Algorithmic trading system
A Python program connected live to a trading platform. It opens and manages positions on its own, around the clock, following a certain pre-programmed strategy.
The idea
Some trading strategies, to be profitably executed, require the precision of being able to open and close deals within milliseconds. No human reacts that fast, meaning a pre-programmed software program may be an amazing solution. Therefore, when a trading strategy like the one currently executed by the bot arose, the development of this side-project began.
How it works
The bot itself is written in Python, currently sitting on an Amazon (AWS) server, which is set up to wake up only when the conditions for opening a trade are met, and to fall asleep as soon as the last deal closes. The system streams live market data from the exchange, checks it continuously against the strategy conditions and places orders through the CCXT library when everything lines up. This project heavily depends on multi-threading to follow different assets simultaneously.
Reports from the best-performing months.
Where it stands
Not all months have been the same, and the ones where the strategy was being refined sometimes brought losses due to bugs. The most prominent months, like January, March and April shown in the reports above, were the best periods, where the strategy was executed flawlessly. Unfortunately, recent months suggest the strategy may be showing signs of dying out, and total profitability has started to decline. Currently the bot is still active, with new strategies being tested.
What it taught me
This project was mainly a test of not rushing the process and of testing the product before deploying. The last lesson came in terms of my own money: when real trade volumes were rushed without proper testing, real money was lost. Only in recent months have I successfully developed the patience needed not to rush updates.