Skip to content
AG Algo LabAG Algo Lab

Personal research · crypto · SOL and BTC

Reversal Engine: seeing the reversal before it shows.

A system that watches SOL and BTC on 15-minute candles, recognises reversal signatures in price action, computes a confidence score and, beyond a threshold, sends the signal to Telegram then places the order with the broker. No manual step.

2assets trackedSOL / USDC · BTC / USDT
15 mintime frameone decision per candle
AutoexecutionTelegram signal, order at the broker

The engine, running

One candle every fifteen minutes, a reversal when the structure says so.

Simulation on synthetic data: the mechanics are real, the prices and scores are not. A candle forms in front of you; when a local extremum is confirmed, the engine marks it, scores it, and only sends the signal above the threshold.

SOL / USDC · 15 minsimulation · synthetic data

How it works

From the data to the order, in three automatic steps.

Real-time ingestion

Price and order flow received continuously on SOL and BTC, over websocket, on 15-minute candles.

  • WEBSOCKET
  • OHLCV
  • ORDER FLOW

Pattern recognition

A proprietary model recognises structural reversal signatures, invisible to classic analysis, and computes a confidence score per signal before anything is sent.

  • CONFIDENTIAL ARCHITECTURE
  • CONFIDENCE SCORE

Automatic execution

When the threshold is crossed, the signal goes to Telegram with the entry price and the score, and the order is sent to the broker by API.

  • TELEGRAM
  • BROKER API
  • NO MANUAL STEP

Personal research project. It is neither a service, nor an offer, nor investment advice. Performance is not published and the model's architecture remains confidential.

What this project says about my work

Systems that run on their own, from signal to action.

The same mechanics serve the platforms I build: data that comes in, a rule that decides, an action that leaves with nobody behind the screen.