stable/scripts/plot_dataframe.py

77 lines
2.6 KiB
Python
Raw Normal View History

2017-11-09 21:29:23 +00:00
#!/usr/bin/env python3
2018-01-06 07:38:47 +00:00
import sys
import argparse
2017-11-09 21:29:23 +00:00
import matplotlib # Install PYQT5 manually if you want to test this helper function
matplotlib.use("Qt5Agg")
import matplotlib.pyplot as plt
from freqtrade import exchange, analyze
2018-01-09 11:06:28 +00:00
from freqtrade.misc import common_args_parser
2017-11-09 21:29:23 +00:00
2018-01-06 07:38:47 +00:00
def plot_parse_args(args ):
2018-01-12 05:09:04 +00:00
parser = common_args_parser(description='Graph utility')
2018-01-06 07:38:47 +00:00
parser.add_argument(
'-p', '--pair',
help = 'What currency pair',
dest = 'pair',
default = 'BTC_ETH',
type = str,
)
parser.add_argument(
'-i', '--interval',
help = 'what interval to use',
dest = 'interval',
default = '5',
type = int,
)
2018-01-06 07:38:47 +00:00
return parser.parse_args(args)
2018-01-12 10:49:50 +00:00
def plot_analyzed_dataframe(args):
2017-11-09 21:29:23 +00:00
"""
Calls analyze() and plots the returned dataframe
:param pair: pair as str
:return: None
"""
# Init Bittrex to use public API
exchange._API = exchange.Bittrex({'key': '', 'secret': ''})
ticker = exchange.get_ticker_history(args.pair,args.interval)
2017-12-17 12:14:57 +00:00
dataframe = analyze.analyze_ticker(ticker)
2017-11-09 21:29:23 +00:00
dataframe.loc[dataframe['buy'] == 1, 'buy_price'] = dataframe['close']
dataframe.loc[dataframe['sell'] == 1, 'sell_price'] = dataframe['close']
2017-11-09 21:29:23 +00:00
# Two subplots sharing x axis
fig, (ax1, ax2, ax3) = plt.subplots(3, sharex=True)
fig.suptitle(args.pair + " " + str(args.interval), fontsize=14, fontweight='bold')
2017-11-09 21:29:23 +00:00
ax1.plot(dataframe.index.values, dataframe['close'], label='close')
# ax1.plot(dataframe.index.values, dataframe['sell'], 'ro', label='sell')
ax1.plot(dataframe.index.values, dataframe['sma'], '--', label='SMA')
ax1.plot(dataframe.index.values, dataframe['tema'], ':', label='TEMA')
ax1.plot(dataframe.index.values, dataframe['blower'], '-.', label='BB low')
ax1.plot(dataframe.index.values, dataframe['buy_price'], 'bo', label='buy')
ax1.legend()
ax2.plot(dataframe.index.values, dataframe['adx'], label='ADX')
ax2.plot(dataframe.index.values, dataframe['mfi'], label='MFI')
# ax2.plot(dataframe.index.values, [25] * len(dataframe.index.values))
ax2.legend()
ax3.plot(dataframe.index.values, dataframe['fastk'], label='k')
ax3.plot(dataframe.index.values, dataframe['fastd'], label='d')
ax3.plot(dataframe.index.values, [20] * len(dataframe.index.values))
ax3.legend()
# Fine-tune figure; make subplots close to each other and hide x ticks for
# all but bottom plot.
fig.subplots_adjust(hspace=0)
plt.setp([a.get_xticklabels() for a in fig.axes[:-1]], visible=False)
plt.show()
if __name__ == '__main__':
2018-01-06 07:38:47 +00:00
args = plot_parse_args(sys.argv[1:])
plot_analyzed_dataframe(args)