SatSense (Mempool Intelligence)

A full-stack Bitcoin analytics platform that correlates real-time market sentiment with mempool congestion patterns using a state-of-the-art 2026-regime ML model.

Overview

SatSense is designed to predict Bitcoin fee trends by analyzing the interplay between social catalysts (news, social media) and on-chain transaction pressure. It features a high-performance FastAPI backend for ML processing and a sleek Next.js dashboard for real-time visualization.

Key Features

• 2026 Fee Market Modeling: Calibrated for the post-L2-dominance era where institutional settlement and retail Lightning usage define the fee floor.
• Sentiment-Congestion Correlation: Uses FinBERT to score real-time news feeds, combined with the Fear & Greed Index and GitHub developer momentum, to correlate market mood with mempool depth.
• Stateful Markov Simulation: A robust synthetic fallback engine that maintains realistic dashboard activity even when upstream APIs (like mempool.space) are unreachable.
• Predictive Analytics: Random Forest models trained on historical 2026 data to project fees for 1, 3, and 6 block confirmation targets.

Architecture

• Frontend: Next.js 14, TailwindCSS, Recharts, Framer Motion.
• Backend: FastAPI (Python), FinBERT (Transformers), Scikit-Learn.
• Database: Supabase (PostgreSQL) for persistence and time-series snapshots.
• Automation: Internal background scheduler for 5-minute periodic snapshots and model retraining.

Getting Started

1. Prerequisites

• Python 3.10+
• Node.js 18+
• Supabase Account

2. Backend Setup (ML)

cd ml
pip install -r requirements.txt
# Configure .env with SUPABASE_URL and SUPABASE_SERVICE_KEY
python -m uvicorn main:app --host 0.0.0.0 --port 8000

3. Frontend Setup

cd frontend
npm install
# Configure .env with NEXT_PUBLIC_SUPABASE_URL and NEXT_PUBLIC_SUPABASE_ANON_KEY
npm run dev -- -p 3001

ML Logic & 2026 Regime

The system operates on a Three-Layer Fee Model calibrated for April 2026:

1. Filler (1-3 sat/vB): Baseline demand from Ordinals and low-priority consolidations.
2. Utility (4-15 sat/vB): Daily retail activity, exchange withdrawals, and L2 channel funding.
3. Priority (15-50 sat/vB): Institutional settlement batches and L2 rollup commits.

Real-Time Adaptivity

If external APIs block requests, the engine automatically switches to a Stateful Simulation that models block mining events every 10 minutes (V-shape mempool patterns) and maintains momentum-based fee transitions to ensure the dashboard remains realistic and alive.

Database Schema

• mempool_snapshots: Raw on-chain metrics (tx count, size, median fee).
• sentiment_snapshots: Scored catalysts and article volume.
• features: Integrated feature sets used for ML training.
• predictions: Model outputs for future block targets.
• actuals: Ground truth data used to measure model accuracy.

---

Built for the 2026 Bitcoin Economy.