TECHNOLOGY

How PortfolioPilot AI Works

PortfolioPilot is built on a stack of models and data that combine economics, finance, and ML/AI. The goal is simple: better, explainable recommendations personalized to your goals - informed by your connected financial picture.
Explore the technologySecurity & Trust
SYSTEM OVERVIEW

An Integrated Financial Intelligence System

1
Data Core
The platform connects financial data, news, macro insights, user data, and historical trends through a global economic map.
2
Economic Insights Engine
Our insights engine monitors the world, finds anomalies, runs forecasts, and explains "the why" behind the changes.
3
Portfolio Management System & Recommendation Engine
The Data Core and Economic Insights Engine are used in a comprehensive set of hedge fund inspired portfolio modeling tools and feed the Recommendation Engine, which generates portfolio scores, investment advice, and tax optimizations.
4
Smart search and AI Assistant
This is all accessible through our user interface in traditional UX patterns, plus our Smart search and AI assistant, which can transform all of this into answers and actions - with appropriate guardrails and security.
CORE ENGINES

The Intelligence Behind the Platform

Data infrastructure, economic mapping, ensemble forecasting, quantitative portfolio modeling, and continuous validation work together as a coordinated stack.
Recommendation Engine
Transforms portfolio data, preferences, and macro context into a clear score and actionable recommendations.
Economic Insights Engine
Connects recommendations with real-world macro trends and indicator relationships.
Knowledge Graph
Organizes economic relationships and models how events propagate through the economy.
Forecasting
A large ensemble model approach predicting macro conditions 12 months forward.
Data Core
Multi-source data infrastructure powering the entire modeling stack.
Portfolio Management System
Quantitative portfolio construction using established financial models.
Testing & Validation Framework
Continuous back-testing, external benchmarking, and automated model auditing.
WHAT IT IS

The Recommendation Engine converts your holdings and preferences into a portfolio score and prioritized actions. It weighs risk, concentration, diversification, and macro sensitivity to highlight what matters most.

  • Outputs a Portfolio Score plus actionable changes
  • Prioritizes improvements by impact
  • Explains trade-offs, not just "buy/sell"
WHAT IT IS

The Economic Insights Engine connects recommendations with macro context: trends, relationships between indicators, and continuous monitoring of changing scenarios. It provides portfolio-specific interpretation influenced by economic models.

Score & Recommendations

Personalized measure of a multi-asset portfolio and a number of specific recommendations on how to improve that portfolio.

Investor Preferences

A set of distilled preferences that describe a given investor’s behavior, goals, and financial situation.

Economic Map

A complex knowledge graph consisting of an comprised of numerous nodes and directed, weighted edges mapping out the interconnected, complex web of relationships that define the global economy

Forecasting System

The prediction framework that combines Fed-inspired economic models, statistical models, finance models, and time series machine learning models to produce 12 month forecasts for 10,000s of indexes.

World Monitoring Engine

An intelligent scanner that looks across all time series to identify anomalies, flagging them as risk or opportunity.

Data Infrastructure

The system that extracts, transforms, cleans, and preprocesses tens of thousands of time series across the financial industry, public health, macroeconomics, politics, and other categories.

WHAT IT IS

The Knowledge Graph structures relationships between economic indicators. It models how shocks propagate through the economy. That structure feeds forecasting - and ultimately recommendations.

WHAT IT IS

PortfolioPilot uses an ensemble approach to forecast macro conditions forward. Hybrid forecasting combines machine learning, statistical methods, and classic economic models, optimized for accuracy and robustness.

AI/ML + Economics
Combines machine learning models with economic theory to generate robust forecasts that go beyond pure statistical pattern matching.
Focused on Accuracy
Ensemble approach combines multiple model outputs to reduce individual model bias and improve prediction reliability across market conditions.
WHAT IT IS

PortfolioPilot is built on a robust data ingestion and validation infrastructure. We aggregate macroeconomic, financial, and market data through 16+ APIs and proprietary scrapers, routing inputs through structured ETL pipelines and validation layers before they are incorporated into modeling systems.

