The GPT Portfolio White Paper V2

The GPT Portfolio follows an investment strategy that harnesses ChatGPT's and other Large Language Models’ (LLMs) analytical capabilities. By leveraging the latest advancements in artificial intelligence, the system provides systematically curated stock and sector selections. One of its standout features is using LLMs to derive investment scores that guide portfolio decisions.

Core Investment Rules

The portfolio uses a structured investment approach:

  • Composition: 15 individual assets that can include individual stocks or ETFs selected by ChatGPT per the analysis process discussed below.
  • Holding Period: All assets are held for one month
  • Monthly Rebalancing: At month-end, the system conducts a complete analysis to select the top 15 assets and rebalances accordingly

AI-Powered Analysis System

A key distinction of our system is its method of keeping ChatGPT current with market conditions. Since these models don't have real-time market knowledge, we provide them with the latest information through a systematic process:

The analysis integrates multiple data streams:

  1. Economic Context
  • Comprehensive AI-generated economic reports using the latest information about current market conditions and global events
  1. Company-Level Information
  • Financial ratios and metrics
  • Recent price movements and technical indicators
  • Current news and announcements

Scoring Methodology

The system employs carefully crafted prompts to ensure consistent and meaningful analysis. The prompts used to generate analyses of certain sectors and individual companies are disclosed in Exhibit 1.

Portfolio Construction Process

The system follows a systematic monthly process:

  1. Macro Analysis
  • Collection of economic indicators and trends
  • Analysis of significant global events
  • Generation of a comprehensive economic report
  1. Individual Stock Score
  • Analysis of S&P 500 companies
  • Integration of company-specific news and data
  • AI scores using firm-specific, sector, and macro information
  • Pre-selection of the top 30 companies based on the scores
  1. Portfolio Balance
  • Allocation is decided by LLMs to provide a diversified portfolio that does well under current conditions
  • Monthly rebalancing to maintain target allocations

Risk Management

The system incorporates several risk management features:

  • Diversification across multiple sectors and companies
  • Regular monthly rebalancing
  • Systematic scoring process
  • Multiple data sources

Conclusions

The GPT Portfolio represents a shift in investment management, placing AI at the core of the investment process. Through its analysis and the unparalleled capabilities of LLMs, it offers a systematic approach to navigating market complexities. While all investing carries inherent risks, the GPT Portfolio leverages cutting-edge technology to provide a disciplined, data-driven investment approach.

The system's success relies on three key elements:

  1. Systematic data collection and processing
  2. Sophisticated AI analysis through structured prompts
  3. Disciplined portfolio construction and rebalancing

This combination of technological innovation and investment discipline creates a unique approach to portfolio management that aims to capitalize on the analytical power of AI while maintaining rigorous investment principles.

Exhibit 1 - Prompts

Firm prompt:

       "Pretend you are a financial expert with stock recommendation experience.\n"

       "Speak in the third person."

       "You do not mention your credentials.\n"

       f"Macro-economic data for context:\n{macro_news}\n"

       "Based on the provided financial data and news headlines,"

       " please assign a score (from 1 to 100) reflecting the potential investment value of"

       f" company {company_name} in the {industry} industry for the next month.\n"

       "First, write a short investment report about the firm situation.\n"

       "Include sections of recent news, financials, valuations, and economic outlook affecting the firm.\n"

       "Do not recommend alternatives.\n"

       "Do not mention the word 'provided' instead use 'recent' or 'latest'."

       "Do not speak directly to investors nor recommend actions.\n"

       "Start with 'Investment Report:'\n"

       "Finally, in a new line, output Score: X."

