Alpaca AI Trading Agents Hackathon (Aug–Sep 2026)
RazorStack Trading
Multi-Agent Quantitative Paper-Trading Platform
AI-powered quantitative paper-trading platform built for Alpaca AI Trading Agents Hackathon (Aug–Sep 2026), featuring a multi-agent research desk, deterministic risk engine, and real-time paper-trade execution.
System ID
07
razorstack-trading
Architecture
M-01Multi-Agent
Research Desk
Guardrails
M-02Deterministic
Risk Engine
Execution
M-03Real-Time
Alpaca Paper API
Backend
M-04FastAPI + Next
Python + TypeScript
Section 01
Problem Statement
Autonomous LLM trading demos often place unconstrained orders without separating multi-agent research synthesis from deterministic position and risk limits.
Section 02
System Constraints
- >Enforce deterministic risk checks prior to any broker order submission
- >Coordinate multi-agent research signals with real-time Alpaca paper-trade execution
Section 03
System Architecture & Dataflow
Multi-Agent Research Desk
Deterministic Risk Engine
Alpaca Trading API
Next.js Trading Console
Section 04
Key Engineering Decisions
- 01Decoupled a multi-agent research desk from a deterministic Python risk engine so LLM hypotheses cannot bypass sizing limits.
- 02Connected FastAPI execution services directly to the Alpaca Trading API and a Next.js monitoring UI.