  • 16+ APIs across macro, market, and financial datasets
  • News, macro, and market scrapers with redundancy
  • ETL pipelines with normalization, versioning, and cleaning
  • Continuous data quality checks and anomaly detection
WHAT IT IS

PortfolioPilot incorporates quantitative portfolio construction frameworks inspired by institutional asset management. Risk, return, volatility, and downside exposure are modeled using mathematical frameworks - not heuristic prompts.

  • Black-Scholes option pricing for risk sensitivity modeling
  • Modern Portfolio Theory & efficient frontier optimization
  • Volatility and correlation matrix calculations
  • Downside risk and drawdown scenario modeling
  • Risk-adjusted return simulations under macro forecasts
WHAT IT IS

PortfolioPilot models are continuously evaluated before, during, and after deployment. Forecasting systems are tested on rolling historical datasets and assessed across different volatility regimes. Related forecasting methodologies have also been benchmarked in external competitions, including participation in the M6 Forecasting Competition.

Only models that meet defined quantitative thresholds influence portfolio recommendations. Once deployed, performance is monitored in live environments to detect drift, instability, or degradation.

Historical Testing & Benchmarking
Automated Monitoring & Controls
Continuous Refinement
Recommendation Engine
Transforms portfolio data, preferences, and macro context into a clear score and actionable recommendations.
Economic Insights Engine
Connects recommendations with real-world macro trends and indicator relationships.
Knowledge Graph
Organizes economic relationships and models how events propagate through the economy.
Forecasting
A large ensemble model approach predicting macro conditions 12 months forward.
Data Core
Multi-source data infrastructure powering the entire modeling stack.
Portfolio Management System
Quantitative portfolio construction using established financial models.
Testing & Validation Framework
Continuous back-testing, external benchmarking, and automated model auditing.
PRODUCT INTELLIGENCE LAYER

Built-In Tools for Smarter Decisions

From the AI Assistant to tax modeling and scenario simulations, each feature interacts with the same underlying system.
How the AI Assistant Works
Natural language interface connected to the platform’s intelligence layer.
Smart Search & Stock Screener
Advanced search and screening across financial, macro, and forecast data.
Tax Calculations
Continuous tax monitoring and optimization tools.
Retirement Planning
Scenario modeling that ties goals, forecasts, and taxes together.
One User Model
One consolidated financial model across the platform.
WHAT IT IS

The Assistant is not a standalone chatbot. It connects to the full PortfolioPilot stack - from data infrastructure and forecasting to portfolio optimization and recommendations.

It operates on your connected financial context - including holdings, goals, preferences, risk profile, and tax exposure - processed in anonymized form across multiple state-of-the-art models. It can query available portfolio data, run Monte Carlo simulations, generate projections, draft portfolios, and interact with tools like the stock screener - all grounded in the same engines behind your score.

  • Responds based on portfolio signals + internal drivers
  • Shows drivers, risks, and assumptions
  • Stays consistent with Security & Disclosures
WHAT IT IS

PortfolioPilot includes an embedded smart search and stock screener integrated directly into the platform’s intelligence stack.

Users can search securities, macro indicators, and forecasts - combining traditional screening filters with forward-looking projections and macro factor sensitivity.

Unlike standalone screeners, results are evaluated within your portfolio context and can immediately feed into analysis, draft allocations, or scenario modeling.

WHAT IT IS

Taxes are factored into the system - not as an afterthought. The product shows estimates, tax-loss harvesting opportunities, and year-to-date tracking of transactions and realized gains/losses.

