Merge branch 'develop' into timeperiod

This commit is contained in:
kryofly 2018-01-14 10:23:54 +01:00
commit f61012097c
12 changed files with 157 additions and 34 deletions

View File

@ -60,6 +60,21 @@ Example:
python3 ./freqtrade/main.py backtesting --timeperiod -200
```
**Update testdata directory
To update your testdata directory, or download into another testdata directory:
```bash
mkdir freqtrade/tests/testdata-20180113
cp freqtrade/tests/testdata/pairs.json freqtrade/tests/testdata-20180113
cd freqtrade/tests/testdata-20180113
Possibly edit pairs.json file to include/exclude pairs
python download_backtest_data.py -p pairs.json
```
The script will read your pairs.json file, and download ticker data
into the current working directory.
For help about backtesting usage, please refer to
[Backtesting commands](#backtesting-commands).

View File

@ -2,20 +2,70 @@
#### I have waited 5 minutes, why hasn't the bot made any trades yet?!
Depending on the buy strategy, the amount of whitelisted coins, the situation of the market etc, it can take up to hours to find good entry position for a trade. Be patient!
Depending on the buy strategy, the amount of whitelisted coins, the
situation of the market etc, it can take up to hours to find good entry
position for a trade. Be patient!
#### I have made 12 trades already, why is my total profit negative?!
I understand your disappointment but unfortunately 12 trades is just not enough to say anything. If you run backtesting, you can see that our current algorithm does leave you on the plus side, but that is after thousands of trades and even there, you will be left with losses on specific coins that you have traded tens if not hundreds of times. We of course constantly aim to improve the bot but it will _always_ be a gamble, which should leave you with modest wins on monthly basis but you can't say much from few trades.
I understand your disappointment but unfortunately 12 trades is just
not enough to say anything. If you run backtesting, you can see that our
current algorithm does leave you on the plus side, but that is after
thousands of trades and even there, you will be left with losses on
specific coins that you have traded tens if not hundreds of times. We
of course constantly aim to improve the bot but it will _always_ be a
gamble, which should leave you with modest wins on monthly basis but
you can't say much from few trades.
#### Id like to change the stake amount. Can I just stop the bot with /stop and then change the config.json and run it again?
#### Id like to change the stake amount. Can I just stop the bot with
/stop and then change the config.json and run it again?
Not quite. Trades are persisted to a database but the configuration is currently only read when the bot is killed and restarted. `/stop` more like pauses. You can stop your bot, adjust settings and start it again.
Not quite. Trades are persisted to a database but the configuration is
currently only read when the bot is killed and restarted. `/stop` more
like pauses. You can stop your bot, adjust settings and start it again.
#### I want to improve the bot with a new strategy
That's great. We have a nice backtesting and hyperoptimizing setup. See the tutorial [[here|Testing-new-strategies-with-Hyperopt]].
That's great. We have a nice backtesting and hyperoptimizing setup. See
the tutorial [here|Testing-new-strategies-with-Hyperopt](https://github.com/gcarq/freqtrade/blob/develop/docs/bot-usage.md#hyperopt-commands).
#### Is there a setting to only SELL the coins being held and not perform anymore BUYS?
#### Is there a setting to only SELL the coins being held and not
perform anymore BUYS?
You can use the `/forcesell all` command from Telegram.
### How many epoch do I need to get a good Hyperopt result?
Per default Hyperopts without `-e` or `--epochs` parameter will only
run 100 epochs, means 100 evals of your triggers, guards, .... Too few
to find a great result (unless if you are very lucky), so you probably
have to run it for 10.000 or more. But it will take an eternity to
compute.
We recommend you to run it at least 10.000 epochs:
```bash
python3 ./freqtrade/main.py hyperopt -e 10000
```
or if you want intermediate result to see
```bash
for i in {1..100}; do python3 ./freqtrade/main.py hyperopt -e 100; done
```
#### Why it is so long to run hyperopt?
Finding a great Hyperopt results takes time.
If you wonder why it takes a while to find great hyperopt results
This answer was written during the under the release 0.15.1, when we had
:
- 8 triggers
- 9 guards: let's say we evaluate even 10 values from each
- 1 stoploss calculation: let's say we want 10 values from that too to
be evaluated
The following calculation is still very rough and not very precise
but it will give the idea. With only these triggers and guards there is
already 8*10^9*10 evaluations. A roughly total of 80 billion evals.
Did you run 100 000 evals? Congrats, you've done roughly 1 / 100 000 th
of the search space.

View File

@ -40,7 +40,7 @@ cd freqtrade
```
4. Copy `config.sample` to `config.json`
```bash
cp config.json.example config.json
cp -n config.json.example config.json
```
To edit the config please refer to the [Bot Configuration](https://github.com/gcarq/freqtrade/blob/develop/docs/configuration.md) page
5. Create your DB file (Optional, the bot will create it if it is missing)
@ -194,7 +194,7 @@ The following steps are made for Linux/mac environment
## 4. Prepare the bot
```bash
cd freqtrade
cp config.json.example config.json
cp -n config.json.example config.json
```
To edit the config please refer to [Bot Configuration](https://github.com/gcarq/freqtrade/blob/develop/docs/configuration.md)

