Just built a trading software

Hey everyone, Hope you’re all good! So, as the title suggests, I’ve been working on a new trading software. Took me over a month of relentless coding and refinement. I aimed to discover structural alpha, but not just for one specific asset. I wanted something broader, something that could be applied to any crypto coin and I think I’ve achieved that. This alpha seems to tie into leverage and liquidation cascades, though I’m not 100% certain. One of the things I’m hoping for is that this could potentially boost market liquidity and help mitigate those chaotic liquidation events if that’s indeed the cause. At this point, I have a model that I trust for FARTCOIN. I started with data from Bybit since they lead this coin. Right now, I’m in the process of downloading Binance data to create models for assets where Binance takes the lead. If you fiddle with models for different coins, you’ll see that some strategy parameters tend to overlap quite a bit this is actually what you want if you're trading on a structural, market-wide level. The program works with gradient boosting using DART (this turned out to be the best for order book features). It looks at various features, employing normalized technical analysis just for context, making the model price-agnostic. I’ve developed microstructure and Level 2 order book features along with broader context features to help it adapt to various market conditions (inventory + context). My main focus was on making it robust enough to handle shifts in market regimes even under extreme circumstances. Currently, the model is performing well. I’ve rolled out live strategies with actual money, and the results have been consistent with the backtests. However, I haven’t fully put the latest models through their paces in live execution yet (since I’ve added more complexity).

### Disclaimer   This software is provided "as is", without warranty of any kind. Use at your own risk. Trading involves substantial   risk and you are solely responsible for any decisions and losses.

New release: https://github.com/wizrdfab/PurpleSky I added a lot of changes. New model architecture is now state of the start with deep learning + gradient boosting. Bear in mind this project was made 100% by AI plus my knowledge. You can also make your own strategies by tunning the config plus other parameters. Don't hesitate to contact me if case you need help running it!