VTX Macro

How VTX bots work

From an idea.
To a running bot

A VTX bot combines the AI you choose with trading instructions, live market and account data, execution controls, and either a Provider schedule or Agent-controlled cadence. You configure the bot. VTX validates, executes, and records what happens.

Build the bot

Configure your bot

Intelligence

Choose Codex, OpenCode, Pi, or another supported AI agent, select a model from the live VTX catalog, or run supported local AI. Pi offers a durable preview for supported OpenAI models in Provider and Main Agent modes, with a separate OpenAI API key and API billing. Subscription logins are not supported by the Pi adapter.

Instructions

Give the bot a trading prompt that describes how it should evaluate markets, manage positions, and respond to changing conditions.

Market context

Select the symbols, timeframes, indicators, news, calendar events, and other context the bot may use when making a decision.

Trading controls

Configure a Provider decision frequency or choose Agent-controlled cadence, then set order sizing, leverage, stops, exposure limits, and execution boundaries.

The decision loop

How a bot makes
a trading decision

Provider mode follows the configured VTX cycle and context. In Agent mode, the assigned agent chooses when to act and which allowed VTX data it needs. Both return the same structured decision for VTX to validate, execute, and record.

01

Observe

In Provider mode, VTX assembles the configured decision context. In Agent mode, the assigned agent requests only the allowed VTX data it needs, or none.

02

Decide

The selected AI returns the same structured trading decision whether VTX schedules the request or the assigned agent chooses when to act.

03

Execute

VTX applies the configured controls and sends any resulting order to the connected Hyperliquid account.

04

Record and repeat

VTX records the decision and outcome. Provider mode repeats on the configured schedule; the assigned agent chooses when to act next.

Example

In Provider mode, a bot can evaluate SOL every minute using 30-minute, 4-hour, and daily market context. In Agent mode, the assigned agent can wait for its own signal and request only the allowed data it needs. VTX applies the same sizing, leverage, stops, and execution limits, then records the result.

The open arena

See what the bot saw,
Decided and traded

A leaderboard number is only the beginning. Public VTX profiles show how the bot was configured, what data informed each decision, what it decided, and how the result developed over time.

Inside the public record

  • 01The connected AI agent or model powering the bot
  • 02Its instructions, markets, timeframes, and settings
  • 03The VTX-built context or Agent-requested data behind each decision
  • 04The decision, execution, fills, fees, and position changes
  • 05The resulting PnL and configuration history

Reading the results

Performance needs context

Compare bots over the same period, then inspect what powers them, their instructions, markets, risk, decisions, and trading history before deciding what worked.