Follow Slashdot blog updates by subscribing to our blog RSS feed

 


Forgot your password?
Close
typodupeerror
AI Businesses

Hedge Fund Startup That Replaced Analysts With AI Beats the Market (msn.com) 69

Posted by msmash from the new-frontier dept.
A hedge fund startup that uses AI to do work typically handled by analysts has outperformed the global stock market in its first six months while slashing research costs. From a report: The Sydney-based firm, Minotaur Capital, was founded by Armina Rosenberg and Thomas Rice. Rosenberg previously managed a global equities portfolio for tech billionaire Mike Cannon-Brookes and ran Australian small-company research for JPMorgan Chase & Co. when she was 25. Rice is a former portfolio manager at Perpetual. The duo's bets on global stocks returned 13.7% in the six months ending January, versus 6.7% for the MSCI All-Country World Index. Minotaur has no analysts on staff, with Rosenberg saying AI models are far quicker and cheaper.

"We're looking at about half the price" in terms of cost of AI versus a junior analyst salary, Rosenberg, 37, said of the firm's program. Minotaur is among a growing number of hedge funds experimenting with ways to improve returns and cut expenses with AI as the technology becomes increasingly sophisticated. Still, the jury is still out on the ability of AI-driven models to deliver superior returns over the long run.
This discussion has been archived. No new comments can be posted.

Hedge Fund Startup That Replaced Analysts With AI Beats the Market More

Hedge Fund Startup That Replaced Analysts With AI Beats the Market

Comments Filter:

  • Hardly surprising (Score:5, Informative)

    by RUs1729 ( 10049396 ) on Friday February 14, 2025 @03:25PM (#65166883)
    A while ago it was reported that a chimp throwing darts at different financial strategies performed as well as professional financial experts.

    • Re: (Score:2, Insightful)

      by Anonymous Coward

      Another thing. If everyone starts using "AI techniques" like this hedge fund does to gain an edge, then the trade fails because it becomes an exploitable pattern.
      Somewhat off-topic...
      Long before LLMs became a thing, the second news or a quarterly report hit the wire, speculators were already using machine learning tools to make instant buying or selling decisions on some security (equity, currency, interest rate futures, etc.)
      I remember one notable event in a USD/CAD trade several (10+) years ago, where peo

      • Seeing as a lot of trading is technical analysis rather than knowledge of fundamentals, I question whether the entire industry is heading for a fall? Eventually AI will just be trading against itself and company valuation or success will be completely irrelevant

        • Comment removed (Score:4, Insightful)

          by account_deleted ( 4530225 ) on Friday February 14, 2025 @07:27PM (#65167459)
          Comment removed based on user account deletion

        • Re: Hardly surprising (Score:4, Insightful)

          by alvinrod ( 889928 ) on Friday February 14, 2025 @09:10PM (#65167647)
          Index funds are a better return than almost anything else unless you have insider information. Investment portfolios have always been a scam. The big firms create hundreds of them and kill off the worst performers every year. Even if you're throwing darts randomly, if you throw enough you'll have a hot streak. The portfolio that outperformed the market by xx% over the last five years is the other side of the curve and the hundreds of other portfolios that died out along the way.

          The analysts and firms that create these portfolios don't make their money by being right. They make their money by getting a cut of all of the trades. Whether the portfolio wins or loses, they get their cut. The overperforming portfolio is to get you to invest so they can get a cut regardless of whether it keeps outperforming the market or takes a nose dive. An AI performing the same trick is hardly surprising. Just have it build enough portfolios and one is bound to look good and if you ignore all the losers and hype up the AI aspect you can probably draw in a lot of marks.

          • While I agree with your sentiment, I still feel like a fool for not investing into Nvidia a few years ago when it was rather clear this hallucinating AI wasn't going anywhere. Being more on the cautious side, I just thought, incorrectly, that people would be smart enough to realize most of this is just a glorified search engine.

            I also regret not investing heavily into oil during the pandemic when the market for a barrel basically dropped out the bottom. There was zero doubt in my mind that oil would bounce

            • 1. Create 10 ETFs or mutual funds
              2. Have each one take very risky bets on different parts of the market
              3. Close the ones which went down
              4. Advertise the 1 or 2 winners and bring in more assets.