View File

@ -67,6 +67,18 @@ SET is_open=0, close_date='2017-12-20 03:08:45.103418', close_rate=0.19638016, c
WHERE id=31;
```
## Insert manually a new trade
```sql
INSERT
INTO trades (exchange, pair, is_open, fee, open_rate, stake_amount, amount, open_date)
VALUES ('BITTREX', 'BTC_<COIN>', 1, 0.0025, <open_rate>, <stake_amount>, <amount>, '<datetime>')
```
**Example:**
```sql
INSERT INTO trades (exchange, pair, is_open, fee, open_rate, stake_amount, amount, open_date) VALUES ('BITTREX', 'BTC_ETC', 1, 0.0025, 0.00258580, 0.002, 0.7715262081, '2017-11-28 12:44:24.000000')
```
## Fix wrong fees in the table
If your DB was created before

View File

@ -1,5 +1,5 @@
""" FreqTrade bot """
__version__ = '0.14.3'
__version__ = '0.15.1'
class DependencyException(BaseException):

View File

@ -123,10 +123,8 @@ class Bittrex(Exchange):
message=data['message'],
pair=pair))
if not data.get('result') \
or not data['result'].get('Bid') \
or not data['result'].get('Ask') \
or not data['result'].get('Last'):
if not data.get('result') or\
not all(key in data.get('result', {}) for key in ['Bid', 'Ask', 'Last']):
raise ContentDecodingError('{message} params=({pair})'.format(
message='Got invalid response from bittrex',
pair=pair))

View File

@ -15,6 +15,11 @@ from freqtrade import __version__
logger = logging.getLogger(__name__)
def file_dump_json(filename, data):
with open(filename, 'w') as fp:
json.dump(data, fp)
class State(enum.Enum):
RUNNING = 0
STOPPED = 1

View File

@ -212,7 +212,18 @@ def test_exchange_bittrex_get_ticker_bad():
wb = make_wrap_bittrex()
fb = FakeBittrex()
fb.result = {'success': True,
'result': {'Bid': 1}} # incomplete result
'result': {'Bid': 1, 'Ask': 0}} # incomplete result
with pytest.raises(ContentDecodingError, match=r'.*Got invalid response from bittrex params.*'):
wb.get_ticker('BTC_ETH')
fb.result = {'success': False,
'message': 'gone bad'
}
with pytest.raises(btx.OperationalException, match=r'.*gone bad.*'):
wb.get_ticker('BTC_ETH')
fb.result = {'success': True,
'result': {}} # incomplete result
with pytest.raises(ContentDecodingError, match=r'.*Got invalid response from bittrex params.*'):
wb.get_ticker('BTC_ETH')
fb.result = {'success': False,

View File

@ -1,29 +1,38 @@
#!/usr/bin/env python3
"""This script generate json data from bittrex"""
import sys
import json
from os import path
from freqtrade import exchange
from freqtrade.exchange import Bittrex
