Freqtrade is a free and open source crypto trading bot written in Python

Last update: Jan 07, 2023

Overview

Freqtrade

Freqtrade is a free and open source crypto trading bot written in Python. It is designed to support all major exchanges and be controlled via Telegram. It contains backtesting, plotting and money management tools as well as strategy optimization by machine learning.

Disclaimer

This software is for educational purposes only. Do not risk money which you are afraid to lose. USE THE SOFTWARE AT YOUR OWN RISK. THE AUTHORS AND ALL AFFILIATES ASSUME NO RESPONSIBILITY FOR YOUR TRADING RESULTS.

Always start by running a trading bot in Dry-run and do not engage money before you understand how it works and what profit/loss you should expect.

We strongly recommend you to have coding and Python knowledge. Do not hesitate to read the source code and understand the mechanism of this bot.

Supported Exchange marketplaces

Please read the exchange specific notes to learn about eventual, special configurations needed for each exchange.

Community tested

Exchanges confirmed working by the community:

Bitvavo

Documentation

We invite you to read the bot documentation to ensure you understand how the bot is working.

Please find the complete documentation on our website.

Features

Quick start

Freqtrade provides a Linux/macOS script to install all dependencies and help you to configure the bot.

git clone -b develop https://github.com/freqtrade/freqtrade.git 
cd freqtrade
./setup.sh --install

For any other type of installation please refer to Installation doc.

Basic Usage

Bot commands

usage: freqtrade [-h] [-V]
                 {trade,create-userdir,new-config,new-hyperopt,new-strategy,download-data,convert-data,convert-trade-data,backtesting,edge,hyperopt,hyperopt-list,hyperopt-show,list-exchanges,list-hyperopts,list-markets,list-pairs,list-strategies,list-timeframes,show-trades,test-pairlist,plot-dataframe,plot-profit}
                 ...

Free, open source crypto trading bot

positional arguments:
  {trade,create-userdir,new-config,new-hyperopt,new-strategy,download-data,convert-data,convert-trade-data,backtesting,edge,hyperopt,hyperopt-list,hyperopt-show,list-exchanges,list-hyperopts,list-markets,list-pairs,list-strategies,list-timeframes,show-trades,test-pairlist,plot-dataframe,plot-profit}
    trade               Trade module.
    create-userdir      Create user-data directory.
    new-config          Create new config
    new-hyperopt        Create new hyperopt
    new-strategy        Create new strategy
    download-data       Download backtesting data.
    convert-data        Convert candle (OHLCV) data from one format to
                        another.
    convert-trade-data  Convert trade data from one format to another.
    backtesting         Backtesting module.
    edge                Edge module.
    hyperopt            Hyperopt module.
    hyperopt-list       List Hyperopt results
    hyperopt-show       Show details of Hyperopt results
    list-exchanges      Print available exchanges.
    list-hyperopts      Print available hyperopt classes.
    list-markets        Print markets on exchange.
    list-pairs          Print pairs on exchange.
    list-strategies     Print available strategies.
    list-timeframes     Print available timeframes for the exchange.
    show-trades         Show trades.
    test-pairlist       Test your pairlist configuration.
    plot-dataframe      Plot candles with indicators.
    plot-profit         Generate plot showing profits.

optional arguments:
  -h, --help            show this help message and exit
  -V, --version         show program's version number and exit

Telegram RPC commands

Telegram is not mandatory. However, this is a great way to control your bot. More details and the full command list on our documentation

/start: Starts the trader.
/stop: Stops the trader.
/stopbuy: Stop entering new trades.
/status |[table]: Lists all or specific open trades.
/profit: Lists cumulative profit from all finished trades
/forcesell |all: Instantly sells the given trade (Ignoring minimum_roi).
/performance: Show performance of each finished trade grouped by pair
/balance: Show account balance per currency.
/daily : Shows profit or loss per day, over the last n days.
/help: Show help message.
/version: Show version.

Development branches

The project is currently setup in two main branches:

develop - This branch has often new features, but might also contain breaking changes. We try hard to keep this branch as stable as possible.
stable - This branch contains the latest stable release. This branch is generally well tested.
feat/* - These are feature branches, which are being worked on heavily. Please don't use these unless you want to test a specific feature.

Support

Help / Discord / Slack

For any questions not covered by the documentation or for further information about the bot, or to simply engage with like-minded individuals, we encourage you to join our slack channel.

Please check out our discord server.

You can also join our Slack channel.

Bugs / Issues

If you discover a bug in the bot, please search our issue tracker first. If it hasn't been reported, please create a new issue and ensure you follow the template guide so that our team can assist you as quickly as possible.

Feature Requests

Have you a great idea to improve the bot you want to share? Please, first search if this feature was not already discussed. If it hasn't been requested, please create a new request and ensure you follow the template guide so that it does not get lost in the bug reports.

Pull Requests

Feel like our bot is missing a feature? We welcome your pull requests!

Please read our Contributing document to understand the requirements before sending your pull-requests.

Coding is not a necessity to contribute - maybe start with improving our documentation? Issues labeled good first issue can be good first contributions, and will help get you familiar with the codebase.

Note before starting any major new feature work, please open an issue describing what you are planning to do or talk to us on discord or Slack. This will ensure that interested parties can give valuable feedback on the feature, and let others know that you are working on it.

Important: Always create your PR against the develop branch, not stable.

Requirements

Up-to-date clock

The clock must be accurate, synchronized to a NTP server very frequently to avoid problems with communication to the exchanges.

Min hardware required

To run this bot we recommend you a cloud instance with a minimum of:

Minimal (advised) system requirements: 2GB RAM, 1GB disk space, 2vCPU

Software requirements

Python 3.7.x
pip
git
TA-Lib
virtualenv (Recommended)
Docker (Recommended)

Comments

Bot couldn't sell because of insufficient fund

I was synced to HEAD yesterday (but I had the same issue some time ago).

I had an order that the bot couldn't sell because of insufficient fund. I think it's because either the bot did wrong calculation of the amount or the exchange is doing some funny stuff.

Logs when the order was fulfilled:

2018-11-29 03:05:08,831 - freqtrade.freqtradebot - INFO - Found open order for Trade(id=xx, pair=BCHSV/USDT, amount=xxx, open_rate=99.10000000, open_since=just now)
2018-11-29 03:05:09,740 - freqtrade.freqtradebot - WARNING - amount xxx does not match amount xxx
2018-11-29 03:05:09,740 - freqtrade.freqtradebot - WARNING - could not update trade amount: Half bought? Amounts don't match
2018-11-29 03:05:09,740 - freqtrade.persistence - INFO - Updating trade (id=xx) ...
2018-11-29 03:05:09,741 - freqtrade.persistence - INFO - LIMIT_BUY has been fulfilled for Trade(id=xx, pair=BCHSV/USDT, amount=xxx, open_rate=99.10000000, open_since=just now).