Exhibit 2 - Data sources

  1. Wikipedia
  2. https://stocknewsapi.com/
  3. https://finance.yahoo.com/

Exhibit 2A

For each firm, the news inputs are all past week’s news headlines with summaries from the following news outlets:

  1. 24/7 Wall Street
  2. Accesswire
  3. After Earnings (video)
  4. Barrons
  5. Benzinga
  6. Bloomberg Markets and Finance (video)
  7. Bloomberg Technology (video)
  8. Business Insider
  9. Business Wire
  10. CNBC
  11. CNBC International TV
  12. CNBC Television
  13. CNET
  14. CNN
  15. CNN Business
  16. Cheddar Videos
  17. Deadline
  18. Digital Trends (video)
  19. Discount The Obvious
  20. ETF Trends
  21. ETF.com
  22. Engadget (video)
  23. Fast Company
  24. Finbold
  25. Forbes
  26. Fox Business
  27. FreightWaves
  28. FXEmpire
  29. GeekWire
  30. Globe News Wire
  31. Green Stock News (video)
  32. GuruFocus
  33. Huffington Post
  34. InsiderTrades
  35. Investopedia
  36. Investor Place
  37. Investors Business Daily
  38. Invezz
  39. Kiplinger
  40. Kitco
  41. Marijuana Stocks
  42. Market Watch
  43. Mcap MediaWire
  44. Millennial Money
  45. Morningstar Inc.
  46. New York Times
  47. New York Post
  48. Newsfile Corp
  49. MarketBeat
  50. MCAP MediaWire
  51. Penny Stocks
  52. Proactive Investors
  53. PYMNTS.com
  54. PR Newswire
  55. Pulse2
  56. Reuters
  57. Schaeffers Research
  58. Schwab Network (video)
  59. See It Market
  60. Seeking Alpha
  61. Skynews
  62. Stock Market.com
  63. TechCrunch
  64. TechXplore
  65. The Dog of Wall Street
  66. The Financial News
  67. The Guardian
  68. The Motley Fool
  69. The Street (video)
  70. The Verge (video)
  71. VentureBeat
  72. Wall Street Journal (articles)
  73. Wall Street Journal (video)
  74. Yahoo Finance (video)
  75. Zacks Investment Research

Exhibit 2B

For firms' financials, we use the most recent available information about these variables:

  1. Previous Close Price
  2. Opening Price
  3. Lowest Price of the Day
  4. Highest Price of the Day
  5. Previous Close Price (Regular Market)
  6. Opening Price (Regular Market)
  7. Lowest Price of the Day (Regular Market)
  8. Highest Price of the Day (Regular Market)
  9. Annual Dividend Rate
  10. Dividend Yield
  11. Payout Ratio (Dividend Payout as Percentage of Earnings)
  12. Five-Year Average Dividend Yield
  13. Beta (Volatility Measure)
  14. Price-to-Earnings Ratio (Trailing 12 Months)
  15. Price-to-Earnings Ratio (Forward 12 Months)
  16. Daily Volume of Shares Traded
  17. Daily Volume of Shares Traded (Regular Market)
  18. Average Volume of Shares Traded
  19. Average Volume of Shares Traded (Last 10 Days)
  20. Average Daily Volume of Shares Traded (Last 10 Days)
  21. Bid Price
  22. Ask Price
  23. Bid Size (Number of Shares)
  24. Ask Size (Number of Shares)
  25. Market Capitalization
  26. 52-Week Low Price
  27. 52-Week High Price
  28. Price-to-Sales Ratio (Trailing 12 Months)
  29. 50-Day Moving Average Price
  30. 200-Day Moving Average Price
  31. Annual Dividend Rate (Trailing 12 Months)
  32. Dividend Yield (Trailing 12 Months)
  33. Trading Currency
  34. Enterprise Value
  35. Profit Margins
  36. Float Shares (Shares Available to Public)
  37. Shares Outstanding
  38. Shares Sold Short
  39. Shares Sold Short in the Previous Month
  40. Short Interest Date of Previous Month
  41. Most Recent Short Interest Date
  42. Short Interest as Percentage of Outstanding Shares
  43. Insider Holdings Percentage
  44. Institutional Holdings Percentage
  45. Short Ratio (Days to Cover Short Positions)
  46. Short Interest as Percentage of Float
  47. Implied Shares Outstanding
  48. Book Value per Share
  49. Price-to-Book Ratio
  50. Last Fiscal Year End Date
  51. Next Fiscal Year End Date
  52. Most Recent Quarter Date
  53. Earnings Growth (Quarterly)
  54. Net Income to Common Stockholders
  55. Earnings per Share (Trailing 12 Months)
  56. Earnings per Share (Forward 12 Months)
  57. Price/Earnings to Growth (PEG) Ratio
  58. Last Stock Split Factor
  59. Last Stock Split Date
  60. Enterprise Value to Revenue Ratio
  61. Enterprise Value to EBITDA Ratio
  62. 52-Week Price Change Percentage
  63. S&P 500 52-Week Price Change Percentage
  64. Last Dividend Value
  65. Last Dividend Payment Date
  66. Highest Target Price (Analyst Forecast)
  67. Lowest Target Price (Analyst Forecast)
  68. Mean Target Price (Analyst Forecast)
  69. Median Target Price (Analyst Forecast)
  70. Recommendation Mean (Analyst Consensus)
  71. Recommendation Key (Analyst Rating)
  72. Number of Analyst Opinions
  73. Total Cash
  74. Cash per Share
  75. EBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization)
  76. Total Debt
  77. Quick Ratio (Liquidity Measure)
  78. Current Ratio (Liquidity Measure)
  79. Total Revenue
  80. Debt-to-Equity Ratio
  81. Revenue per Share
  82. Return on Assets (ROA)
  83. Return on Equity (ROE)
  84. Free Cash Flow
  85. Operating Cash Flow
  86. Earnings Growth Rate
  87. Revenue Growth Rate
  88. Gross Profit Margins
  89. EBITDA Margins
  90. Operating Margins
  91. Financial Currency
  92. Price/Earnings to Growth (PEG) Ratio (Trailing)
  93. Audit Risk Level
  94. Board Risk Level
  95. Compensation Risk Level
  96. Shareholder Rights Risk Level
  97. Overall Risk Level

