stable/scripts/plot_profit.py

161 lines
5.1 KiB
Python
Raw Normal View History

2018-01-12 09:55:49 +00:00
#!/usr/bin/env python3
import sys
import json
import matplotlib.pyplot as plt
import numpy as np
import freqtrade.optimize as optimize
import freqtrade.misc as misc
from freqtrade.strategy.strategy import Strategy
2018-01-12 09:55:49 +00:00
2018-01-23 05:20:17 +00:00
def plot_parse_args(args):
2018-01-12 09:55:49 +00:00
parser = misc.common_args_parser('Graph utility')
# FIX: perhaps delete those backtesting options that are not feasible (shows up in -h)
2018-01-12 09:55:49 +00:00
misc.backtesting_options(parser)
parser.add_argument(
'-p', '--pair',
2018-01-23 05:20:17 +00:00
help='Show profits for only this pairs. Pairs are comma-separated.',
dest='pair',
default=None
2018-01-12 09:55:49 +00:00
)
return parser.parse_args(args)
2018-01-12 21:15:50 +00:00
# data:: [ pair, profit-%, enter, exit, time, duration]
# data:: ['BTC_XMR', 0.00537847, '1511176800', '1511178000', 5057, 1]
# FIX: make use of the enter/exit dates to insert the
# profit more precisely into the pg array
def make_profit_array(data, px, filter_pairs=[]):
pg = np.zeros(px)
2018-01-12 09:55:49 +00:00
# Go through the trades
# and make an total profit
# array
for trade in data:
pair = trade[0]
if filter_pairs and pair not in filter_pairs:
2018-01-12 09:55:49 +00:00
continue
profit = trade[1]
2018-01-12 21:15:50 +00:00
tim = trade[4]
dur = trade[5]
2018-01-12 09:55:49 +00:00
pg[tim+dur-1] += profit
# rewrite the pg array to go from
# total profits at each timeframe
# to accumulated profits
pa = 0
2018-01-23 05:20:17 +00:00
for x in range(0, len(pg)):
2018-01-12 09:55:49 +00:00
p = pg[x] # Get current total percent
pa += p # Add to the accumulated percent
pg[x] = pa # write back to save memory
return pg
def plot_profit(args) -> None:
"""
Plots the total profit for all pairs.
Note, the profit calculation isn't realistic.
But should be somewhat proportional, and therefor useful
in helping out to find a good algorithm.
"""
# We need to use the same pairs, same tick_interval
# and same timeperiod as used in backtesting
# to match the tickerdata against the profits-results
filter_pairs = args.pair
2018-01-12 09:55:49 +00:00
config = misc.load_config(args.config)
config.update({'strategy': args.strategy})
# Init strategy
strategy = Strategy()
strategy.init(config)
2018-01-12 09:55:49 +00:00
pairs = config['exchange']['pair_whitelist']
if filter_pairs:
filter_pairs = filter_pairs.split(',')
pairs = list(set(pairs) & set(filter_pairs))
print('Filter, keep pairs %s' % pairs)
2018-01-12 09:55:49 +00:00
tickers = optimize.load_data(args.datadir, pairs=pairs,
ticker_interval=strategy.ticker_interval,
2018-01-12 09:55:49 +00:00
refresh_pairs=False)
dataframes = optimize.preprocess(tickers)
# Make an average close price of all the pairs that was involved.
# this could be useful to gauge the overall market trend
# FIX: since the dataframes are of unequal length,
# andor has different dates, we need to merge them
# But we dont have the date information in the
# backtesting results, this is needed to match the dates
# For now, assume the dataframes are aligned.
max_x = 0
for pair, pair_data in dataframes.items():
n = len(pair_data['close'])
max_x = max(max_x, n)
# if max_x != n:
# raise Exception('Please rerun script. Input data has different lengths %s'
# %('Different pair length: %s <=> %s' %(max_x, n)))
2018-01-23 05:20:17 +00:00
print('max_x: %s' % (max_x))
2018-01-12 09:55:49 +00:00
# We are essentially saying:
# array <- sum dataframes[*]['close'] / num_items dataframes
# FIX: there should be some onliner numpy/panda for this
avgclose = np.zeros(max_x)
2018-01-12 09:55:49 +00:00
num = 0
for pair, pair_data in dataframes.items():
close = pair_data['close']
maxprice = max(close) # Normalize price to [0,1]
2018-01-23 05:20:17 +00:00
print('Pair %s has length %s' % (pair, len(close)))
for x in range(0, len(close)):
avgclose[x] += close[x] / maxprice
# avgclose += close
num += 1
2018-01-12 09:55:49 +00:00
avgclose /= num
# Load the profits results
# And make an profits-growth array
filename = 'backtest-result.json'
with open(filename) as file:
2018-01-23 05:20:17 +00:00
data = json.load(file)
pg = make_profit_array(data, max_x, filter_pairs)
2018-01-12 09:55:49 +00:00
#
# Plot the pairs average close prices, and total profit growth
#
fig, (ax1, ax2, ax3) = plt.subplots(3, sharex=True)
fig.suptitle('total profit')
ax1.plot(avgclose, label='avgclose')
ax2.plot(pg, label='profit')
ax1.legend(loc='upper left')
ax2.legend(loc='upper left')
2018-01-12 09:55:49 +00:00
# FIX if we have one line pair in paris
# then skip the plotting of the third graph,
# or change what we plot
# In third graph, we plot each profit separately
for pair in pairs:
pg = make_profit_array(data, max_x, pair)
2018-01-12 09:55:49 +00:00
ax3.plot(pg, label=pair)
ax3.legend(loc='upper left')
# black background to easier see multiple colors
ax3.set_facecolor('black')
2018-01-12 09:55:49 +00:00
# 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__':
args = plot_parse_args(sys.argv[1:])
plot_profit(args)