Logs when the bot was trying to sell:

2018-11-29 03:09:29,427 - freqtrade.freqtradebot - INFO - checking sell
2018-11-29 03:09:29,910 - freqtrade.freqtradebot - WARNING - Unable to sell trade: Insufficient funds to create limit sell order on market BCHSV/USDT.Tried to sell amount xx at rate 100.02 (total xx).Message: binance {"code":-2010,"msg":"Account has insufficient balance for requested action."}

The code that is throwing the error, any reason that instead of using the latest amount getting from exchange, we still choose to use the one saved when buying? https://github.com/freqtrade/freqtrade/blob/200484ab8ba85308e6e759bbec613dcba0e83d37/freqtrade/freqtradebot.py#L603

Bug CCXT

opened by pan-long 64

Edge Positioning
Summary

Currently FT analyzes pairs in whitelist randomly (or by order in config.json) and takes a position as soon as it finds a buy signal with a fixed stake_amount. a better approach would be:

Calculating the chance of winning for each pair and sorting whitelist accordingly.

Adjusting the position size according to risk/reward.

This PR addresses the above issues by adding a package called "Edge". Edge will run a process every X minutes (default 30 min) using the mainloop to go through all pairs in order to:

Calculate WinRate and Expected Risk Reward ratio for a range of stoplosses.

Finding the best WinRate/Expected Risk Reward ratio for a pair. finding the best stoploss value.

Forcing the stoploss value (calculated above) into the strategy.

The process uses the backtesting download script to fetch historical data. and cache a matrix of [pair, stoploss, win_rate, expected_risk_reward, required_risk_reward]. an example of the cached matrix:

| pair | stoploss | winrate | risk_reward_ratio | required_risk_reward | |-----|----------|---------|------------------|---------------------- | | NEO/BTC | -0.04 |0.666667 | 3.071181 | 0.500000 | |STEEM/BTC | -0.02 | 0.500000 | 2.399500 | 1.000000 | |DASH/BTC | -0.01 | 0.625000 | 1.181620 | 0.600000 | | BNB/BTC | -0.08 | 0.625000 | 1.128478 | 0.600000| | CMT/BTC | -0.05 | 0.583333 | 0.952720 | 0.714286| | XEM/BTC | -0.03 | 0.500000 | 1.194416 | 1.000000 |

WinRate is the number of winning trades divided by number of trades (e.x. if 4 trades out of 10 are profitable then the winrate is 40%).

Risk Reward Ratio is the ratio between your loss and your take profit. as an example lets say you take a position with an entry price of 100$. you put your stoploss at 90$ (i.e. 10%) and your take profit (sell point) is 120$. your risk reward would be 2 (your profit would be two times your loss).

While going through historical data we are only interested in finding two and only two things:

Amount/Number of losses if the stoploss hit.

Amount/Number of profits if sell signal is triggered.

TODO list before merging the PR:

[X] Fixing .travis script which breaks tests currently.

[X] Fixing already broken tests.

[X] Finding an alternative for py_find_1st, otherwise try to make its installation clean.

[X] Finding a proper way to sort the matrix according to WinRate and RRR.

[X] Make backslap function more readable

[X] Writing tests

[X] Writing documentation

Will replace #1147 and #1114
Enhancement
opened by mishaker 58
Avoid look-ahead in backtesting

Use buy/sell signal from previous candle, not current to avoid seeing to the future. Quick improvement to one issue mentioned in https://github.com/freqtrade/freqtrade/issues/746

I will update/fix tests only if we decide to go with this change.

opened by vertti 50

(sqlite3.OperationalError) database is locked

I was fiddling with the web ui (only viewing: reloading at the order tab: no actual trading or changes) and the 'regular' software crashed at that moment while it was trying to order something (it went through at the exchange but not at the database).

freqtrade/freqtradebot.py", line 625, in execute_buy
    Trade.session.flush()
  File "<string>", line 2, in flush
sqlalchemy/orm/session.py", line 3233, in flush
    self._flush(objects)
sqlalchemy/orm/session.py", line 3373, in _flush
    transaction.rollback(_capture_exception=True)
sqlalchemy/util/langhelpers.py", line 70, in __exit__
    compat.raise_(
sqlalchemy/util/compat.py", line 198, in raise_
    raise exception
sqlalchemy/orm/session.py", line 3369, in _flush
    transaction.commit()
sqlalchemy/orm/session.py", line 836, in commit
    trans.commit()
sqlalchemy/engine/base.py", line 2257, in commit
    self._do_commit()
sqlalchemy/engine/base.py", line 2421, in _do_commit
    self._connection_commit_impl()
sqlalchemy/engine/base.py", line 2392, in _connection_commit_impl
    self.connection._commit_impl()
sqlalchemy/engine/base.py", line 930, in _commit_impl
    self._handle_dbapi_exception(e, None, None, None, None)
sqlalchemy/engine/base.py", line 1929, in _handle_dbapi_exception
    util.raise_(
sqlalchemy/util/compat.py", line 198, in raise_
    raise exception
sqlalchemy/engine/base.py", line 928, in _commit_impl
    self.engine.dialect.do_commit(self.connection)
sqlalchemy/engine/default.py", line 634, in do_commit
    dbapi_connection.commit()
sqlalchemy.exc.OperationalError: (sqlite3.OperationalError) database is locked
(Background on this error at: http://sqlalche.me/e/14/e3q8)
freqtrade.commands.trade_commands - INFO - worker found ... calling exit

Triage Needed

opened by epigramx 47

ModuleNotFoundError: No module named 'freqtrade'

hi sorry the bot gives me an error Error : Traceback (most recent call last): File "./freqtrade/main.py", line 11, in from freqtrade import OperationalException ModuleNotFoundError: No module named 'freqtrade'
Question

opened by enea1982 47
No option to short, Limits bot to 50% potential, increased risk during bear as counter trend trading
Crypto markets experience both bull and bear periods mid / long term and up/down price movement intraday.

A lack of capability to trade short during bear periods limits strategies to trading against the trend which is hard work and carries in increased risk.

To be able to flip a position from long to short when day trading doubles to entry points opportunities.

By way of examples:

Longterm low risk trade: A simple long term trend trade on hr4 would make 150% on 1 short/bear flip from Feb to April, and 60% on the following bull.