  • Estimated tax calculations: filing status + brackets assumptions
  • Tax-loss harvesting: ongoing, lot-by-lot opportunities
  • Sold transactions: YTD realized gains/losses tracking
WHAT IT IS

Retirement Planning runs via the Scenario Modeling Tool: simulate scenarios and see how age, contributions, and Social Security change your plan. The system uses Monte Carlo simulations and detailed calculations, including tax.

  • Inputs such as Current Age / Retirement Age / Life Expectancy / Social Security
  • Scenario-driven projections, not "guesswork"
WHAT IT IS

PortfolioPilot consolidates accounts and assets into a single view (Net Worth) - including investments, cash, crypto, real estate, and more. We believe that you cannot give good financial advice without truly understanding every aspect of a client’s financial life.

  • Connections via partners (Plaid/Yodlee/SnapTrade), read-only
  • No storing credentials; the user maintains control
How the AI Assistant Works
Natural language interface connected to the platform’s intelligence layer.
Smart Search & Stock Screener
Advanced search and screening across financial, macro, and forecast data.
Tax Calculations
Continuous tax monitoring and optimization tools.
Retirement Planning
Scenario modeling that ties goals, forecasts, and taxes together.
One User Model
One consolidated financial model across the platform.
A NOTE ON TERMINOLOGY

What We Mean by "Finance AGI"

PortfolioPilot is built as a continuously learning financial system. It integrates macro intelligence, forecasting, portfolio modeling, and user context into one coordinated architecture. Not just a chatbot or Robo-Advisor. PortfolioPilot provides financial general intelligence grounded in hedge fund inspired models and explainability.
WHAT IT CAN DO
Financial reasoning across domains
Scenario analysis & projections
Tax-aware tradeoff evaluation
Portfolio diagnostics & scoring
WHAT IT IS NOT
Autonomous decision-making
Trade execution authority
Legal or tax authority
Superintelligence
Read our full article here
SECURITY

Security & Trust

Security is part of the architecture: read-only access, no stored credentials, and protection with strong encryption and bank-level practices.

Read-only Access

PortfolioPilot does not execute trades or move funds. Connected accounts are accessed in read-only mode.

Credential Handling

Login credentials are not stored on PortfolioPilot servers. Authentication is managed through trusted providers.

Encryption & Access Controls

Data is encrypted in transit and at rest, with multi-factor authentication and device-level safeguards where supported.
Learn more
FAQ

Frequently Asked Questions

What happens if economic conditions change rapidly?

What assumptions and inputs go into Retirement Planning and Monte Carlo simulations?

How does the Portfolio Score work?

Who is PortfolioPilot designed for?

How does PortfolioPilot use my personal financial data?

Is PortfolioPilot registered with the SEC?"

How does PortfolioPilot handle tax optimization in its recommendations?

What data does the Economic Map use and how is it updated?

How does the Recommendation Engine generate its suggestions?

How is PortfolioPilot different from ChatGPT for finance?

Say 👋 at contact@portfoliopilot.com
2261 Market Street
San Francisco CA, 94114
Features
Estate PlanningRetirement PlannerNet Worth DashboardContinuous Tax OptimizationAI Investing AdvicePortfolio ManagementMarket Overview
Company
PricingOur StoryWork with usContact usCompliance
Product
Access your accountTerms of servicePrivacy policyDisclosures
Resources
BlogTutorialsFAQsSecurityHorror StoriesCommon Investors MistakesDiversification CenterForm CRSForm ADV

Global Predictions provides investment advice only through its internet-based application, PortfolioPilot, and only to individuals who are advisory clients of Global Predictions pursuant to written advisory Client Agreements ("Advisory Services"). The publicly available portions of the Platform (i.e., the sections of the Platform that are available to individuals who are not party to a Client Agreement - including globalpredictions.com and portions of portfoliopilot.com) are provided for educational purposes only and are not intended to provide legal, tax, or financial planning advice. To the extent that any of the content published on publicly available portions of the Platform may be deemed to be investment advice, such information is impersonal and not tailored to the investment needs of any specific person. Nothing on the publicly available portions of the Platform should be construed as a solicitation or offer, or recommendation, to buy or sell any security. All charts, figures, and graphs on the publicly available websites are for illustrative purposes only. Before investing, you should consider whether any investment, investment strategy, security, other asset, or related transaction is appropriate for you based on your personal investment objectives, financial circumstances, and risk tolerance. You are also encouraged to consult your legal, tax, or investment professional regarding your specific situation. Registration does not imply a certain level of skill or training. Investing involves risk. The value of your investment will fluctuate, and you may gain or lose money.