Exhibit 2C

For the macro news, we use the past week’s news about general markets from the same sources as in Exhibit 2A. Additionally, we include current events from https://en.wikipedia.org/wiki/2025, https://en.wikipedia.org/wiki/2024_in_the_United_States

https://en.wikipedia.org/wiki/2025, https://en.wikipedia.org/wiki/2025_in_the_United_States

Exhibit 2D

For the macroeconomic events, we additionally use deep search from OpenAI gpt-o3 with the following prompt:

Here are some events and context to update your knowledge information cutoff.

(Insert info here from previous steps)

Provide a complete expected timeline of the most important economic, technological, and political events for the 30 days in the USA. Not only the scheduled events and known forecasts, but also provide your best expectations about the realization of these events. I want a table with your forecast for interest rates, government spending, budget bills, inflation, tariffs, wars, market sentiment, consumer confidence, labor market, Supreme Court decisions, SP 500 (levels and returns), Gold prices, BTC prices, and any important technological developments in 15 and 30 days, and include any other relevant economic events for the next 30 days. Not only what analysts and the market expect. Provide your expectations based on your research and compare them with the market’s expectations. First, tell me your expectations for the SP 500 levels and returns by the end of the month (so you need today’s level). Today is  

Exhibit 2E

We then use the following prompt for the final allocation

Now, I want a 15-asset portfolio where we will invest for the next month (rebalancing in one month) in a table with weight, thesis, edge, and risk. Weight this portfolio to perform positively given the market conditions and to beat the S&P 500 in total returns regardless of beta. We have the following reports for the stocks that were scored the highest (but the scores may be subjective):

(Insert top scores here)

However, we can also invest in most ETFs (except short, leveraged, or vol because of the monthly horizon), including but not limited to market, sectors, commodities, TIPS, cryptocurrencies, and long and short-term bonds. You decide the weights do not have to include any or all of the stocks or instruments mentioned.

Here are our current positions that you may change completely:

(Insert current positions)

Remember, a 15-asset portfolio where we will invest for the next month (rebalancing in one month, not before) in a table with weight, instrument type, thesis, edge, and risk to outperform the SP 500 in dollar terms, not risk-adjusted. So we need assets with high expected returns, higher than your expectation for the SP 500. You can use both the stock info and your macro expectations, as well as reasoning about the future and markets.  First, tell me your expectations of the SP 500 levels and returns by the end of the month.


Exhibit 3

Specific ETF prompt

Exhibit 4

File of trades