Day trading example: Indicators flip to short at 7600, but no means to short BTC on the drop.

2) Enhancement
opened by creslinux 47

Hyperopts

An error warning comes up when trying

python3 freqtrade -c config.json hyperopt --customhyperopt HyperOpts -e 5000 --print-all

/Users/nsm/.env/lib/python3.7/site-packages/sklearn/externals/joblib/__init__.py:15: DeprecationWarning: sklearn.externals.joblib is deprecated in 0.21 and will be removed in 0.23. Please import this functionality directly from joblib, which can be installed with: pip install joblib. If this warning is raised when loading pickled models, you may need to re-serialize those models with scikit-learn 0.21+.
  warnings.warn(msg, category=DeprecationWarning)
Traceback (most recent call last):
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "freqtrade/__main__.py", line 15, in <module>
    main.main(sys.argv[1:])
  File "/Users/nsm/freqtrade/freqtrade/main.py", line 34, in main
    args.func(args)
  File "/Users/nsm/freqtrade/freqtrade/optimize/hyperopt.py", line 362, in start
    config = setup_configuration(args)
  File "/Users/nsm/freqtrade/freqtrade/optimize/hyperopt.py", line 338, in setup_configuration
    config = configuration.load_config()
  File "/Users/nsm/freqtrade/freqtrade/configuration.py", line 78, in load_config
    config = deep_merge_dicts(self._load_config_file(path), config)
  File "/Users/nsm/freqtrade/freqtrade/configuration.py", line 117, in _load_config_file
    conf = json.load(file)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/json/__init__.py", line 296, in load
    parse_constant=parse_constant, object_pairs_hook=object_pairs_hook, **kw)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/json/__init__.py", line 348, in loads
    return _default_decoder.decode(s)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/json/decoder.py", line 337, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/json/decoder.py", line 353, in raw_decode
    obj, end = self.scan_once(s, idx)
json.decoder.JSONDecodeError: Expecting ',' delimiter: line 175 column 13 (char 4125)

Question Hyperopt

opened by gitgitgat-zz 45

Web Ui Not available
Describe your environment

Operating system: windows

Python Version: 3.9-3.10 (python -V)

CCXT version: _____ (pip freeze | grep ccxt)

Freqtrade Version: freqtrade 2022.3(freqtrade -V or docker-compose run --rm freqtrade -V for Freqtrade running in docker)

Your question

So every time I start the bot nothing shows up in the web UI and yes it is enabled
Question Docker
opened by Dumnersm580 44
Buy Signal not meet criteria after updating to 2021.1 release due to pandas upgrade

Hi,

After updating bot to 2021.1 version, a strategy that worked well before , right now is not working properly, throwing buy signals that doesn't meet criteria.

After install bot began to buy pairs and this was not the normal behaviour of this strategy.

Buy signal is using qtpylib and Bollinger bands and seems that is not calculating this indicator properly.

Has been an update of this library in last release?

Thanks for your support.
Question Dependencies

opened by jirbizz 41
position mgmt (Position stacking / pyramiding)
buy or sell in several steps according to the signal.

For example as bb band strategy,

buy signal: cross up low band, buy 50% position

buy signal again: cross up middle band, buy another 50% position.

sell signal: cross down high band, sell 50% 4, sell signal again: cross down middle band, sell 50% position.

FT supports these use cases?
Enhancement
opened by yaotian 39
Missed or badly triggered computed Actions: BUY and SELL. Significant Difference Between Backtest and LIVE.
## Step 2: Describe your environment

Python Version: Python 3.6.4

Branch: Develop

Last Commit ID: b0536db (git log --format="%H" -n 1)

## Step 3: Describe the problem: Hi everyone, I decided to finaly create an issue because i'm struggling with a problem since a few weeks: There is an important difference between the signal that are computed on Backtest and the reality of actions in Live. After a few observation I can tell that this is not related to a normal lag due to the network. The behavior seems to occur when a trigger is slightly triggered. Exemple: Sell When "Close Under Open Of Previous Candle" dataframe['close'] < dataframe['open'].shift() Theoricaly Failing case 0.00005155 < 0.00005156 (Those values are too close, the condition is not fullfilled) Theoricaly Working case 0.00005155 < 0.00005160 (the gap between the two value is large enought, the condition is fullfilled)

Steps to reproduce:

Use 1 minute Time frame

launch freqtrade on live with a strategy that use directly price or averages to trigger BUY or SELL actions

Wait for a price movement that "slightly" correspond to the signal (BUY or SELL)

The signal is not triggered or is badly triggered .

When a few moment later there is a bigger price move that trigger the signal more Frankly, the signal is triggered.

Observed Results:

What happened? The signal is not triggered sometimes, when there is a small gap between compared values that are used to take the descision

What did you expect to happen? The signal should be triggered in live more or less at the same moment that is computed on the backtest (less than 1min drift tolerated on live)

Relevant code exceptions or logs:
Note: On plot-dataframe I modified the code to display only the SELLs that are consecutive to BUYs, to have more clarity.
Suspected cause

Loss of prescision due to python floating point: the problem with that theory is that I don't understand why the backtest is working as expected. The problem should be also present in that case...

What I tried to fix that issue

1- Convert all collumns related to triggers to Decimal 2- Rise Ta-lib prescision (https://github.com/mrjbq7/ta-lib/issues/151)

The issue is still present.

Thank you for your help. Regards, Axel
Bug Enhancement Good first issue Backtest
opened by Axel-CH 39
update calmar, sharpe, and sortino hyperopt losses to use latest formula
Summary

update calmar, sharpe, and sortino hyperopt losses to use latest formula

Solve the issue: #___

Quick changelog

<change log 1>

<change log 1>

What's new?
opened by stash86 1

ReinforcementLearner - raised exception ValueError. Message: Observation spaces do not match

Describe your environment

Operating system: ubuntu
Python Version: 3.9 (python -V)
CCXT version: 2.4.60 (pip freeze | grep ccxt)
Freqtrade Version: freqtrade 2022.12 (freqtrade -V or docker compose run --rm freqtrade -V for Freqtrade running in docker)

Describe the problem:

when trading with ReinforcementLearner and continual_learning=true the bot showing the following error: `freqtrade.freqai.freqai_interface - WARNING - Training ALGO/USDT raised exception ValueError. Message: Observation spaces do not match: Box([[-1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1. -1.

