hackquest logo

Pond3r

Pond3r is an AI-powered perpetuals trading terminal built on GMX that enables users to create, test, and execute both single-position and portfolio-level strategies. By combining AI-driven research, personalized screeners, backtesting, and one-click execution, Pond3r removes much of the complexity and manual effort associated with perpetual trading, allowing users to move seamlessly from investment ideas to live deployment while retaining full control of their assets.

Videos

Tech Stack

Next
Web3
Solidity
Node
Python

Description

Today, many DeFi users trade perpetuals, but most lack the expertise required to design effective trading strategies or manage them efficiently. Even experienced traders face a highly manual process, requiring constant monitoring of market conditions, positions, and transactions.

While AI-powered tools can help users analyze markets and rebalance positions, they remain limited in scope and in enabling AI to execute on-chain transactions have constrained these solutions, and none currently support portfolio-level perpetual trading strategies.

Powered by GMX, Pond3r is an AI-powered Terminal that abstracts the operational difficulties of trading, lowering the "expertise barrier" by providing the interface and intelligence, GMX handles the actual execution behind the scenes

Users can build and backtest strategies, configure personalized market screeners, and execute trades both single-position and portfolio-level perpetual strategies directly from the platform, with GMX powering execution behind the scenes.

This is how it works:

The trader deposits USDC once, and Pond3r splits it across every position in the strategy in a single transaction — no manual leg-by-leg trading.

Because trading on-chain means trusting your funds, Pond3r is non-custodial by design. Your money stays in your own vault, and a keeper automatically rebalances your strategy as markets move — bound by limits you set: which markets it can trade, how much leverage it can take, and the ability to reduce risk but never withdraw a cent. You stay in control; the keeper handles the rebalancing, and Pond3r handles the operational work.

The strategy can then be refined, saved, and deployed to the vault in one click, sending all orders in a single transaction. A keeper monitors the bot and rebalances when needed—enabling long/short strategies and more complex setups.

The result is the full power of GMX delivered through a simple, intuitive interface. By abstracting away much of the complexity traditionally associated with perpetual trading, Pond3r lowers the expertise barrier required to access advanced trading strategies.

At the same time, it enables users to construct, execute, and manage coordinated portfolio-level strategies across multiple markets from a single interface—capabilities that would otherwise require significant manual effort. This allows traders to focus on their investment objectives while Pond3r handles the operational complexity of execution and monitoring.

Progress During Hackathon

During the buildathon we rebuilt the product from a small proof-of-concept into a working end-to-end system across three layers: Smart contracts. We wrote StrategyVault.sol from scratch — a single-user, upgradeable collateral vault that holds the owner's collateral and executes owner-approved GMX v2 perp orders (increase / decrease / cancel). Strategy logic stays off-chain; the vault only executes. Deployed via a beacon-proxy factory so each user gets their own vault. Solidity 0.8.26, Foundry. Agent backend. We refactored the agents from a single monolithic "strategist" into a deterministic, step-based workflow controller (built on the opencode SDK) that runs a fixed pipeline: interpret brief → select templates → select universe → select window → propose candidates → backtest & validate → decide → finalize. Backtests run on a Python compute layer (the bt library) over ingested GMX + CoinGecko data, with realistic costs. The controller enforces retry/iteration caps and always salvages a best-fit winner instead of failing. Frontend. We revamped the entire Next.js frontend: a chat copilot plus a beginner wizard, a run inspector to trace each pipeline step, and a one-click "Deploy Vault" CTA that deploys the StrategyVault on-chain.

Fundraising Status

Self funded
Team Leader
IIvan Alberquilla
Project Link
Sector
DeFiInfraAI