AI and Trading: Why You Should Think Twice

Discussion in 'Artificial Intelligence' started by STS, Jun 27, 2025 at 2:25 PM.

  1. Dismiss Notice

Do you try to use AI for advice & Strategy building

This poll will close on Sep 27, 2025 at 2:25 PM.
  1. Yes

    0 vote(s)
    0.0%
  2. Sometimes

    0 vote(s)
    0.0%
  3. Hell no

    4 vote(s)
    80.0%
  4. I use it to review strategy ideas

    0 vote(s)
    0.0%
  5. I only use it for research

    1 vote(s)
    20.0%
  1. STS

    STS

    AI and Trading: Why You Should Think Twice
    I’ve seen a lot of traders jump on the AI trend, hoping Layman LLMs like ChatGPT and other simple tools will hand them an edge. I believe that to be extremely poor judgement. Here’s why reliance on AI platforms does more harm than good if you want to create or find a profitable trading strategy. Seen a lot trying to cut corners so here you go.


    AI & Layman LLMs: The Consensus Problem
    [​IMG]
    Majority Opinion doesn’t equate to edge

    LLM models such as GPT digest a motherlode of public data from several sources derived from all around the entire internet. Sure, there’s a lot of high quality, but a lot of garbage too. When talking trading, answers provided are based on LLM-induced groupthink, not what actually works.

    On the surface consulting an LLM for strategy building guidance seems smart but it’s the easiest way to become dumb money. It’s extremely saturated too.

    In markets, following the crowd usually means buying high and selling low (-edge)


    The Yes-Man / Echo-Chamber Effect

    These chat bots are built to be friendly and helpful i.e agree with you. You feed them your ideas and they can’t help but nod along.

    A consultant that only tells you what you want to hear is useless. Treat your GPT the exact same way.


    The Misalignment in Perception & Overfitting

    Layman Chatbots vs Institutional Systems

    Free to use and easily accessed chatbots like GPT are nowhere close to systems institutions build &/or have access too.

    Don’t confuse simple chatbots with proprietary machine learning models.

    Using AI to build strategies using historical data is also the easiest way to create overfitted systems; AI has biases you couldn’t even fathom.


    The alternative for those struggling

    Look at how markets actually work. Order flow mechanics, OHLC candlesticks, market microstructure to create your own ideas.

    Form your own ideas, convert them into clear, concise, testable rules then backtest & forward test them honestly. No intuitive gaps. Spell out everything from entry to exit including position management.


    Main points

    Be Your Own Critic

    Avoid the passive role AI places you in; stay curious and sceptical at all times

    Treat your trading like a science experiment. Run fair tests and question every assumption.

    AI can be useful but take every output with a tablespoon of salt.


    Thanks for reading - Sentient Trading Society