Steps to reproduce:

run trade in "dry run" mode trade -c user_data/config.json --strategy rl_test --freqaimodel ReinforcementLearner
use my config (nothing special)

"freqai": {
        "enabled": true,
        "purge_old_models": false,
        "save_backtest_models": true,
        "continual_learning": true,
        "train_period_days": 16,
        "backtest_period_days": 8,
        "live_retrain_hours": 0,
        "identifier": "rl_test",
        "rl_config": {
            "train_cycles": 10,
            "add_state_info": true,
            "randomize_starting_position": true,
            "max_trade_duration_candles": 600,
            "max_training_drawdown_pct": 0.4,
            "cpu_count": 8,
            "model_type": "PPO",
            "policy_type": "MlpPolicy",
            "model_reward_parameters": {
                "rr": 1,
                "profit_aim": 0.8
            }
        },
        "feature_parameters": {
                "include_timeframes": [
                "15m",
                "1h"
            ],
            "include_corr_pairlist": [
            "BTC/USDT"
            ],
            "label_period_candles": 20,
            "include_shifted_candles": 2,
            "DI_threshold": 0.9,
            "weight_factor": 0.9,
            "principal_component_analysis": false,
            "use_SVM_to_remove_outliers": true,
            "use_DBSCAN_to_remove_outliers": true,
            "indicator_periods_candles": [
                10,
                20
            ],
            "plot_feature_importances": 0
        },
        "data_split_parameters": {
            "test_size": 0.25,
            "random_state": 1235
        },
        "model_training_parameters": {}
    }

Observed Results:

seems that the problem is in the train function that executes dk.normalize_data(data_dictionary) in freqtrade/freqai/data_kitchen.py:318 when looping over data_dictionary["train_labels"].keys() the if condition to is wrong as it filters only dtype == object, but in ReinforcementLearner the target field is type int64 which causing NaN all over that df.

Triage Needed freqAI

opened by CryptoRahino 0

Improve backtest detail speed
Summary

Improve backtesting speed when using --timeframe-detail on strategies that trade rarely. this will have little effect on strategies that are constantly in positions with most pairs.

Quick changelog

only use detail timeframe when actually necessary.

Backtest
opened by xmatthias 1

ZeroDivisionError when df has no data from the date of backtest timerange

Describe your environment

Operating system: ubuntu
Python Version: 3.9 (python -V)
CCXT version: 2.4.60 (pip freeze | grep ccxt)
Freqtrade Version: freqtrade 2022.12 (freqtrade -V or docker compose run --rm freqtrade -V for Freqtrade running in docker)

Note: All issues other than enhancement requests will be closed without further comment if the above template is deleted or not filled out.

Describe the problem:

error:

2023-01-05 10:19:22,926 - freqtrade.freqai.data_kitchen - INFO - dropped 0 training points due to NaNs in populated dataset 0.
2023-01-05 10:19:22,927 - freqtrade - ERROR - Fatal exception!
Traceback (most recent call last):
  File "/opt/freqtrade/freqtrade/main.py", line 42, in main
    return_code = args['func'](args)
  File "/opt/freqtrade/freqtrade/commands/optimize_commands.py", line 58, in start_backtesting
    backtesting.start()
  File "/opt/freqtrade/freqtrade/optimize/backtesting.py", line 1336, in start
    min_date, max_date = self.backtest_one_strategy(strat, data, timerange)
  File "/opt/freqtrade/freqtrade/optimize/backtesting.py", line 1237, in backtest_one_strategy
    preprocessed = self.strategy.advise_all_indicators(data)
  File "/opt/freqtrade/freqtrade/strategy/interface.py", line 1203, in advise_all_indicators
    return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy()
  File "/opt/freqtrade/freqtrade/strategy/interface.py", line 1203, in <dictcomp>
    return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy()
  File "/opt/freqtrade/freqtrade/strategy/interface.py", line 1221, in advise_indicators
    return self.populate_indicators(dataframe, metadata)
  File "/opt/freqtrade/freqtrade/user_data/strategies/rl_test.py", line 171, in populate_indicators
    dataframe = self.freqai.start(dataframe, metadata, self)
  File "/opt/freqtrade/freqtrade/freqai/freqai_interface.py", line 157, in start
    dk = self.start_backtesting(dataframe, metadata, self.dk)
  File "/opt/freqtrade/freqtrade/freqai/freqai_interface.py", line 315, in start_backtesting
    self.model = self.train(dataframe_train, pair, dk)
  File "/opt/freqtrade/freqtrade/freqai/RL/BaseReinforcementLearningModel.py", line 101, in train
    features_filtered, labels_filtered = dk.filter_features(
  File "/opt/freqtrade/freqtrade/freqai/data_kitchen.py", line 241, in filter_features
    if (1 - len(filtered_df) / len(unfiltered_df)) > 0.1 and self.live:
ZeroDivisionError: division by zero

this issue happens when the --timerange starts before the data, as the data doesn't exists unfiltered_df is cut and stays empty what causing its len() == 0

although the issue is on the user side too, i'd expect it to cut the df first to be with the available data dates on only then tries to manipulate the df.

Steps to reproduce:

download ALGO data from binance, use timerange 20200101-20220101
the data will be recieved, starts from 20200616
run backtest on that data with freqai backtesting -c user_data/config.json --strategy rl_test --timerange 20200101-20221001 --freqaimodel ReinforcementLearner

Observed Results:

the error mentioned above of ZeroDivisionError: division by zero

Triage Needed freqAI

opened by CryptoRahino 0

Bump ccxt from 2.4.60 to 2.5.30
Bumps ccxt from 2.4.60 to 2.5.30.

Commits

47b8066 2.5.30

c45f54d Merge pull request #14206 from ttodua/precisionFromString-update

8ec5001 2.5.29

a764369 minor regex edit

8ac1438 minor-edit-clienttrait

0bd4c8e 2.5.28

f31b93d Merge pull request #16059 from sc0Vu/add-woo-edit-order

abfc243 2.5.27

4e75dc7 Merge pull request #16159 from sc0Vu/add-binance-watch-tickers

18edf5d Merge branch 'master' of github.com:ccxt/ccxt into add-binance-watch-tickers

Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

@dependabot rebase will rebase this PR

@dependabot recreate will recreate this PR, overwriting any edits that have been made to it

@dependabot merge will merge this PR after your CI passes on it

@dependabot squash and merge will squash and merge this PR after your CI passes on it

@dependabot cancel merge will cancel a previously requested merge and block automerging

@dependabot reopen will reopen this PR if it is closed

@dependabot close will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually

@dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

Dependencies Python
opened by dependabot[bot] 1
Funding fee explodes with --timeframe-detail option
Describe your environment

Operating system: linux

Python Version: 3.10

CCXT version: 1.93.66

Freqtrade Version: 2012.12

Note: All issues other than enhancement requests will be closed without further comment if the above template is deleted or not filled out.

