cryptocurruency trading bot "Sigmabot" is a project aimed to trade based on the real time data.

The purpose of this release is to demonstrate the capabilities of a frequency analysis and probabilistic approach in the field of the analysis of the price action for trading. Traditional algorythmic trading bots use indicators - Movig average(MA), Relative strength index(RSI) and templates of candlestic patterns which is a popular and simple solution for the integration. The problem of such approach is that the analysis is performed using the tools which are easy for humans and do not invovlve the techniques like Markov process and frequency characteristics of a function etc. The probabilistic research is a powerful tool but it reqires complex computations, they work very fast thus impossible for humans to use in a real time. But a computer can be programmed and it is capable of using the advantegeous techniques.

Currentlry Sigmabot has only one probability calculus method. The core principle is the definition of the function based on the function propagation. Such approach is superiour to the traditional technichniques beecause they are very demanding for the historical data and sigmabot relies upon as little as 10 hours of the market data. Which makes it very flexible. Besides the solution of the disorder problem produce a reliable result. Such algorithm can be used for the search of optimal entry point for trading for long or short positions as well as to red flag opened positions during spontaneous price hikes. MA, RSI and other technical analysis tools are incapable to produce such results.

Unlike parametric or conditional trading strategies for trading bots who use MA and RSI, Sigmabot uses CLT(central limit theorem) or Six Sigma to define the function state in a real time. Such approach is also called a price action strategy in trading.

Decision making in such method is based on the solution of disorder problem. The implementation of analysis is targeted towards bitcoin because it has the biggest market cap and there is a strong correlation between the functions of bitcoin and altcoins.

Step by step algorithm is the following:

For each direction of the price action we detect abnormalities in speed rate.
Determine  the frequency of the outlier occurrence for a given timeframe.
Based on the frequencies from upward and downward movement we produce a frequency distribution table.
Transform statistical data in a usable form by filtering conditional terms (predefined parameters) 
for the frequency distribution table, while satisfying definition consider we are at the end of a function price spike
otherwise the current level is not at the end of the price hike or no spikes has occurred for the frequency. 
We take the decision to buy, sell or stall(in case of a risk) solely by the definition of the present function.
Finally we record information for control and future analysis.

note: we can find different points of the functions by setting different parameters.

The pictures shows the moments this algorithm can find. In this particular case it searches the end of hikes for the downward trend. It works with the degree of certainty for different sets.

The image shows the order point with the green dot and sell order with the red dot. The image demonstrates the selection of a spot at the bottom of the price drop with the green dot

30402 frequency up 39  down  198  price 128.14
30403 frequency up 39  down  199  price 128.04
30404 frequency up 39  down  200  price 127.49
30404 made order at lvl  127.49
outlies   outliers up:  39   down :  200
lvl:  127.49  SL increment:  2
 n lvl: 129.49
lvl:  129.49  SL increment:  1
 n lvl: 130.49
lvl:  130.49  SL increment:  1
 n lvl: 131.49
lvl:  131.49  SL increment:  1
 n lvl: 132.49
lvl:  132.49  SL increment:  1
 n lvl: 133.49
order closed in profit
lvl:  133.49 index_ 0

This is a daraft from the console. You may see the example of the program exectution.

Sigmabot use OKEX REST API for trading https://github.com/jane-cloud/Open-API-SDK-V5 . At the moment this is the only supported trading platform. There are 3 files, two of which works togeter and the downloader is separate from them. You will need to have them both running.

This program is designed to run on the machine with python 3.8-3.10 You need to install libraries like numpy, pandas etc.

We need a historical(master) data. 'downloader.py' surves the purpose of a data collection using REST API. Records BTC, LTC, MATIC price and ofcourse timestamp. Some other data for future research. It has to be started before you run the main program. To use it for trading you need okex api key.

First you need to search for some spesified rows in a program and configure the path to to your folder.

1. Check 'downloader.py' file row 48 and put your path
2. put filename for a new '.csv' file in 'downloader.py' row 22
3. start 'downloader.py' it and give 14-20 hours to absorb enough information
4. run execute.py it will start to produce output in a console

Interested in contributing to the sigmabot project? Thanks so much for your interest! Contributions from open-source developers are greatly appreciated. My contact is [email protected]

Upon your decision the sponsorship can be reflected here.

Lean Six Sigma in Service applications and case studies edited by Sandra L. Furterer ISBN 978-1-4200-7888-6

Python for Finance Yves Hilpsch O'REILLY ISBN 978-1-491-94528-5

Sigmabot is a free and open-source software licensed under the GNU General Public License v3.0.