Agentic Trading Engine

The Teaglu market engine is an agent-based framework for automated intraday trading above the High Frequency Trading (HFT) timescale.  Market analysis and response is typically in the 40ms range, with additional time required for trade execution.

The same features fed to the machine learning engine are sent to training files along with time-shifted actual return data - allowing a continuous learning loop at the weight level.  The more it trades the more it learns.

Features

• TensorFlow based machine learning models

• Generation of training files while trading

• Hundreds of indicators for machine learning features

• Fluid dynamics inspired monte-carlo risk engine

• Volatility-based risk management

• Optional reverse martingale sizing strategy

• Capital and VaR shared budget model

• Live news feed from Insight Sentry

• Live news ingestion and impact analysis via ChatGPT

• Live scanner with automated spin-up of agents

• Replay based on recorded market data OR massive.com

Availability

We use our own platform knowing that every gain and loss is our responsibility. When trading systems are sold as products, responsibility for outcomes inevitably becomes ambiguous. Rather than blur that line, we’ve chosen not to commercialize this system.