from freqtrade import misc
PAIRS = [
'BTC_BCC', 'BTC_ETH', 'BTC_MER', 'BTC_POWR', 'BTC_ETC',
'BTC_OK', 'BTC_NEO', 'BTC_EMC2', 'BTC_DASH', 'BTC_LSK',
'BTC_LTC', 'BTC_XZC', 'BTC_OMG', 'BTC_STRAT', 'BTC_XRP',
'BTC_QTUM', 'BTC_WAVES', 'BTC_VTC', 'BTC_XLM', 'BTC_MCO'
]
TICKER_INTERVAL = 5 # ticker interval in minutes (currently implemented: 1 and 5)
OUTPUT_DIR = path.dirname(path.realpath(__file__))
parser = misc.common_args_parser('download utility')
parser.add_argument(
'-p', '--pair',
help='JSON file containing pairs to download',
dest='pair',
default=None
)
args = parser.parse_args(sys.argv[1:])
TICKER_INTERVALS = [1, 5] # ticker interval in minutes (currently implemented: 1 and 5)
PAIRS = []
if args.pair:
with open(args.pair) as file:
PAIRS = json.load(file)
PAIRS = list(set(PAIRS))
print('About to download pairs:', PAIRS)
# Init Bittrex exchange
exchange._API = Bittrex({'key': '', 'secret': ''})
for pair in PAIRS:
data = exchange.get_ticker_history(pair, TICKER_INTERVAL)
filename = path.join(OUTPUT_DIR, '{}-{}.json'.format(
pair,
TICKER_INTERVAL,
))
with open(filename, 'w') as fp:
json.dump(data, fp)
for tick_interval in TICKER_INTERVALS:
print('downloading pair %s, interval %s' % (pair, tick_interval))
data = exchange.get_ticker_history(pair, tick_interval)
filename = '{}-{}.json'.format(pair, tick_interval)
misc.file_dump_json(filename, data)

23
freqtrade/tests/testdata/pairs.json vendored Normal file
View File

@ -0,0 +1,23 @@
[
"BTC_ADA",
"BTC_BAT",
"BTC_DASH",
"BTC_ETC",
"BTC_ETH",
"BTC_GBYTE",
"BTC_LSK",
"BTC_LTC",
"BTC_NEO",
"BTC_NXT",
"BTC_POWR",
"BTC_STORJ",
"BTC_QTUM",
"BTC_WAVES",
"BTC_VTC",
"BTC_XLM",
"BTC_XMR",
"BTC_XVG",
"BTC_XRP",
"BTC_ZEC"
]

View File

@ -11,7 +11,7 @@ scikit-learn==0.19.1
scipy==1.0.0
jsonschema==2.6.0
numpy==1.14.0
TA-Lib==0.4.14
TA-Lib==0.4.15
pytest==3.3.2
pytest-mock==1.6.3
pytest-cov==2.5.1
@ -19,7 +19,7 @@ hyperopt==0.1
# do not upgrade networkx before this is fixed https://github.com/hyperopt/hyperopt/issues/325
networkx==1.11
tabulate==0.8.2
pymarketcap==3.3.145
pymarketcap==3.3.148
# Required for plotting data
#matplotlib==2.1.0

View File

@ -10,7 +10,7 @@ from freqtrade.misc import common_args_parser
def plot_parse_args(args ):
parser = common_args_parser(args, 'Graph utility')
parser = common_args_parser(description='Graph utility')
parser.add_argument(
'-p', '--pair',
help = 'What currency pair',