              Mutual fund companies have been doing this for decades.

              This is a variant on the diversification sales pitch, your advisor, fund company, whatever always wants one or more winners to talk about when you ask them about the losers.

          • Yes. Outperforming the market once is easy. Consistently outperforming it is hard. It is also easier to perform well with a small pot of money than with a large one.

      • Re: (Score:3)

        by HiThere ( 15173 )

        You're assuming every AI uses the same model. This is probably not the way things would work out. If different AIs are using different models, will the all have the same inputs? If so, one may turn out better than the rest (at least over a certain period of time). If they're operating off different inputs, though, you get a situation rather like the current one for the larger players.

    • Re: (Score:2)

      by chthon ( 580889 )

      More than a while ago, I think somewhere at the end of the '90s. I think it was in Belgium that the weekly magazine Humo had an article or a series of articles about that.

    • I've never heard of "MSCI All-Country World Index" but if it's an index typically those are just a basket of stocks that has little intervention beyond the odd swapping of stocks that comprise it (no real buying and selling).

      It is very true that almost anything can (and does) outperform your typical mass market fund, but it's a little nuanced. Definitely, mutual funds are the worst of the worst places to put your money, but that's largely because of fees. Second, for any public fund big enough, or manage
    • Comment removed based on user account deletion

  • AI baby names. (Score:4, Funny)

    by geekmux ( 1040042 ) on Friday February 14, 2025 @03:37PM (#65166913)

    A hedge fund startup that uses AI to do work..

    Why do I have this feeling they named it Cashy McCashface..

    • Re:AI baby names. (Score:4, Interesting)

      by awwshit ( 6214476 ) on Friday February 14, 2025 @04:32PM (#65167073)

      Crashy McCrashface

    • Re: (Score:1)

      by shanen ( 462549 )

      Mod parent funny, but when someone actually figures out how to do this "properly", we'll only find out about it the next morning after it has sucked up all the "value" in all of the markets. And I mean all of the markets. Stocks, future, derivatives, bonds, crypto exchanges, what have you. That will be an interesting mess and won't be so funny.

      Needs a funny name for the AI for the Subject... How about "Sick Smart" as in "I told you it was sick smart!"

      • Re: (Score:2)

        by HiThere ( 15173 )

        You're assuming that the optimal strategy is a LOT better than any current strategy. This MAY be true, but it may not be. There may be a huge amount of random noise in the system.

        • Small investers can trade the random noise and make good money. That is essentially what I do. I day trade with dividend paying blue chip stocks only. That way I cannot really lose - at worst my trade money gets stuck in blue chip stocks earning dividends till the market turns again.

        • Re: (Score:2)

          by shanen ( 462549 )

          No, I'm just assuming that the stupidity can be overwhelmed by sufficient intelligence. Won't actually matter if it's human stupidity or the stupidity of insufficiently "intelligent" machines. But the price fluctuations certainly look like noise to us because the complexity is quite high. Did they ever figure out what caused the flash crash? I'm worrying about an AI that does "understand" what happened and can figure out how to do it again for maximum profit... Also it would have to figure out how to diseng

  • Oh Joy (Score:3)

    by Tailhook ( 98486 ) on Friday February 14, 2025 @03:38PM (#65166917)

    This is an earthquake if true. The greedheads of speculation will lose their fucking minds and there will be a nuclear reactor behind every blade of grass to run their nvidia machines. And no, it doesn't matter that once all speculation is AI powered then they'll just be back at square one. They don't think like that. Not even a little.

    • What is not clear to me yet is how successfull they are with AI. The thing, when you are playing with investor money, hallucinations can be very bad.

      • The thing, when you are playing with investor money, hallucinations can be very bad.

        From their perspective, it's better that it's the investors' money than their own.

