AlphaLabs

Back Testing Battle
Backtest Arena
Forward Test Arena
Dashboard
Landing Page
Landing Page

Inspiration

From Chaos to Code: ALPHALABS
The Problem That Kept Me Up at Night
I wanted to build a platform where anyone could create AI trading agents, backtest strategies, and prove their performance.
The challenge: coordinating AI models, real-time market data, WebSocket streams, and financial calculations.
Discovering Kiro’s specs feature changed everything.

The Inspiration: nof1.ai Meets Algorithmic Trading
Traditional algorithmic trading relies on rigid rules tied to a single trading point. You write conditions like “buy when RSI < 30 and MACD crosses above the signal line,” and the system executes exactly that.

AI trading is different. Instead of hardcoded conditions, AI can:

Analyze multiple indicators simultaneously
Consider broader market context
Adapt its reasoning to inputs
Make nuanced decisions beyond binary conditions

AlphaLabs allows users to provide tickers, indicators, and context. The AI makes decisions using holistic analysis instead of a single rule check.

What it does

AlphaLabs enables:

AI-driven contextual trading decisions
Backtesting with historical data
Real-time WebSocket-based streaming
Forward testing (planned)
Council Mode using 4–5 LLMs for consensus
Custom indicator engine with 22+ indicators
Risk-managed position handling
Certificate generation (PDF + PNG)
Comprehensive analytics

It blends AI reasoning, financial logic, and scalable architecture into a production-ready trading intelligence platform.

How we built it

Why Kiro Was My Secret Weapon

1. Specs: The Blueprint That Saved Me
The .kiro/specs directory became the backbone of the project. Instead of scattered notes, I had structured specs for:

Backend architecture and trading engine
FastAPI migration
Code quality and refactoring
Custom indicator engine

Each spec included: Requirements, Design, Tasks
This enabled consistent progress across more than 20 services, easy onboarding, and a clear development roadmap.

2. The Frankenstein Architecture
AlphaLabs combines components that rarely coexist:

AI + Trading Logic

OpenRouter integration for multiple models
JSON-formatted trading decisions (LONG, SHORT, CLOSE, HOLD)
Contextual reasoning
Retry logic, timeouts, circuit breakers
Graceful fallback to HOLD

Real-Time WebSocket Communication

Live streaming of backtests
Candle updates, reasoning logs, position changes
Concurrent sessions
Heartbeat + reconnection logic

Technical Indicator Engine

22+ indicators using pandas-ta
Two modes: Monk (RSI + MACD) and Omni (full suite)
Custom indicator engine with JSON formulas

Position Management & Risk Controls

Real-time PnL
Auto stop-loss / take-profit
Safety mode with ~2% liquidation protection
Leverage support

Certificate Generation

PDF certificates
Shareable PNG summaries
Verification codes

Council Mode

Query 4–5 models simultaneously
Aggregate decisions via voting/consensus
Compare reasoning across models

Challenges we ran into

AI Response Consistency
Solution: schema enforcement, validation, retry logic, fallback to HOLD.
Real-Time Synchronization
Solution: structured WebSocket events + session management.
Performance at Scale
Solution: precompute indicators and cache results.
Risk Management
Solution: auto SL/TP, safety mode, leverage limits.
Council Mode Coordination
Solution: parallel API calls, consensus algorithms, decision aggregation.