Describe the problem:

The trade's funding fee is miscalculated in back-testing when trade adjustment is enabled and --timeframe-detail option is used.

The assumption made here: https://github.com/freqtrade/freqtrade/blob/8e5b4750d64c265c138f14f420d79c39627afa75/freqtrade/persistence/trade_model.py#L204

doesn't hold true when using --timeframe-detail because is_first is not always True here: https://github.com/freqtrade/freqtrade/blob/8e5b4750d64c265c138f14f420d79c39627afa75/freqtrade/optimize/backtesting.py#L699

This results in wrong calculation of the funding fees and produces overflow in trades with many adjustments.
Backtest Triage Needed Non-spot
opened by raphaelstar 3

Releases(2022.12)

2022.12(Dec 29, 2022)

Highlighted changes

Enhanced Producer logic, reducing memory usage while improving performance (:warning: this is a breaking change - producer and consumer must be updated together).
--timeframe-detail now properly simulates all candles at detail timeframe, significantly increasing backtesting accuracy.
New pairlist RemotePairList - allowing a pairlist to be loaded from file or from an API.
Removal of deprecated trade.enter_side (use trade.entry_side instead).
FreqUI
- Auto-refresh chart.
- Order action notifications (fill / cancel).
- Backtesting support for freqAI strategies.
FreqAI enhancements:
- Reinforcement learning.
- Improvements to Tensorboard logging.
JOSS paper has been accepted. Great job @robcaulk and everyone else involved in making this happen!

How to update

As always, you can update your bot using one of the following commands:

docker-compose

docker-compose pull
docker-compose up -d

Installation via setup script

# Deactivate venv and run 
./setup.sh --update

Plain native installation

git pull
pip install -U -r requirements.txt

Expand full changelog

8e8f71ade Version bump 2022.12
5cb8fe1a5 Add JOSS badge to freqtrade docs and readme.md
c52910f28 Improve resiliance against invalid data
4d112def1 Remove binance AD from docs page
cd4faa9c5 keep max_stake_amount through backtests
62c4675e2 Remove some deprecated fields from the API
cb66663fd show max_stake_amount in API
55001bf32 Keep max_stake_amount (only relevant for DCA orders).
6f2c3e252 Split migration and persistence tests
20901c833 Improve `purge_old_models` explanation
882e68c68 Rename backtest-result from new to "not new".
6a15a9b41 Update backtest-result_new
1cef40a13 Bump numpy from 1.23.5 to 1.24.1
63f114395 is_short should be a boolean
aaeeb8662 Bump ccxt from 2.4.27 to 2.4.60
c5b246af8 Bump isort from 5.11.3 to 5.11.4
00112d81d Bump types-requests pre-commit
9a556d263 Remove all mac conflicts
18709406c use link overwrite
9ea8792d3 Attempt brew fix
3993bd7c1 Bump types-requests from 2.28.11.5 to 2.28.11.7
e0f60e175 Bump pre-commit from 2.20.0 to 2.21.0
b1bf6d8dc Bump nbconvert from 7.2.6 to 7.2.7
ce13ce4b1 Update binance stoploss order types
524da3c7a Don't actually load models to avoid random failures
ad0d7c9a9 Don't allow DCA trades to go beyond max order size
2a7369b56 fix macos CI
73792fd6c Don't attempt to convert None to dict
70531224e Allow setting identifier via UI
07606a9e2 Simplify APi backtest config merging
6d9f1fafb allow backtest_cache to be provided via backtest API
256fac2a2 Add test for freqaimodels endpoint
5dbd5c235 Add endpoint for freqAI models
a119fbd89 Small error-message finetuning
ebf60d85d self._init_done placed wrong. fixed
43f5a1600 parse exception handling, remove info, cache change
4bad2b5c0 Apply suggestions from code review
5b9e3af27 improve wording
5405d8fa6 add discussion and tips for Base3ActionRLEnvironment
a276ef4b0 ensure long only RL is tested
86b30d2d6 Improve emc test resiliancy
0c8d657d9 update types-dateutil precommit
fa87e0807 Bump ccxt from 2.2.92 to 2.4.27
7216d140d Bump torch from 1.13.0 to 1.13.1
06225b950 Bump types-python-dateutil from 2.8.19.4 to 2.8.19.5
d86885c7f Bump blosc from 1.10.6 to 1.11.1
b61fc161b Bump isort from 5.10.1 to 5.11.3
6380c3d46 reduce duplicate code, fix cache check
bb33b96ba init cache on first iteration, init checks, limit length and charmap to info replace if invalid, move filter logic
1f4cc145c Move trade docs to advanced section
eda72ef26 Add documentation for Order object
a439488b7 Add initial trade_object documentation
bad6fe77d Remove deprecated trade property
329a0a3f4 Update docs/freqai.md
c293401b2 Add can_short to freqai base model
e60404715 Enable RL tests on arm mac
a8c9aa01f Add 3ac test
7727f3150 Add 3 Action RL env
dde363343 Add can_short param to base env
e4284f4e7 add citation to freqai doc. Update credits
36948e2a7 fix base4 env done condition
935275010 Remove some unused fixtures
32d57f624 delisted bibox following ccxt PR https://github.com/ccxt/ccxt/pull/16067
6fa3db3a1 Fix failing tests
cd1b8b9ce single space removed for the unit test to pass..
1d5c66da3 + Unit Tests
581a5296c fix docstrings to reflect new env_info changes
7b4abd5ef use a dictionary to make code more readable
7a0eadbdf Don't recalc profit on closed trades
33dce5cf1 Clarify partial exit calculation messaging
ca2a878b8 Update test naming
3af2251ce Fix add_state_info backtesting bug
2018da076 Add env_info dict to base environment
fa260e656 Move "replace or append" decision to dataprovider
dac1c8ab8 fix isort
2285ca7d2 add dp to multiproc
de19d1cfb fix doc minimal_roi
97fee3707 Improve emc test
7f3524949 - print
d52c1c755 Add unit tests
1d92db780 Change CI to actually run one 2 randomized point.
fed46d330 Revert "Bump scikit-learn from 1.1.3 to 1.2.0"
c042d0146 Don't run gc_setup during tests
0dd3836cc fix rpc method docstring
1c0c4fd42 Improve test
a693495a6 Improve external_candle aggregation
96edd3145 Test add_external_candle
414c0ce05 change unused var
6717dff19 update overlapping candle handling, move append to misc
0602479f7 minor changes, update candle appending to support overlaps
f1ebaf473 fix tests
49f6f4066 remove comment
0d5b2eed9 fix same candle handling
d376bf405 fix indefinite reconnecting
ccd1aa70a change log calls to debug, handle already received candle
c050eb8b8 add candle difference calculation to dataprovider
89338fa67 allow specifying channel send throttle
d2c8487ec update add_external_candle, fix breaking on ping error, handle empty dataframes
fce1e9d6d update analyzed df request to allow specifying a single pair
36a00e8de update add_external_candle returns
4cbb3341d change how missing candles will be handled
9660e445b use new channel apis in emc, extend analyzed df to include list of dates for candles
abc3badfb Improve shutdown behavior
5c984bf5c Temporarily downgrade blosc for arm64
6f92c58e3 add docs, add bearer token.
f940280d5 Fix tensorboard_log incrementing note
f9b7d3590 add increment param for tensorboard_log
f6b90595f remove html. change var names.
de9784267 Bump filelock from 3.8.0 to 3.8.2
915e0ac62 Bump ccxt from 2.2.67 to 2.2.92
bc2b9981d Bump python-telegram-bot from 13.14 to 13.15
2647c35f4 Bump pypa/gh-action-pypi-publish from 1.6.1 to 1.6.4
034420337 Bump sqlalchemy from 1.4.44 to 1.4.45
5a7b493d3 Bump xgboost from 1.7.1 to 1.7.2
562564801 Bump pytest-asyncio from 0.20.2 to 0.20.3
a35111e55 Bump nbconvert from 7.2.5 to 7.2.6
63d3a9ced Bump prompt-toolkit from 3.0.33 to 3.0.36
434eec733 Bump blosc from 1.10.6 to 1.11.0
78c40f053 Bump scikit-learn from 1.1.3 to 1.2.0
0fd8e214e add documentation for tensorboard_log, change how users interact with tensorboard_log
cb8fc3c8c custom info to tensorboard_metrics
8c7ec0795 ensure predict_proba follows suit. Remove all lib specific params from example config
85f22b5c3 fix bug in MultiOutput* with conv_width = 1
6b9f3f279 Fix test validation
980a5a9b5 Fix docs typo
1da8ad69d improve more tests by freezing time
bbedc4b63 Stop clock to avoid random failures on slow CI runs
3d3a7033e Improve Docker documentation wording
74e623fe5 Improve kraken test resiliance
66412bfa5 Remove unnecessary loop
7efcbbb45 Local File Loading
da2747d48 Add Local .json file Loading
7b3406914 flip add_state_info
9b4364ddc ensure that add_state_info is deactivated during backtesting
b144a6357 Remove Duplicate
547a75d9c Fix Info
607d5b2f8 Split to fetch_pairlist function, Info Message
48160f3fe Flake 8 fix, Json Fetching
199fd2d07 +Remote Pairlist
58604c747 cleanup tensorboard callback
227cdb093 Change dict update order
26a61afa1 Move base tf logic to config validation
bc48099e4 Revert changes
62c69bf2b fix custom_info
72472587d Increase test range for api version test
7c27eedda Bump API version
24edc276e Simplify new_candle message
d30a872ed Move message-silencing list next to enum
687eefa06 Improve emit_df testcase
5e533b550 Emit a simple "new candle" message to allow UI's to refresh charts
189fa6405 Add more dynamic to directory change
730fba956 Ensure base tf included in include_timeframes
e734b3992 Make model_training_parameters optional
102ab91fa Bump orjson from 3.8.2 to 3.8.3
179adea0e Bump ccxt from 2.2.36 to 2.2.67
82d4dca18 Bump mkdocs-material from 8.5.10 to 8.5.11
2eb8f9f02 Bump pypa/gh-action-pypi-publish from 1.5.1 to 1.6.1
66bb2c525 Bump fastapi from 0.87.0 to 0.88.0
caae4441e Bump jsonschema from 4.17.1 to 4.17.3
441069f36 Bump pandas from 1.5.1 to 1.5.2
16bad8dca Bump pymdown-extensions from 9.8 to 9.9
133a081a3 Bump pytest-random-order from 1.0.4 to 1.1.0
d8565261e ignore initializer type
24766928b reorganize/generalize tensorboard callback
f7b4fc5bb Update freqai-reinforcement-learning.md
38d3b4cab add details to doc plot_feature_importance doc
b2edc5808 fix flake8
d6f45a12a add multiproc fix flake8
469aa0d43 add state/action info to callbacks
075c8c23c add state/action info to callbacks
7ddf7ec0a Update freqai-parameter-table.md
05424045b Temporarily disable since binance blocks US
77dc2c92a performance improvevemnts - backtest freqai from saved predictions
2b3e166dc fix fees RL
eb81ccced add download-data command change directory fix relative config path
396e666e9 Keep old behavior of model loading
4a9982f86 Fix sb3_contrib loading issue
95651fcd5 Improve/simplify telegram exception handling
59c7ce02f binance restricted locations and server location
79821ebb3 fix flake8 errors
e7f72d52b bring back market side setting in get_state_info
17cf3c7e8 bug fixes and removed fillna from fit_live_predictions
10a45474e Update data-analysis.md
4571aedb3 consolidate and clean code
3c322bf7d Improve forceenter validation messages
8ea58ab35 change BT prediction files to feather format
b87545cd1 Bump torch from 1.12.1 to 1.13.0
5500c10f7 Improve CI file layout
2c75b5e02 Extract "live" test from regular tests
8efa8bc78 Update stable-baselines3 to 1.6.2
e891c4176 Fix typo in ipynb, too.
9cbfa1201 Directly set model_type in base RL model
05a7fca24 Fix utils docs for backtesting-analysis
9880e9ab6 Fix typo in strategy_analysis_example.md
dc03317cc Bump jsonschema from 4.17.0 to 4.17.1
d73fd4276 Fix flake8 error introduced with 6.0 update
9c28cc810 Bump cryptography from 38.0.1 to 38.0.4
348731598 Bump ccxt from 2.1.96 to 2.2.36
a46b09d40 Bump prompt-toolkit from 3.0.32 to 3.0.33
924bbad19 Bump pyarrow from 10.0.0 to 10.0.1
5aec51a16 Bump urllib3 from 1.26.12 to 1.26.13
7e75bc8fc Bump sb3-contrib from 1.6.1 to 1.6.2
49e41925b Bump flake8 from 5.0.4 to 6.0.0
f21dbbd8b Update imports of custom model
56518def4 isort
7fd6bc526 add randomize_starting_position to the rl_config
25e041b98 sneak in small change to FreqaiExampleHybridStrategy docstring and startup count
64d4a52a5 Improve the RL learning process
67d946927 small wording fix
a02da0806 Fix typo
320535a22 improve tests doc wording
a85602eb9 add "how to run tests"
5b5859238 Fix typo
fe00a6516 FIx custom reward link
cf000a4c0 Bump develop version to 2022.12-dev
e4a3efc7d Don't use strategy.stoploss too often
1a3f88c7b Replace separate start/end date option with usual timerange option
51d21b413 Fix 4ac update_total_profit
4790aaaae Implement cli options for backtesting-analysis date filtering
a26b3a9ca change sleep call back to 0.01
b52f05923 fix list to array in constants.py
be890b52f remove np import
aaaa5a5f6 add documentation for net_arch, other small changes