      • I know, but it won't stop them. You'd think lawyers would have learned after I think it was the case in NY where an AI made up a case and it was submitted to the judge who was less than understanding. But nope, the reg (https://www.theregister.com/2025/02/14/attorneys_cite_cases_hallucinated_ai/?td=rt-3a) is reporting another lawyer with egg on their face. The AI imagined 8 of the 9 citations that the judge was not happy about. The judge is now deciding on what sanctions may come. Personally I hope the judg

    • Re: (Score:2)

      by _merlin ( 160982 )

      Not really. People have been trying to apply "AI" techniques to trading (e.g. technical analysis, expert systems, etc.) for years. Even basic broker execution strategies have some form of "intelligence" in that they attempt to match or beat VWAP.

  • So? (Score:4, Funny)

    by nealric ( 3647765 ) on Friday February 14, 2025 @03:38PM (#65166919)

    There's a 50/50 chance I can beat the stock market by taking your investments to Vegas and putting it all on the roulette wheel coming up black.

    • Re: (Score:1)

      by Anonymous Coward
      9/19

    • There's a 50/50 chance I can beat the stock market by taking your investments to Vegas and putting it all on the roulette wheel coming up black.

      If you did that, you'd have a 50/50 chance -- actually 9/19 as AC pointed out -- that you would win big. But you would also have a 10/19 chance of losing everything.

      Investing in the stock market doesn't work that way. You don't just put your money on black and wait for one turn of the wheel. You buy shares in a company or a collection of them. Shares that may go up or down, but you can get out with a partial loss if things go against you. It's rare that you "lose everything" on a trade -- you'd have to be v

      • You are missing the point here. A single fun outperforming an index for a single year is basically indistinguishable from a lucky bet. Many funds have even beat the market for years on end only to revert to the mean.

        • You are missing the point here. A single fun outperforming an index for a single year is basically indistinguishable from a lucky bet. Many funds have even beat the market for years on end only to revert to the mean.

          No, you are missing my point: investing in the stock market is not gambling. The two are most definitely "distinguishable." And there is no "revert to the mean" when you gamble, because over the long term, the house wins and you don't. Whereas when you invest (sensibly) in the stock market, you win. You may go up or down in any given year, depending on your choices, but over time, you will win, i.e., see a return. Even if you choose a fund that doesn't beat the market, but makes a fair showing, you still wi

          • No, when you gamble in Vegas you may win, even over your lifetime, and you may lose. You only lose in gambling when you revert to the mean or are always below it.

            • No, when you gamble in Vegas you may win, even over your lifetime, and you may lose. You only lose in gambling when you revert to the mean or are always below it.

              Take a look at any of the games offered in Vegas. Any. Single. One. They all have one thing in common: the odds favor the house, even if only slightly.

              In Vegas, you cannot help but lose over the long term. That is the mean. And believe me, if somehow you're beating the house's odds, they will escort you off the premises and kindly ask you not to come back.

              • As the adage goes, Vegas was not built on winners.
              • My point is that the variation in games of chance, even with a house advantage, requires some reversion to the mean to assume you will always lose. Entirely contrary to the post I was replying to. Elon Musk could lose by doubling his bet at the roulette wheel each time he loses assuming no limit. But those odds are pretty long, like one in a billion. Those other cases? He's a winner. I'll have to ask you to trust me: we can't wait the thousands upon thousands of years for him to revert to the mean.

          • I don't disagree, but it's not really responsive to my point.

            Of course the odds in Vegas are not in your favor. Over a long enough time horizon and sufficnet number of bets you will always go bust. The odds of gaining money in the stock market are in your favor, but unlike a game set by a casino the odds are not fixed. Over a sufficiently long time horizon, you'd probably also go bust because the society that created the market will cease to exist. However, the odds over normal human timescales are generall

        • It didn't outperform for a single year. It outperformed for six months. The story isn't a scientific analysis, it's a sales pitch. The stock fund manager wants funds to manage, and that's why you are hearing about the story (and why the second half of the summary looks like a resume).

          Wittgenstein described this sort of marketing situation subtly: "Typically it constructs. It is occupied with building an ever more complicated structure. And even clarity is sought only as a means to this end, not as an end

    • Re: (Score:2)

      by ceoyoyo ( 59147 )

      There is not.

