Skip to main content
Hyperoru is a production trading platform that you control through a web app and an API. This page is a friendly, non-technical tour of what happens inside — so you know exactly where your money goes, what decisions get made, and what the platform can and cannot do on your behalf. You do not need to read this page to use Hyperoru. Read it if you want to understand the machinery.

​
The big picture

At the highest level, Hyperoru is a loop that repeats continuously for every AI trader you own:
1

Collect market data

Prices, klines, open interest, funding rates, news — Hyperoru pulls them from Hyperliquid, Binance, and public news feeds and stores them in its own database.
2

Compute indicators

From raw prices Hyperoru derives higher-level signals: factor scores, market regimes, momentum and mean-reversion metrics, and so on.
3

Check your signals

Every few seconds, Hyperoru asks: does any signal your trader is subscribed to fire right now? If nothing fires, nothing happens. No API calls, no spending, no trades.
4

Run your strategy

When a signal fires, Hyperoru packages the relevant market context and runs your strategy — either by prompting your language model or by running your Python program in a sandbox.
5

Make a decision

The strategy returns a decision: buy, sell, close, or hold, along with size, leverage, and take-profit/stop-loss levels.
6

Execute the trade

Hyperoru sends the order to the exchange on your behalf using the API wallet you connected, then monitors it until it fills or is cancelled.
7

Record everything

Every decision, prompt, model response, order, and fill is saved so you can review it later in Arena and Analytics.

​
What drives each decision

When your strategy runs, it sees a rich picture of the market. This is the context Hyperoru assembles for you:

Live prices and klines

Latest ticks plus 1m, 5m, 15m, 1h, 4h, and 1d candles from the exchange you are trading on.

Position and account state

Your current equity, margin used, open positions, and recent fills.

Indicators

Moving averages, RSI, MACD, Bollinger Bands, and whatever your strategy explicitly requests.

Factor rankings

How your watchlist compares on momentum, carry, mean-reversion, and volatility.

Market regime

Breakout, trending, ranging, volatile, or quiet.

Recent news

Articles classified by relevance, sentiment, and affected symbols.
Prompt strategies receive this as natural language embedded in the trading prompt. Program strategies receive it as a Python object with typed fields.

​
What Hyperoru does NOT do

Hyperoru does not hold your funds

Your margin and open positions always live at the exchange (Hyperliquid or Binance), in your own account. Hyperoru never custodies USDC, USDT, HYPE, or any other asset.

Hyperoru cannot withdraw funds

The API keys you connect are scoped to trading only. Withdrawals, deposits, or transfers to other wallets require your manual action on the exchange.

Hyperoru does not trade without a signal

Nothing happens until a signal you configured fires. If you disable signals, the trader sits idle.

Hyperoru does not hide decisions

Every decision is logged with its full prompt and reasoning. You can audit why a trade happened at any time.

​
The data behind the decisions

Market data is not magic — it comes from publicly available sources, processed into a form strategies can consume.
KindSourceRefresh
Spot and perpetual pricesHyperliquid, Binance Futuresevery few seconds
K-lines (candles)Hyperliquid, Binance Futuresevery minute
Open interest, funding ratesExchange APIsevery few minutes
News articlesCryptoPanic, Crypto Vision, curated feedsevery 5 minutes
Factor scoresComputed internallyevery 15 minutes
Market regimeComputed internallycontinuously
Hyperoru keeps a rolling historical window so backtests and analytics can look back in time.

​
Where AI fits in

Large language models (LLMs) are used for two things:
  1. Prompt strategies — the model reads the assembled market context and returns a trading decision in JSON.
  2. HyperAI — the in-app conversational assistant that helps you build strategies, review performance, and understand what happened in a given trade.
Hyperoru itself is not a language model. It is a trading platform that uses language models when you ask it to. You bring your own API key, and you control which provider and model are used for every trader. Providers currently supported include OpenAI, Anthropic, Google (Gemini), DeepSeek, and any OpenAI-compatible endpoint (including self-hosted proxies and the new generation of “LLM gateway” aggregators).

​
What you see in the app

Dashboard

Overall equity, PnL, win rate, and recent decisions across all your traders.

AI Traders

The list of traders you own, with start/stop toggles and quick stats.

Arena

Live feed of decisions, prompts, and reasoning — the place to watch a trader in action.

Analytics

Trade attribution, factor breakdown, drawdown analysis, and backtest comparisons.

​
What you see in the API

Everything visible in the app is also reachable through the API. This enables:
  • Custom dashboards and mobile apps.
  • Integrations with Telegram or Discord bots.
  • Automated scripts that spin up traders in response to external events.
  • Exports of your trade history to spreadsheets or databases.
Start at the API reference introduction.

​
Next

Trust and safety

The guardrails around your keys, funds, and data.

FAQ

Quick answers to the questions new users ask.