Building an Intelligent Alpha ETF with ChatGPT | Doug Clinton

Поділитися
Вставка
  • Опубліковано 30 січ 2025

КОМЕНТАРІ • 27

  • @HepCatJack
    @HepCatJack 14 днів тому +1

    Any idea of why Copilot was excluded from these tests ? Since a lot of investing is quantitative, it would seem that it, as a coding assistant may be suited to tailor a good portfolio.

    • @ExcessReturns
      @ExcessReturns  14 днів тому +1

      Copilot is more of a productivity tool within Office than an AI model. So it wouldn't work for this use case.

    • @filippxx
      @filippxx 14 днів тому

      Copilot is a Jake of all trades. There are models got coding that eat its GPU lunch

  • @passtheguak5403
    @passtheguak5403 14 днів тому +1

    I can't wait for the AI Warren Buffet and AI Benjamin Graham Voice + Portfolios
    I can totally see this happening. People gunne be like "Why need a financial advisor when I have an AI Warren Buffet as my financial advisor?" Hey Shwab, what would ____ say about this stock? Create a Portfolio in totay's market based on who _____ would buy? I at least hope simple passive investors will stop only buying S&P

  • @nlabanok
    @nlabanok 12 днів тому +1

    I've attempted many times to convey a philosophy, specific metrics, and a subset of the stock universe to look at to try and get chatgpt to wade through the bottomless pit of available data and I've found chatgpt is just so f'ing lazy...it literally will do almost anything to avoid any actual data acquisition and analysis of large datasets.

    • @jawsh5628
      @jawsh5628 9 днів тому

      Exactly, that's why I've been strong contemplating just building my model locally off Llamas. But it would be so much less powerful. But the Project DIGITS announcement definitely peaked my interest

  • @sfarber12345
    @sfarber12345 14 днів тому

    Curious why you called it the Livermore ETF? Does this strategy replicate Livermore's approach in any way?

    • @ExcessReturns
      @ExcessReturns  14 днів тому

      I think it was just to get at the idea that it is based on the strategies of great investors.

  • @lmmortalZodd
    @lmmortalZodd 13 днів тому +1

    Some comments on the o3 model spectacular test results:
    1. It achieved 87 of I remember right
    2. The questions were designed to be difficult to a non-thinking ai.
    3. The people that designed the test claimed that a score of 70 is probably agi and years away.
    4. There's disagreement on weather this means openai has achieved agi.
    If you ask me, I'd say that if you can't really tell if it's AGI, it's probably AGI or really really close. Hold on to your socks people.