Twitter Alpha Sentiment Tracker V2

Real-time crypto trading signals from Smart Money Twitter accounts. Never miss a move.

The problem. Most crypto sentiment tools—LunarCrush, Santiment, and others—measure crowd sentiment. By the time retail turns bullish, Smart Money has often already moved. The signal arrives too late.

The solution. Twitter Alpha Sentiment Tracker V2 monitors the accounts that move first: on-chain analysts, whale trackers, and institutional researchers. When they post, you receive a BUY or SELL signal in your Telegram in under 2 seconds.

What you get. An open-source Python project built with FinBERT and VADER for financial NLP, using the X (Twitter) API v2. Configure your watchlist, set confidence thresholds, and run in paper mode before going live. Automatic asset tagging for BTC, ETH, SOL, and 40+ tickers. No subscription required—a free alternative to paid crypto sentiment platforms.

How It Works

Most Twitter sentiment tools read the firehose and drown in noise. This tracker watches a curated list of Smart Money accounts and fires only when they post:

[X / Twitter]  →  [Account Filter]  →  [NLP: FinBERT + VADER]  →  [BUY/SELL Signal]  →  [Telegram]

• Smart Money filter — only pre-screened accounts enter the pipeline
• FinBERT + VADER — bullish/bearish with 0.00–1.00 confidence
• Configurable threshold — set minimum score (e.g. 0.75) before firing
• Telegram delivery — formatted alerts in under 2 seconds

Live Signal Output Sample

What lands in your Telegram when a high-confidence signal fires:

 BUY SIGNAL DETECTED
 Asset:       $ETH
 Source:      @smart_money_account
 Tweet score: 0.91 Bullish (FinBERT+VADER)
 Reach:       847K followers | Tier 1

 Raw: "Accumulating ETH below $2,400. Load the boat."

 SELL SIGNAL DETECTED
 Asset:       $SOL
 Source:      @on_chain_analyst
 Tweet score: 0.87 Bearish (VADER + FinBERT)
 Reach:       312K followers | Tier 2

 Raw: "Reducing exposure here. Distribution zone."

How This Compares to Everything Else

Crowd-sentiment tools measure after Smart Money has moved. This tracker monitors the accounts that move first.

Key Features

• Smart Money Account Filter — only pre-screened accounts; no crowd noise
• FinBERT + VADER Hybrid — financial NLP with consensus scoring
• Asset Auto-Tagging — detects $BTC, $ETH, $SOL and 40+ tickers
• Tiered Weighting — Tier 1 vs Tier 2 account alerts
• Paper Mode — test without sending live alerts before connecting to Telegram
• Secrets in .env — store tokens in .env (copy from .env.example)

Installation

git clone https://github.com/Rezzecup/twitter-alpha-sentiment-tracker-v2.git
cd twitter-alpha-sentiment-tracker-v2
pip install -r requirements.txt

Set up credentials (store all tokens in .env):

cp .env.example .env
# Edit .env: TWITTER_BEARER_TOKEN, HF_TOKEN, TELEGRAM_BOT_TOKEN, TELEGRAM_CHAT_ID

Configure and run:

cp config/config.example.yaml config/config.yaml
python scripts/download_models.py
python main.py --demo          # Test on sample tweets
python main.py --mode paper    # Live monitoring, log only
python main.py --mode live     # Live + Telegram

Configuration

Store all secrets in .env. See config/config.example.yaml for pipeline settings such as the confidence threshold and watchlist paths.

Roadmap

• v0.1–0.6 Shipped: scraper, NLP, Telegram, asset tagging, paper mode
• v0.7–0.8 Active: multi-account consensus, backtesting
• v0.9+ Planned: Discord, web dashboard, exchange integration

Full Private Build

The public repo ships the engine. The private build includes the 85-account Smart Money watchlist, multi-account consensus logic, and backtesting module. Contact @Rezzecup on GitHub for details.

Risk Disclaimer

This is beta software and does not constitute financial advice. Sentiment signals carry inherent noise. Use paper mode first. Trading involves risk of loss. Past performance does not guarantee future results.