Executive Summary

Nexus Investing is an autonomous trading engine powered by 8 specialized AI agents across 5 LLM families, managing 6 trading profiles across 6 brokerage accounts—built to think, debate, and trade like a hedge fund.

Unlike traditional automated trading systems that rely on rigid algorithms or single-model approaches, Nexus combines multiple state-of-the-art language models with real-time market data to deliver adaptive, intelligence-driven trading. The platform has validated 38,284 backtests and achieved 78% CAGR with its APEX Elite profile while maintaining 0% ruin probability.

Platform Overview

Nexus Investing operates on the principle that the best trading decisions emerge from diverse, specialized intelligence working in concert. The platform's eight-agent council includes:

• Market Analyst (GPT-4o) — Processes technical indicators, price action, and chart patterns
• Fundamentals Analyst (DeepSeek R1) — Evaluates company financial health and valuation metrics
• News & Sentiment Analyst (Gemini 2.5 Pro) — Grades market catalysts and investor sentiment
• Calendar Agent (DeepSeek R1) — Tracks earnings, economic events, and market-moving dates
• Bull Researcher (GPT-4o) — Constructs the optimistic investment thesis
• Bear Researcher (GPT-4o) — Constructs the pessimistic counterargument
• Risk Manager (Claude Sonnet 4.5) — Identifies portfolio-level threats and position risks
• Trader Agent/CIO (Claude Opus 4.5) — Synthesizes all inputs and renders final decisions

Each agent uses the optimal AI model for its specific task, ensuring superior intelligence while optimizing costs through strategic model selection.

Six Trading Profiles

The platform supports six distinct trading profiles, each optimized for different investor goals and risk tolerances:

Key Innovations

Multi-Model Intelligence

Unlike systems that rely on a single AI model, Nexus assigns the best-in-class model for each cognitive function. Claude Opus 4.5 handles strategic synthesis as the Chief Investment Officer. DeepSeek R1 excels at fundamental analysis and calendar tracking. Gemini 2.5 Pro processes news with its massive context window. GPT-4o handles routine market analysis tasks. This hybrid approach delivers superior intelligence while optimizing operational costs.

Two-Lane Architecture

The platform separates high-frequency intraday trading (Fast Lane) from longer-term position management (Strategic Lane). The Fast Lane uses pure Python signal processing for sub-second decisions on time-sensitive day trades. The Strategic Lane employs the full AI council for deep analysis on positions held for weeks or years. This dual-engine design ensures optimal performance across all trading timeframes.

Backtest-Validated Strategies

The APEX Elite profile deploys three proven intraday strategies:

• VWAP Mean Reversion — Long positions based on price deviation from volume-weighted average price (1.275 profit factor, 55% win rate)
• Lunch Range Reversion — Captures mean reversion during the midday lull (1.637 profit factor, 60% win rate)
• Power Hour Continuation — Momentum continuation in final trading hour (2.5+ profit factor, activated during crisis conditions)

Each strategy has been validated through thousands of backtests across multiple market regimes to ensure statistical robustness.

Systematic Discipline

The platform enforces trading discipline through code rather than willpower:

• Circuit Breakers: Automatic trading halt when daily loss limits are reached
• Regime Detection: VIX-based filters adjust strategy when volatility exceeds thresholds
• Position Limits: Maximum allocation per trade prevents overconcentration
• Loss Streak Protection: Position sizes automatically reduce after consecutive losses
• Stop Losses: Every position has predefined exit points based on technical levels

This eliminates the emotional biases that plague human traders: fear, greed, revenge trading, and the disposition effect.

Technology Stack

Nexus integrates best-in-class technologies across its architecture:

• AI Models: GPT-4o, Claude Opus 4.5, Claude Sonnet 4.5, Gemini 2.5 Pro, DeepSeek R1
• Orchestration: LangGraph for agent coordination, state management, and decision flow
• Data Sources: Financial Modeling Prep (fundamentals), Alpaca (execution), Theta Terminal (options data)
• Deployment: Docker containers, Python 3.11+, Streamlit dashboard
• Database: SQLite for trade logs, JSON for configuration

Platform Capabilities

Mission Control — Autopilot Engine

Set your risk profile, capital allocation, and trading hours. The AI council runs autonomously, scanning markets, identifying opportunities, executing trades, and managing positions. Real-time notifications keep you informed of all actions. Full audit trails provide complete transparency.

Dashboard — Portfolio Analytics

Track performance across multiple dimensions: daily P&L, win rate by strategy, risk-adjusted returns, drawdown analysis, and position-level attribution. Monitor the live status of all 8 AI agents and view their real-time analysis.

Deep Dive — AI Research Terminal

Request on-demand analysis of any stock or option. The AI council conducts comprehensive research including technical analysis, fundamental valuation, news sentiment, upcoming catalysts, and risk assessment. Receive a detailed report with actionable recommendations.

Alpha Hunter — Opportunity Scanner

Profile-adaptive scanning delivers opportunities matched to your strategy. Each scan includes confluence scores that quantify setup quality across multiple technical and fundamental factors.

Performance Tracking

Detailed analytics on every trade: entry/exit prices, holding period, P&L, win/loss status, strategy attribution, and post-trade review. Learn from both winners and losers through systematic analysis.

The Lab — System Configuration

Advanced users can customize agent behavior, modify risk parameters, adjust confidence thresholds, enable/disable circuit breakers, and configure data sources. The modular architecture supports custom strategy development.

Transparency & Auditability

Every decision made by Nexus is fully logged and auditable. The "Flight Recorder" system maintains detailed per-run logbooks that capture:

• What data each agent received
• What analysis each agent produced
• How the Bull and Bear researchers debated
• What risks the Risk Manager identified
• Why the Trader Agent reached its final conclusion
• All confidence scores and reasoning chains

This transparency enables post-trade analysis, provides regulatory audit trails, and supports continuous system improvement.

Risk Management Philosophy

Nexus employs multiple layers of protection to safeguard capital:

• Position Limits: Maximum allocation per trade prevents overconcentration
• Daily Loss Caps: Profile-specific limits (3-10% depending on aggression level)
• Regime-Based Gating: VIX thresholds determine market conditions (green/yellow/red/circuit breaker)
• Confluence Scoring: Multi-factor alignment required before entry
• Stop Loss Discipline: Automated exits based on technical levels or ATR
• Correlation Monitoring: Portfolio-level analysis prevents excessive correlation exposure

The Risk Manager agent reviews every proposed trade before execution and can veto high-risk opportunities.

About This Document

This Executive Overview provides a high-level introduction to Nexus Investing. It showcases the platform's capabilities, performance results, and strategic approach without revealing proprietary implementation details.

For complete technical specifications, installation procedures, strategy parameters, agent configuration, and operational guides, subscribe to access the full 300+ page User Manual & Technical Reference.