Skip to main content

​
Zonein Architecture

Zonein is a superior intelligence layer and intent-centric agent launchpad for on-chain traders. It continuously ingests orderbook activity, wallet behavior, derivatives flow, and market microstructure from Hyperliquid and Polymarket — fuses these streams with multi-timeframe technical analysis, sentiment correlation, and mispricing detection — then delivers real-time composite signals through an AI Dashboard, interactive Position Graphs, autonomous Trading Agents, and a public API that any bot or AI assistant can plug into. The architecture is designed for one thing: compressing the OODA loop from hours to milliseconds.

​
System Overview

​
From Data to Signal

The platform treats the market as a continuous stream of events across four asset types: perpetual futures, spot token holdings, HIP‑3 builder‑deployed perps, and prediction markets. Data arrives in real time from Hyperliquid’s on-chain orderbook (positions, trades, funding, HIP‑3 DEXes) and Polymarket’s API (leaderboards, positions, trades). Technical analysis indicators span multiple timeframes. Derivatives data provides open interest, funding rates, long/short ratios, and liquidation flows. Tokens must meet minimum smart money conviction thresholds before signals are computed — the system automatically filters for assets where enough high-credibility wallets have taken positions to produce a statistically meaningful consensus. The signal fusion engine combines smart money telemetry (wallet-count ratio, value ratio, behavioral confidence), multi-timeframe technical analysis (RSI, MACD, SuperTrend, ADX, Bollinger Bands, and more across multiple timeframes), and derivatives flow (funding rate extremes, OI surges, liquidation clustering, taker flow imbalance) into a single directional conviction score. The weighting between sources is adaptive — configurable per agent and per strategy — and confidence increases when multiple independent sources converge on the same direction.

​
What You Use

​
AI Dashboard

Real-time command center. Assets ranked by composite conviction — click any asset for a full analysis page with OHLC charts, SM strength overlay, trade flow bubbles, funding rate, derivatives indicators, and liquidation heatmaps. See AI Dashboard for the full detail page breakdown.

​
Position Graphs

Four interactive bipartite graphs (PM, Perp, Spot, HIP-3) render the live relationship between smart money wallets and their positions. Pre-computed server-side, rendered with D3.js. See Position Graph.

​
MCP API & Telegram

REST API + MCP stdio server for programmatic access. Telegram bot for trade notifications and HITL approval. See MCP API & Integrations.

​
Trading Agents

Autonomous agents that consume composite signals, apply custom strategy and trigger conditions, and execute on Hyperliquid via non-custodial vaults. Auto or HITL mode. See Trading Agents.

​
Data Flow

Data is ingested continuously from on-chain sources, processed into composite signals, and served through the MCP API. Live snapshots power real-time views; historical time-series enable trend analysis and backtesting. Responses are cached in-memory for sub-second latency.

​
ZoneIn’s Endgame

ZoneIn is building toward a future where signal, intent, and execution collapse into a single frictionless loop — verifiable agent reputation (ERC-8004), pay-per-request micropayments (x402), RL-based policies that learn from market feedback, expanded venue coverage, and a strategy marketplace with on-chain performance verification. For the full roadmap, see Core Components — Future Endgame and MCP API — Web3 Trust & Payments. Typical scenarios illustrate how this works in practice: browsing the AI Dashboard to find assets where smart money, TA, and derivatives flow all converge; opening a Position Graph to see which high-conviction wallets share positions in a token you’re researching; deploying a HITL trading agent that proposes trades with full evidence for you to approve from Telegram; or querying your AI assistant with “What are whales doing on BTC?” and getting a composite answer in seconds.