Functional Agents: Providing Objective Analysis and Rules

These agents perform specific, factual analysis tasks that are vital for building a complete and disciplined investment strategy. They give you solid data and enforce critical rules to keep your strategy on track.

Sentiment Analyst: The Market Mood Reader

This agent focuses on interpreting the overall “mood” of the market towards specific stocks.

• What this expert does: It acts like a market psychologist. It gathers text data (financial news headlines, social media trends) and insider trading activities over the past year. Its job is to figure out the overall “mood” of the market towards specific stocks. It looks for positive or negative signs in the news and checks if company insiders are buying (good sign) or selling (bad sign).

• How this expert works: It processes all this text and trading data, then combines the positive and negative signals. It assigns different weightings to sources (e.g., news sentiment might have a higher weight, 0.7, than insider trading activity, 0.3). It then quantifies this into an overall sentiment signal (bullish, bearish, or neutral) and a confidence score (0-100%) in its reading.

• Its report: An overall sentiment signal, a confidence score, and a structured report detailing its findings (e.g., number of bullish insider trades, positive news articles).

Risk Manager: Your Portfolio’s Safety Guardian

This agent is a crucial protector for your investments. It focuses only on measuring and limiting risk based on clear, mathematical rules, not on opinions or market mood.

• What this expert does: It’s like your personal financial safety inspector. It obtains historical price data for all relevant assets (user-selected and existing portfolio positions). It then rigorously analyzes past stock prices to understand volatility (how much prices swing) and correlation (how different assets’ prices tend to move together). Based on these objective facts, it calculates and imposes deterministic position limits. These are strict rules defining the maximum quantity or value of a specific asset you can hold.

• How this expert works: Using mathematical models, it determines that if a stock is very “jumpy” (high volatility, e.g., > 30% annualized) or often moves with other stocks you own (high correlation, e.g., average > 0.6), you should hold less of it. These limits are unbreakable rules that ensure your strategy stays within safe boundaries. This advice is mandatory—the system will not allow trades that go against these safety limits.

• Its report: Precise “remaining position limits” for each asset (how much more can be bought/sold safely), current prices, and detailed metrics on volatility and correlation. This output mandatorily influences the final investment decisions.

Portfolio Manager: The Final Decision-Maker

This agent is the ultimate authority for making simulated trading decisions. It synthesizes all inputs and rigorously adheres to risk constraints.

• What this expert does: This agent is the leader of your AI team, making the final simulated trading decisions. It takes all the advice (signals and reports) from the other AI experts and strictly follows the safety rules set by the Risk Manager. It also considers your current simulated portfolio (cash, existing long/short positions) and current stock prices.

• How this expert works: First, it deterministically calculates exactly what trades are possible for each stock (buy, sell, short, cover, hold) and the maximum quantity, based on available cash, current positions, and the Risk Manager’s limits. If no smart trade is possible, it suggests “hold.” Then, it sends this validated information (compressed signals and allowed actions) to an LLM. This LLM acts as a portfolio manager, guided by strict instructions to choose the best allowed action and quantity for each stock, and provide a brief reasoning, always staying within the allowed limits.

• Its final order: It issues the final simulated trading instructions (e.g., buy X shares, sell Y shares, or hold), the exact quantity, a confidence score (0-100%), and a clear, concise explanation for the decision.

Fundamentals Analyst: The Financial Health Checker

This agent is a general expert in fundamental analysis, focusing on a company’s financial health, profitability, and growth potential.

• What this expert does: It acts like a financial auditor. It collects a broad range of past financial reports and metrics (up to 10 periods of TTM data) like Return on Equity (ROE), profit margins, revenue and EPS growth, debt levels (Current Ratio, D/E), and cash flow (FCF conversion). It performs quantitative analyses on these to assess the company’s profitability, growth trends, financial stability, and current valuation ratios (P/E, P/B, P/S).

• How it works: It uses specific financial formulas and predefined thresholds to score different aspects of a company’s fundamentals. For example, it checks if ROE is above 15%, Net Margin above 20%, or Debt-to-Equity is below 0.5. Each check contributes to an overall fundamental health score and a corresponding signal.

• Its report: An overall signal (bullish, bearish, or neutral), a confidence score (0-100%), and detailed reasoning explaining the financial metrics that led to its conclusion (e.g., “ROE: 18%, Net Margin: 22%, D/E: 0.4. Strong profitability and conservative debt.”).

Technicals Analyst: The Market Pattern Reader

This agent is a specialized expert in technical analysis, focusing on identifying trading signals from price charts and market patterns.

• What this expert does: It acts like a chart pattern expert. It collects historical price data (open, high, low, close, volume) for a stock over a specified period. It then applies multiple advanced technical analysis strategies simultaneously. This includes identifying market trends (using Exponential Moving Averages and ADX), looking for mean reversion opportunities (using Bollinger Bands and RSI), assessing price momentum (over 1, 3, 6 months), analyzing price volatility, and detecting statistical anomalies (e.g., using Hurst Exponent for mean reversion/trending behavior).

• How it works: It runs various mathematical algorithms on the price data. For instance, it calculates EMAs to determine trend direction or Z-scores to see if a price is overextended. It then combines the signals from all these different technical strategies, giving different weightings (e.g., Trend and Momentum often have higher weights) to calculate a final composite signal and confidence.

• Its report: An overall signal (bullish, bearish, or neutral), a confidence score (0-100%), and a detailed breakdown of the signals and metrics from each technical strategy (e.g., “Strong buy signal from trend following (ADX 45), but neutral from mean reversion (RSI 52).”).

Valuation Analyst: The Comprehensive Value Assessor

This agent is a comprehensive valuation expert that assesses a company’s intrinsic worth using a combination of different financial models.

• What this expert does: It acts like a team of valuation specialists. It collects extensive financial metrics and detailed line items (up to 8 periods of TTM data) including free cash flow, net income, debt, cash, EBIT, EBITDA. It then runs four different valuation models in parallel: Owner Earnings Valuation (Buffett-inspired), an Enhanced Discounted Cash Flow (DCF) with scenario analysis, Enterprise Value/EBITDA valuation, and the Residual Income Model.

• How it works:

  • WACC & Scenarios: It first calculates the company’s Weighted Average Cost of Capital (WACC), then performs the DCF valuation under “bear,” “base,” and “bull” market scenarios to provide a range of potential values and an probability-weighted expected value.
  • Weighted Aggregation: It combines the results from all four models, giving different, predefined weights to each (e.g., DCF and Owner Earnings often get higher weights). It then calculates a “weighted gap” between the calculated intrinsic value and the current market price.

• Its report: An overall signal (bullish, bearish, or neutral), a confidence score (0-100%), and a detailed reasoning that includes the results from each valuation model, the calculated valuation gap, and a summary of the DCF scenario analysis.