fcf13580f Revert "offload initial df computation to thread"
7b0a76fb7 Improve typehint
7ebc8ee16 Fix missing Optional typehint
8660ac9aa Fix import in docs
cf2f12b47 Headers between Tables -> Tables can be jumped to directly
bdfedb5fc Improve typehints / reduce warnings from mypy
81fd2e588 ensure typing, remove unsued code
8dbfd2cac improve docstring clarity about how to inherit from ReinforcementLearner, demonstrate inherittance with ReinforcementLearner_multiproc
9f13d99b9 improve parameter table, add better documentation for custom calculate_reward, add various helpful notes in docstrings etc
bd95392ee fix formatted string in warning message :)
4aa4c6f49 change sleep in channel send to 0
f268187e9 offload initial df computation to thread
afc00bc30 log warning if channel too far behind, add docstrings to message stream
391817243 Tidy up complex functions
fc59b0225 prevent ws endpoint from running without valid token
101dec461 close ws channel if can't accept
2e82e6784 move data_provider cleanup to shutdown()
73c458d47 use importlib instead of __import___
00d2a01bf isort
3a07749fc fix docstring
8855e36f5 reduce freqai testing time by reducing retrain frequency and number of features
44b042ba5 remove unused function
8f1a8c752 Add freqairl docker build process
e5fc21f57 Fix broken table rendering
3d26659d5 Fix some doc typos
48242ca02 update catch block in cancel channel tasks
d09157efb update code to use one prediction file / pair
a5442772f ensure only broadcasting to subscribed topics
c01f25ddc update code to freqai_backtest_live_models only from historic predictions
d9d7df70b fix tests, log unknown errors
d2870d48e change typing to async iterator
48a1f2418 update typing, remove unneeded try block, readd sleep
60a167bde add dynamic send timeout
fdc82af88 fix tests - update code to backtest with historic_predictions
3714d7074 smaller throttle in channel send
c1a73a551 move sleep call in send, minor cleanup
80d070e9e update code to use historic_predictions for freqai_backtest_live_models
d02da279f document the simplifications of the training environment
0cb6f71c0 better error handling, true async sending, more readable api
61a859ba4 remove tensorboard req from rl reqs
60fcd8dce fix skipped mac test, fix RL bug in add_state_info, fix use of __import__, revise doc
3903b04d3 save_live_data_backtest - added docs and tests
99bff9cbf backtesting_from_live_saved_files - code refactoring
913749c81 backtesting_from_live_saved_files - code refactoring
b01e4e3db change default value - save_live_data_backtest as false
442467e8a remove old comments and code
d713af045 remove main queue completely
659c8c237 initial revision
bf4d5b432 ensure model_type is defined
6394ef455 fix docstrings
c8d3e5771 add note that these environments are designed for short-long bots only.
c76afc255 explain how to choose environments, and how to customize them
96fafb7f5 remove limit_ram_use
b421521be help default ReinforcementLearner users by assigning the model_type automatically
90f168d1f remove more user references. cleanup dataprovider
f8f553ec1 remove references to "the user"
388ca2120 update docs, fix bug in environment
3c249ba99 add doc for data_kitchen_thread_count`
af9e40056 add test coverage, fix bug in base environment. Ensure proper fee is used.
81f800a79 switch to using FT calc_profi_pct, reverse entry/exit fees
259f87bd4 fix rl test;
e71a8b8ac add ability to integrate state info or not, and prevent state info integration during backtesting
27fa9f1f4 backtest saved dataframe from live
7adca9735 Improve python GC behavior
9c6b97c67 ensure normalization acceleration methods are employed in RL
6746868ea store dataprovider to self instead of strategy
6ff0e66dd ensure strat tests are updated
4f0f3e5b6 removed unnecessary code
8ee95db92 refactoring backtesting_fit_live_predictions function
3e57c18ac add fix_live_predictions function to backtesting
8d9988a94 enforce date column in backtesting freqai predictions files
9c5ba0732 save predictions with date and merge by date
2c1330a4e Update docs to new behavior
ded57fb30 Remove no longer valid test part
d089fdae3 Fix current-time_det calculation
5bd3e54b1 Add test for detail backtesting
0888b53b5 Udpate current_time handling for detail loop
29ba263c3 Update some test parameters
a11d579bc Verify order fills on "detail" timeframe
a2843165e fix leftovers from merge
51b410ac1 Update utils.md
8c39b3722 Update bug_report.md
35cc6aa96 Update data-analysis.md
67850d92a Update question.md
fe3d99b56 Update feature_request.md
11d6d0be9 Update sql_cheatsheet.md
abcbe7a42 Update updating.md
d42722690 Update docker_quickstart.md
8d7adfabe clean RL tests to avoid dir pollution and increase speed
3e258e000 Don't set use_db without resetting it
b9f1872d5 Install RL dependencies as dev dependency
e5204101d add tensorboard back to reqs to keep default integration working (and for docker)
488739424 fix reward inconsistency in template
017e476f4 add extras to setup.py for RL
cf10a76a2 bring back Trades.use_db = True
17fb7f7a3 gym needs 0.21 to match stable_baselines3
ab4705efd provide background and goals for RL in doc
b5dd92f85 remove RL reqs from general FAI reqs
936ca2448 separate RL install from general FAI install, update docs
8c7f47872 Update requirements-freqai.txt
292d72d59 automatically handle model_save_type for user
cf882fa84 fix tests
ab9d781b0 add reinforcement learning page to docs
6e74d4666 Ensure 1 thread available
7ef56e302 Ensure at least 1 thread is available
555cc4263 Ensure 1 thread is available (for testing purposes)
dcf6ebe27 Update BaseReinforcementLearningModel.py
83343dc2f control number of threads, update doc
099137ada remove hasattr calls
9e36b0d2e fix formatting
caa47a2f4 close subproc env on shutdown
647200e8a isort
77c360b26 improve typing, improve docstrings, ensure global tests pass
9c361f442 increase test coverage for RL and FreqAI
95121550e Remove unnecessary models, add model arg
f7dd3045f Parameterize backtesting test
f5cd8f62c Remove unused code from BaseEnv
1c56fa034 add test_models folder
7295ba0fb add test for Base4ActionEnv
f6e9753c9 show advanced users how they can customize agent indepth`
eeebb78a5 skip darwin in RL tests, remove example scripts, improve doc
7b1d409c9 fix mypy/flake8
d056d766e make tests pass
025b98dec bring back doc sentence
3b97b3d5c fix mypy error for strategy
8aac64400 add tests. add guardrails.
48140bff9 fix bug in 4ActRLEnv
27dce20b2 fix bug in Base4ActionRLEnv, improve example strats
240b52953 fix tensorboard path so that users can track all historical models
2493e0c8a Unnecessary lines in Base4, and changes for box space, to fit better for our needs (#7324)
1a8e1362a There was an error in the docs around continual learning and thread count (#7314)
67cddae75 fix tensorboard image
af8f30858 start the reinforcement learning doc
7766350c1 refactor environment inheritence tree to accommodate flexible action types/counts. fix bug in train profit handling
8c313b431 remove whitespace from Dockerfile
baa4f8e3d remove Base3ActionEnv in favor of Base4Action
cdc550da9 Revert the docker changes to be inline with the original freqtrade image
d31926efd Added Base4Action
3199eb453 reduce code for base use-case, ensure multiproc inherits custom env, add ability to limit ram use.
05ccebf9a automate eval freq in multiproc
94cfc8e63 fix multiproc callback, add continual learning to multiproc, fix totalprofit bug in env, set eval_freq automatically, improve default reward
d1bee29b1 improve default reward, fix bugs in environment
a61821e1c remove monitor log
bd870e233 fix monitor bug, set default values in case user doesnt set params
c0cee5df0 add continual retraining feature, handly mypy typing reqs, improve docstrings
b708134c1 switch multiproc thread count to rl_config definition
b26ed7dea fix generic reward, add time duration to reward
280a1dc3f add live rate, add trade duration
f9a49744e add strategy to the freqai object
a2a4bc05d Fix the state profit calculation logic
29f0e01c4 expose environment reward parameters to the user config
d88a0dbf8 add sb3_contrib models to the available agents. include sb3_contrib in requirements.
8b3a8234a fix env bug, allow example strat to short
8cd4daad0 Feat/freqai rl dev (#7)
3eb897c2f reuse callback, allow user to acces all stable_baselines3 agents via config
4b9499e32 improve nomenclature and fix short exit bug
4baa36bdc fix persist a single training environment for PPO
f95602f6b persist a single training environment.
5d4e5e69f reinforce training with state info, reinforce prediction with state info, restructure config to accommodate all parameters from any user imported model type. Set 5Act to default env on TDQN. Clean example config.
7962a1439 remove keep low profit
81b5aa66e make env keep current position when low profit
45218faeb fix coding convention
d55092ff1 Docker building update, and TDQN repair with the newer release of SB+
74e4fd063 ensure config example can work with backtesting RL
b90da46b1 improve price df handling to enable backtesting
2080ff86e 5ac base fixes in logic
16cec7dfb fix save/reload functionality for stablebaselines
0475b7cb1 remove unuse code and fix coding conventions
d60a166fb multiproc TDQN with xtra callbacks
dd382dd37 add monitor to eval env so that multiproc can save best_model
69d542d3e match config and strats to upstream freqai
e5df39e89 ensuring best_model is placed in ram and saved to disk and loaded from disk
bf7ceba95 set cpu threads in config
57c488a6f learning_rate + multicpu changes
48bb51b45 example config added
b1fc5a06c example config added
6d8e838a8 update tensorboard dependency
acf3484e8 add multiprocessing variant of ReinforcementLearningPPO
cf0731095 type fix
1c81ec601 3ac and 5ac example strategies
13cd18dc9 PPO policy change + verbose=1
926023935 make base 3ac and base 5ac environments. TDQN defaults to 3AC.
096533bcb 3ac to 5ac
718c9d044 action fix
9c78e6c26 base PPO model only customizes reward for 3AC
6048f60f1 get TDQN working with 5 action environment
d4db5c328 ensure TDQN class is properly named
91683e1dc restructure RL so that user can customize environment
ecd1f55ab add rl module
70b25461f add rl dependency
9b895500b initial commit - new dev branch
cd3fe4442 callback function and TDQN model added
01232e9a1 callback function and TDQN model added
8eeaab274 add reward function
ec813434f ReinforcementLearningModel
2f4d73eb0 Revert "ReinforcementLearningModel"
c1e7db313 ReinforcementLearningModel
05ed1b544 Working base for reinforcement learning model

