_XMR",
|
|
]
|
|
for pair in _pairs:
|
|
with open('{abspath}/../tests/testdata/{pair}-{ticker_interval}.json'.format(
|
|
abspath=path,
|
|
pair=pair,
|
|
ticker_interval=ticker_interval,
|
|
)) as tickerdata:
|
|
result[pair] = json.load(tickerdata)
|
|
return result
|
|
|
|
|
|
def preprocess(tickerdata: Dict[str, List]) -> Dict[str, DataFrame]:
|
|
"""Creates a dataframe and populates indicators for given ticker data"""
|
|
processed = {}
|
|
for pair, pair_data in tickerdata.items():
|
|
processed[pair] = populate_indicators(parse_ticker_dataframe(pair_data))
|
|
return processed
|