About the Project

A personal project aimed at debunking Telegram trading signal channels, often promotional and potentially fraudulent. Starting from the premise that even shady actors must show minimal performance or knowledge to attract followers, I built an objective analysis system to evaluate the reliability, consistency, and truthfulness of published signals.

Problem

• Explosion of dubious channels promising unrealistic gains
• Widespread opacity: rigged results, signals posted after price moves, name changes to erase history
• Lack of independent tools to assess real signal quality

Proposed Solution

A complete pipeline for ingestion → parsing → verification → scoring:

1. Collection via Telethon

• Authenticated connection to the Telegram API
• Dynamic channel tracking (handling name/ID changes)
• Real-time message capture

2. Intelligent Parsing

• Automatic detection of signal patterns (entry, TP, SL, leverage, etc.)
• Filtering of irrelevant messages (ads, announcements, etc.)

3. Market Verification

• Sync with real market data via Binance API
• Comparison of message timestamp vs actual movement start
• Validation of price levels (entry, TP hit, SL triggered)

4. Metrics & Scoring

| Metric | Description | |--------------------|-------------| | Win Rate | % of winning trades over 1, 3, 6 months | | Frequency | signals per day, week, month | | Confidence Index | Weighted score: signal timing, consistency, absence of retro-engineering | | Fraud Detection | Signals posted after movement, falsified results, name changes |

5. Future Evolutions

Conditional Copy-Trading:
If win rate > 70% + high confidence index over several months → automatically forward new signals to a Binance account (via REST API)

Auto-Learning Filters:
On parsing error for a static channel → send to AI → auto-generate new filter
→ 90–95% reduction in maintenance time

6. Visualization Interface

Lightweight web dashboard (JavaScript + Chart.js):

• Performance charts per channel
• Detailed trade history
• Confidence score evolution
• Real-time alerts on suspicious behavior

Impact & Learnings

• Systematic exposure: 3 out of 5 channels were faking results
• Discovery of reliable signals: 1 channel > 68% success over 4 months, with full transparency
• Solid foundation for a future selective and secure copy-trading system
• Cross-disciplinary skills: web scraping, real-time processing, quantitative finance, data visualization

This project embodies a critical, data-driven approach to online financial information — an antidote to FOMO and misinformation, with real potential to clean up the social media signal ecosystem.