Freqtrade is a free and open source crypto trading bot written in Python

Related tags

Overview

Freqtrade

Disclaimer

Supported Exchange marketplaces

Community tested

Documentation

Features

Quick start

Basic Usage

Bot commands

Telegram RPC commands

Development branches

Support

Help / Discord / Slack

Requirements

Up-to-date clock

Min hardware required

Software requirements

Comments

Summary

Describe your environment

Your question

Steps to reproduce:

Observed Results:

Relevant code exceptions or logs:

Suspected cause

What I tried to fix that issue

Summary

Quick changelog

What's new?

Describe your environment

Describe the problem:

Steps to reproduce:

Observed Results:

Summary

Quick changelog

Describe your environment

Describe the problem:

Steps to reproduce:

Observed Results:

Describe your environment

Describe the problem:

Releases(2022.12)

2022.12(Dec 29, 2022)

Highlighted changes

How to update

docker-compose

Installation via setup script

Plain native installation

2022.11(Nov 27, 2022)

Highlighted changes

How to update

docker-compose

Installation via setup script

Plain native installation

2022.10(Oct 29, 2022)

Highlighted changes

How to update

docker-compose

Installation via setup script

Plain native installation

2022.9.1(Oct 2, 2022)

2022.9.1

Highlighted changes

How to update

docker-compose

Installation via setup script

Plain native installation

2022.9(Sep 29, 2022)

Highlighted changes

How to update

docker-compose

Installation via setup script

Plain native installation

2022.8(Aug 31, 2022)

Highlighted changes

How to update

docker-compose