100% public sources

V4 Server

The data engine behind RoadTo1M. Over a year of relentless work to turn free public feeds into real-time market intelligence.

390K+
Lines of code
45+
Python Services
540+
Real-time pairs
<1s
Analysis latency
7 GB+
Data in database
24/7
Continuous uptime

The V4 Server uses no paid sources. Everything comes from free public APIs: Binance WebSocket for prices, open interest, funding, and real-time trades. CoinGecko complements broader market coverage. No Bloomberg subscription, no Reuters terminal. Just engineering discipline, code, and a long build cycle.

The result is a system capable of analyzing 540+ pairs simultaneously, computing multi-factor anomaly scores, detecting order book spoofing, and producing structured market signals on a single dedicated Hetzner server.

Architecture

4 layers that transform raw feeds into actionable intelligence.

Ingestion
WebSocket
Processing
Workers
Analysis
Services
Storage
3 databases

Tech stack

Python (FastAPI + asyncio)

High-performance async backend

PostgreSQL

7 GB+ of historical and real-time data

MongoDB

300K+ documents (signals, calls, trades)

Redis

3,200+ live keys for real-time cache

The data

Main PostgreSQL tables — all from free sources.

TableSizeDescription
anomaly_scores2.7 GBReal-time anomaly scores, 540+ pairs
ohlcv_aggregated2 GBAggregated multi-timeframe candles
ohlcv_1m600 MB1-minute candles (3-day rotation)
liquidations480 MBLive Binance Futures liquidations
big_trades480 MBLarge trades detected in real-time
futures_metrics400 MBOI, funding, volumes over 3 months

Data sources

All public. All free. Zero subscription.

Binance WebSocket
Binance REST API
CoinGecko API
Internal Calculations

The journey

2025 — Beginning

V1: The prototype

First Python script. Basic price collection via Binance REST API. A single pair. No analysis.

2025 — Evolutions

V2: WebSocket + multi-pairs

Migration to real-time WebSocket. 50 pairs. First anomaly scoring system. PostgreSQL for storage.

2025 — Maturity

V3: The radar

Introduction of the pre-pump radar. Multi-factor scoring. MongoDB for signals. Redis for cache. 100+ pairs.

2026 — Now

V4: The current server

540+ pairs, 45+ async services, spoofing detection, footprint analysis, liquidation tracking, multi-TF CVD, autonomous trading simulator. 390,000+ lines of code.

What it computes in real-time

Anomaly Scoring

Composite score 0-200 per pair, based on 8+ factors.

CVD Delta

Cumulative Volume Delta over 5min, 15min, 1h. Buy vs sell pressure.

Spoofing Detection

Orderbook analysis to detect fake walls and manipulations.

Liquidation Mapping

Liquidation zones above/below price with imbalance ratio and directional bias.

Indicators Engine

RSI, MACD, EMAs, ADX, Bollinger, ATR — computed internally, no external API.

Signal Performance

Post-detection tracking at 15m, 30m, 1h to measure reliability.

All of this, completely free.

Access RoadTo1M
← Back to home