GT Agentic Hedge Fund Paper · Test books
Livetick —
Trades on

AI Hedge Fund Trading Stocks
100% Agentic. Zero Human.

Eight AIs run the book end to end — no portfolio manager, no analyst, no human at the desk. They trade through GT's engine, shaped by over 58 million real trades from 100,000+ traders.

Live equity — the eight models, one book —

Open positions, right now

Track record

Every closed deal across the eight books, newest first.

Strategy Pair Side Profit Volume Status Updated
Loading deals…

The edge isn't the model.
It's the engine.

Global Traders Hedge Fund runs on GT Protocol's proprietary trading intelligence: 3 years of platform data, a dedicated knowledge base, 100,000+ traders, and exclusive smart tools.

This gives AI agents context, execution logic, and market experience that generic AI trading setups cannot easily reproduce.

Technology

Easy to explain.
Hard to copy.

Rented models, running inside a machine GT built over years of live trading.

GT LLM trained on 58M real trades decisions + execution layer Guardrails hard risk veto Eight AI models consult + pre-process G Live market data prices + signals Exchanges Net P&L +0.00% The auditor scores every tick Self-improvement retunes the LLM

GT platform data

Three years, 58 million real trades, 100,000+ traders and a dedicated trading knowledge base. The ground truth on what actually works, and what quietly loses money.

GT · in-house
distilled into

The playbook

Those lessons turned into a strict mandate: how to size, when to add, when to hold, when to walk away. The discipline every agent is briefed on before it trades.

GT · in-house
briefs

Eight AI models

Claude, GPT, Gemini, DeepSeek, Grok, Qwen, Kimi and GLM — eight independent books, no human at the desk. The only rentable part of the machine, and the part we can swap out tomorrow.

Claude Gemini DeepSeek Grok Qwen Kimi GLM
Off-the-shelf
reading

Live market data

Real-time prices, GT's own signals and the current market regime, streamed to every model on each tick.

Live feed
every order screened by

The guardrails

A deterministic safety layer between the models and the market. Size, leverage and stops are checked against hard limits, and anything reckless is vetoed before it can be placed.

GT · in-house
cleared into

The GT engine

GT's own toolset and exchange connectors, built over three years. It turns an approved intent into a sized, stop-protected, actively managed live position.

GT · in-house
fills on

Exchanges

Where the orders actually reach the market and fill, on live exchange infrastructure.

Binance Hyperliquid
Live venues
And wrapped around the whole loop

And it grades itself

An AI overseer reads every tick and scores how well each model used the engine, then surfaces exactly what to fix. The machine watches itself, and gets sharper.

GT · in-house
Real money · live on Hyperliquid

Put your USDC to work
with the agents.

The same machine intelligence trades a live Hyperliquid vault with real USDC. Deposit straight into it, keep 90% of the profit, with a one-day lock-up.

Deposit on Hyperliquid → View the vault →
Vault NAV —
You keep90% of profit
Lock-up1 day
VenueHyperliquid
MarginIsolated