Giter Club home page Giter Club logo

kuegibot's Introduction

Kuegi Bot

a simple tradingbot for BTCUSD with connectors for bybit, bitmex, binance futures and phemex. It just implements my manual tradingstrategy and therefore helps me to stick to the system. No financial advice. Don't use it if you don't understand what it does. You probably will loose money.

I only use the bybit connector myself right now. The other connectors are still there, but might have problems due to API changes in the last months which i am not aware of yet. Feel free to make a Pull Request if you find any.

ref links

if you like it and wanna start a trading account, feel free to use the ref links:

bybit

binance (saves you 10% of fees)

binance futures (saves you 10% of fees)

DigitalOcean The cloud service provider i use for the bot. If you use the reflink you also get a 100$ credit for the first 2 months to try the service risk-free.

disclaimer

right now this is a pure expert tool for my personal use. if you find the code useful thats nice, but don't expect it to be easy to understand or work with it. It also comes with no warranty whatsoever and is for educational purposes only!

roots

started with code from https://github.com/BitMEX/sample-market-maker (connector to bitmex mainly but got to change a lot)

donations

if this helps you in any way, donations are welcome:

BTC: bc1qfdm2z0xpe7lg70try8x3n3qytrjqzq5y2v6d5n

story time

i write twitter threads about my journey and the bot. you can either follow me there @mkuegi or read them here

Getting started

This section describes how to get the environment (i highly recommend virtual environments) set up to run the bot (backtest and trading). As mentioned before, this is highly experimental and use of the bot on real money is not recommended unless you really know what you are doing.

needed modules

first install the needed packages via pip. They are listed in requirements.txt and can be installed via the standard steps

execution

currently the only way to execute the bot is by running the included scripts.

bot and strategies

Currently there are specific bots for the kuegi-strategy and a simpler SFP strategy. Since those two work nicely together, i also created a MultiStrategyBot which can execute multiple strategies within one account and exchange.

Kuegi Strategy

Here i will explain my strategy once i find the time

SFP

a simple swing-failure pattern strategy. I will provide more information when the time is right.

backtest

crawling history

To run backtests you first have to collect data history from the exchange. For the known exchanges i created the history_crawler script. make sure that a folder "history" is existing in the main directory (next to the script) and run

python3 history_crawler.py bybit

if you want to crawl bitmex, you have to replace bybit with bitmex

for future calls adapt the known_history_files in utils/helper to the highest existing index of the M1 files.

The crawler loads M1 data from the exchange in max batchsize and aggregates it in a way to be easily reused from the backtest afterwards. The download can take a long time since its lots of data to fetch.

If you use another exchange or existing data, you need to make sure that the history data is saved in the same structure as the crawler does it.

running the backtest

in backtest.py you find a collection of code snippets and functions for backtesting. they are mostly commented out to be used in the python console.

run

python3 -i backtest.py

to execute the script with interactive console afterwards. there you can for

load bars from the history:

bars = load_bars(<daysInHistory>,<timeFrameInMinutes>,<barOffset>,<exchange>)

where daysInHistory and timeFrameInMinutes should be obvious. barOffset is an option to shift bars. f.e. when the default first H4 bar starts at 00:00, with this parameter you can make him start at 01:00 etc. This is pretty useful to test for stability of the bot. small changes in input (like shifting the bars) shouldn't result in big changes of the performance.

those loaded bars you can then use for backtests. create a tradingBot (a class derived from TradingBot in kuegibot/bots/trading_bot.py). I would recommend using MultiStrategyBot since thats the one i am using in production too.

There are multiple samples in the file how to create a bot and add strategies (for MultiStrategyBot).

Once the bot is created and set up, call b= BackTest(bot,bars).run() this runs the backtest and prints some performance numbers into the logs.

i mainly use the "rel:" number which is the relation between profit (per year) and the maxDD. i consider a relation of greater than 4 a good performance. but you need to decide for yourself

after the run you can call bot.create_performance_plot().show() to create a chart with more detailed performance analysis

or b.prepare_plot().show() to create a chart with the historic bar data and the positions plotted for detailed analysis.

production

disclaimer

As stated before: i DO NOT RECOMMEND to run this bot on your own money out of the box. If you understand what it does, checked the code and are confident that it does what you want, feel free to run it. I am running it on my own money, but i also know every line of the code and did extensive tests over 5 months before scaling up to serious position sizes.

so seriously: don't run it. You risk losing all your money!

settings

all settings should be placed within the settings folder.

When started via the cryptobot script, it first reads default settings from settings/default.json. If a path argument is provided, the first one should be a path to the full settings file with all data needed for the bot.

see a sample file in settings/sample.json.

The settings needs to contain one "bot" directory per exchange. The minimum required changes in the sample file are

  • the API_KEY and API_SECRET set to your key and secret.
  • KB_RISK_FACTOR set to the (average) amount of btc to be risked per trade. WARNING: depending on the strategy this might not be the actual max-risked amount but a target for the average. Also a trade might loose more than this amount because of slipage and execution problems on the exchange.
  • If you really want to run it on a live exchange, also set IS_TEST to false. Do this at your own risk!

realtime usage

if running the bot on a real exchange you need to run it 24/7. For this i recommend getting a server and set the bot up as a daemon that is restarted on any failure. The bot generally is prepared for this usage. Open Positions are stored on the disk to easy pick up after a restart.

The bot also has some security measures to prevent uncovered positions. If he finds an open position on the exchange without a matching position on file and specially without a stoploss, it closes this position. This also means that you must not trade other positions with the account of the bot. otherwise it will close them instantly and you loose money.

kuegibot's People

Contributors

kuegi avatar dek4nt avatar stewolk avatar herrhofrat avatar

Stargazers

ANG.eth avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.