{"id":240696,"date":"2026-03-25T13:45:45","date_gmt":"2026-03-25T17:45:45","guid":{"rendered":"https:\/\/ibkrcampus.com\/campus\/?p=240696"},"modified":"2026-03-26T04:51:58","modified_gmt":"2026-03-26T08:51:58","slug":"ai-will-create-millions-of-quants","status":"publish","type":"post","link":"https:\/\/www.interactivebrokers.com\/campus\/ibkr-quant-news\/ai-will-create-millions-of-quants\/","title":{"rendered":"AI Will Create Millions of Quants"},"content":{"rendered":"\n

The article “AI Will Create Millions of Quants” was originally published on Robot Wealth<\/a> blog.<\/em><\/p>\n\n\n\n

(Most of Them Will Lose Money)<\/h3>\n\n\n\n

AI makes it easier than ever to build trading strategies.<\/p>\n\n\n\n

Prompt a model, run a backtest, optimise some parameters, and suddenly you\u2019ve got a beautiful equity curve staring back at you. It feels like progress. It feels like research.<\/p>\n\n\n\n

I wrote recently about how AI coding assistants tend to prescribe \u201cmore of the disease, faster\u201d<\/a>, skipping the learning that makes trading research actually valuable. This article is the bigger picture version of that argument.<\/p>\n\n\n\n

Because something interesting is happening. The barriers to building quantitative trading strategies are collapsing. And that sounds like it should be great news.<\/p>\n\n\n\n

It isn\u2019t.<\/strong><\/em><\/p>\n\n\n\n

AI makes strategy creation trivial.<\/p>\n\n\n\n

Think about what you used to need to build and test a trading strategy: coding ability, data access, infrastructure, technical knowledge. Acquiring all of that took months or years.<\/p>\n\n\n\n

Now you can prompt an LLM and have a coded, backtested strategy in minutes. That\u2019s a real shift. I\u2019m not dismissing it.<\/p>\n\n\n\n

But lowering the cost of experimentation doesn\u2019t just increase discoveries. It increases false<\/em> discoveries.<\/p>\n\n\n\n

This is the same thing that happened in academic research, factor investing, and machine learning competitions. Cheap experimentation leads to a flood of results that look significant but aren\u2019t.<\/p>\n\n\n\n

The difference is that now it\u2019s happening to anyone with a laptop and a ChatGPT subscription.<\/p>\n\n\n\n

The classic rite of passage of the aspiring quant\/systematic trader is to fall into the Backtest Cycle of Doom:<\/p>\n\n\n\n

    \n
  1. Come up with a strategy idea (or ask AI for one)<\/li>\n\n\n\n
  2. Run a backtest<\/li>\n\n\n\n
  3. Results look mediocre<\/li>\n\n\n\n
  4. Optimise parameters<\/li>\n\n\n\n
  5. Add a filter<\/li>\n\n\n\n
  6. Change the timeframe<\/li>\n\n\n\n
  7. Try a different universe<\/li>\n\n\n\n
  8. Combine some indicators<\/li>\n\n\n\n
  9. Suddenly\u2026 amazing backtest<\/li>\n<\/ol>\n\n\n\n

    It feels like you\u2019re converging on something real.<\/p>\n\n\n\n

    But you\u2019re not. You\u2019re overfitting noise.<\/p>\n\n\n\n

    Each tweak is another implicit hypothesis test. Run enough of them and statistics guarantees<\/em> you\u2019ll find something that looks incredible. If you test 10,000 parameter combinations, some of them will show spectacular performance. Not because they work, but because there will always be something that fit the past somewhere in the noise.<\/p>\n\n\n\n

    AI accelerates every step of that loop. So instead of running 20 bad tests manually, you can run 20,000 with AI assistance.<\/p>\n\n\n\n

    And you\u2019ll find something that looks fantastic.<\/p>\n\n\n\n

    Nice looking backtests are cheap now.<\/p>\n\n\n\n

    This is worth thinking about for a moment.<\/p>\n\n\n\n

    In the age of AI, a beautiful backtest proves almost nothing.<\/p>\n\n\n\n

    The probability that some<\/em> parameter combination produces an amazing equity curve approaches certainty as the number of combinations you try increases.<\/p>\n\n\n\n

    This is just the multiple testing problem, and it\u2019s been well understood in statistics for decades. If you run 1,000 tests at 5% significance, you\u2019d expect about 50 false positives. AI just makes it trivially easy to run those 1,000 tests without even realising you\u2019re doing it.<\/p>\n\n\n\n

    So the thing that used to be hard (producing a good-looking backtest) is now easy. Which means it\u2019s no longer informative.<\/p>\n\n\n\n

    So what actually matters?<\/p>\n\n\n\n

    Before a backtest should even exist, there should be a question: why would the market pay me to do this trade?<\/strong><\/p>\n\n\n\n

    The classic engineer\u2019s mistake is to skip this question (I\u2019m speaking from experience).<\/p>\n\n\n\n

    Engineers work on problems with well-understood physics, and we tend to assume that trading strategies belong in that category. We go straight to data, straight to code, straight to backtests.<\/p>\n\n\n\n

    Eventually, we realise that that\u2019s backwards. We have to do the work to discover the underlying physics first.<\/p>\n\n\n\n

    Answering the question \u201cwhy would the market pay me to do this trade\u201d requires thinking about incentives, market structure, behavioural biases, institutional constraints. In other words, a theory of edge<\/strong><\/em>.<\/p>\n\n\n\n

    Good edges usually exist because of something structural:<\/p>\n\n\n\n