wtf is signal-engine-portfolio?

Signal Engine v5 is a system built to automatically watch and trade cryptocurrency futures. It scans more than 15 crypto coins every 15 minutes, checking 16 different market signals like liquidity sweeps, price retests, and volume changes, and combines them into a single score out of 100. That score is then converted into a letter grade from A+ down to F, and the grade decides what happens next: top graded signals get traded automatically on Binance Futures, middle grades are logged for testing without real trades, and low grades are blocked entirely. After the system has closed 100 real trades, a machine learning model called LightGBM starts filtering signals further, predicting how likely each new signal is to win and rejecting anything under a 65 percent probability. The model keeps retraining itself every 50 new trades so its predictions can improve over time. The results show up on a live web dashboard built with React, updating every two seconds with open positions, profit and loss, and performance charts. A Telegram bot with more than 30 commands lets someone check signals, trades, and statistics, or switch between paper and live trading with two factor protection. The system can also write and post its own Twitter updates about strong signals, using a Groq hosted language model, with a manual approval step through Telegram before anything goes live. The backend runs on Python with FastAPI, Redis, and SQLite, and the whole thing is deployed on an AWS server using Docker containers behind an Nginx reverse proxy. It supports paid subscription tiers that unlock features like live signal timing, entry and exit price levels, and access to the underlying confluence factors. Several of the most important files, including the core scoring engine, the machine learning training code, and the alert scanner, are not included in this version and are described as available only on request, most likely for a job interview or portfolio review. The author states the entire project was built through structured prompting of an AI assistant rather than traditional hand written coding.