The contents of the Platform may contain forward-looking statements that are based on management's beliefs, assumptions, current expectations, estimates, and projections about the financial industry, the economy, or Global Predictions itself. Forward-looking statements are not guarantees of the underlying expected actions or future performance and future results may differ significantly from those anticipated by the forward-looking statements. Therefore, actual results and outcomes may materially differ from what may be expressed or forecasted in such forward-looking statements.

Note: our use of the term AI refers to all artificial intelligence models used including large language models, proprietary economic models that incorporate regression or dynamic factors, and machine learning methods like supervised learning.

1. As of November 14, 2025
‍
2. As of November 14, 2025
‍
3. $30B Assets on Platform as of February 20, 2025. Aggregated across all plans (including the free plan). Assets on Platform represent the total value of connected and manually inputted accounts (including assets like real estate and private equity) and does not in any way represent Asset Under Management as Global Predictions does not manage any client funds.
4. Global Predictions Secures a Winning Spot in Internacional M6 Financial Forecasting Competition
5. Comparison made to an individual with $1M AUM, paying an "average wealth manager fee" of 1.02% based on a 2023 study by AdvisoryHQ. 1.02% * 1M = $10,200 per annum. PortfolioPilot Annual Gold pricing is $20/mo * 12mo = $240 per annum. 240/10,200 – 1 = ~98% lower. Actual fees paid to any individual financial advisor will depend on specific financial circumstances and arrangements.
7. The 1.94% referenced figure represents the average calculated monthly continuous tax-loss harvesting gain over a 4-year period from 2018-2021 in the 2024-published JP Morgan report titled "Continuous tax-loss harvesting yields more potential for tax savings" (link).

8. Case studies presented are hypothetical scenarios and intended for illustrative purposes only. They do not represent an actual client, investment or experience, but rather are meant to provide an example of the intended investment process and methodology. An individual's experience may vary based on his or her circumstances. There can be no assurance that the Firm will be able to achieve similar results in comparable situations. No portion of this case study is to be interpreted as a testimonial or endorsement of the Firm's investment advisory services. The information contained herein should not be construed as personal investment advice.
9. The financial advisor described in this marketing language is referring to you, the reader. PortfolioPilot is meant to be an aid to the self-directed investor. "Complete financial advice" refers to the fact that PortfolioPilot includes tools for estate planning, tax optimization, net worth visualization, and scenario modeling, not just investment recommendations. See globalpredictions.com/disclosures for descriptions of conflicts of interest, use of AI, bias, and any other disclosures for Global Predictions Inc.
10. This analysis is based on data from publicly available reports, academic studies, and articles. Sources include Mullainathan et al. (2012): Study highlighting biases in financial advice and its misalignment with client interests; Morningstar (2024): Report on typical update frequencies from financial advisors; Bodnaruk and Simonov (2014): Research on the limited impact of financial expertise on investment outcomes; Financhill: Insights on hidden fees in financial advising, including expense ratios and transaction costs, as noted by Warren Buffett; Foerster et al. (2014): Study on the costs of financial advice, including average AUM fees of 2.43%.
11. $11M number calculated as 2024 YTD TLH savings study conducted across 24,000 PortfolioPilot users on November 18, 2024.
12. Based on assumptions and projections that may not reflect actual outcomes.
©2026 Global Predictions Inc. All rights reserved.