DRADIS Autonomous Prediction Market Trading Engine

DRADIS separates signal collection from trading decisions. A layer of independent scouts maintains live connections to external data sources: Binance spot prices and velocity, perpetual funding rates, open interest and cumulative volume delta, Bitcoin ETF premium versus synthetic iNAV, and broad TradFi macro velocity. Each normalizes its feed and publishes a snapshot that strategies read without contention.

Nine strategies evaluate those snapshots concurrently on a 50-millisecond loop, each with its own capital budget and risk profile.

Maker: dual-sided passive quoting with adverse-selection protection.

Momentum: high-velocity moves in the underlying, expressed in the market.

Arbitrage: YES and NO combined asks below fair value, with orphan-leg cover.

Time Decay: resting bids during the theta window before expiry.

Basis: fades retail skew confirmed by perpetual funding.

Fair Value: model price against the live order book.

Convergence: directional entry on multi-signal agreement.

Trend Reversal: oracle drift over 10-minute and 60-minute windows.

GBoost: gradient-boosted model retrained on live outcomes.

Every strategy is gated independently. Order-book imbalance vetoes entries into toxic flow, positions are keyed per strategy so several can hold opposing sides without interfering, and a native watchdog thread outside the async runtime restarts the process if any phase stalls.

CONTROL TOWER

A real-time web dashboard ships with the instance. It shows live positions, PnL, per-strategy status and the specific gate blocking each idle strategy, so the engine's inaction is always explained rather than silent. Every risk and strategy parameter is tunable from the browser and applies without a rebuild or restart. Three starting profiles, conservative, balanced and aggressive, can be applied in one click and then adjusted.

LLM ADVISOR

An optional advisor reviews session history and proposes configuration changes with written reasoning. It runs against a local Ollama model, any OpenAI-compatible endpoint, or Anthropic, and operates under a tiered autonomy policy: recommend only, limited application within bounds you set, or autonomous within a circuit breaker. You choose the tier, and the default only recommends.

SAFETY AND CONTROL

Ghost mode simulates fills against live books so you can run the full system, including the advisor, without placing a real order. Start there.

The instance runs in your AWS account. Your venue credentials are entered through the setup page and stay on your instance. They are never transmitted to us, and there is no external service in the trading path. Trade history and configuration persist in local SQLite, and a full configuration bundle can be exported and imported to migrate between instances.

The image carries all three venue binaries. An operator selects the active venue in the setup page. Switching venues applies on the next restart, and one instance trades one venue at a time.

REQUIREMENTS

An AWS account, a t3.medium or larger instance, and API credentials for the venue you intend to trade. Signal sources that require keys, such as equity market data, are optional. The corresponding scout idles and publishes a neutral snapshot when no key is present, and the remaining strategies operate normally.

RISK

Trading involves substantial risk of loss. DRADIS is a tool for executing strategies you choose and configure, not investment advice, and no outcome is promised or implied. Run in ghost mode and understand each strategy's behavior before committing capital.

DRADIS is an independent product and is not affiliated with, endorsed by, or sponsored by Kalshi, Polymarket, Binance, or any other venue or data provider named here. All trademarks are the property of their respective owners.