      There is, however, a 50/50 chance that you can beat the stock market by taking your investments and buying stocks.

  • The real question is how long traditional brokerages and private equity will be the sole beneficiaries of these reduced overhead costs and improved results before some startup gets funds in place to the general public with reduced management fees. Given how heavily regulated the financial markets can be, it's hard to say.

  • So slightly better than S&P500, maybe... (Score:5, Interesting)

    by Amazing Proton Boy ( 2005 ) on Friday February 14, 2025 @03:45PM (#65166943) Homepage

    They provide no real detail on their website, just "+13.6%" with no other data or information. Is that the Nominal price return? Nominal total return? With dividends reinvested? The annualized rate???? Without more detail it's not even possible to tell if they beat a blind investment into an S&P500 index fund. For reference, the same six month period the S&P500 did:

    10.4% Nominal return
    18.3% Annualized return
    11.2% Return with dividends reinvested
    19.8% Annualized with dividends

    Smells like hype from dummies.

    • They provide no real detail on their website, just "+13.6%" with no other data or information.

      We have no idea what the AI model picked. Maybe they picked mostly the S&P 500 which would have been pretty close to their 13.7% gain. If they had picked US tech stocks, they would have gotten way more than their 13.7% gain, so it's not clear that achieving 13.7% is good or bad. Maybe they picked a bunch of high-yielding REITs for the short-term gain. Maybe they picked crypto. The actual picks would have been much more insightful than just a number.

      • We have no idea if they are using LLM or any kind of transformer at all. They could have just read turtle trader, implemented the algos in the book, fired everyone, and called it a day.

    • They are trying to obfuscate this, but they only showed that the AI did better than the stock market in general - which basically is more or less a nothingburger. We have no idea whether the "AI" choices did better / the same / worse than human analysts following a similar strategy.

      Additionally - automated and semi-automated tools that do exactly this have been around for a decade or two. Is this really "AI", or is it just a market-driven rebranding of previously existing market analysis tools? Smart money

    • Re: (Score:2)

      by Pascoea ( 968200 )

      Smells like hype from dummies.

      I would say it smells like hype FOR dummies.

    • It's hype from a salesperson.

  • The AI is only better until they all switch to AI, then they all get the same answer.

    • Sort of, but stock market investing is kind of a herd activity, so currently people are all looking for the same answer anyway.
  • Because more often than not we find out AI is actually just somebody using regular human beings and lying through their teeth.

    I'm not saying it's impossible but I am saying I wouldn't be surprised if in a couple of years we have an article about how the CEO is facing prison time for scamming some rich people

  • https://www.msci.com/documents... [msci.com]

    I do not see the figures they're talking about in the Index. If they mean the primary index it was about 19% annualized which beats them. If they mean so ethically g else, then I'm not seeing a t hi g figure here.

  • This is great, until the AI forgets what money is.

  • These LLMs don’t “think”, they analyze what others write. So the more people move to AI the more often you’ll get Ai analyzing AI output.

    • Re: (Score:3)

      by HiThere ( 15173 )

      Did it say they used an LLM? It didn't in the summary. There are LOTS of AI programs, and the stock market is reasonably easy data to fit to many of them. (I had an uncle who tried to do that sort of thing with simple curve fitting. He didn't lose his shirt, but he also didn't win big. And it was a lot of work.)

  • If this worked, they would not want your money (Score:4, Insightful)

    by ffkom ( 3519199 ) on Friday February 14, 2025 @05:55PM (#65167265)
    Whenever somebody claims having a method of reliably making a profit above market average, just ask yourself: Why would that somebody want YOU to invest YOUR money - instead of just profiting silently alone, without all the hassle of operating a fund?

    The simple truth is: Those companies usually run N different funds. Then advertise their services using the numbers of only those funds that happened to run well for a while. Then they collect the gullible investor's money.

  • [Minotaur Capital] has outperformed the global stock market in its first six months

    Six months is short term results. A coin flip can outperform the market in the short term.

    Over the long term, the best results are from index funds. The longer the time span, the harder it is to "outperform the market".

  • a 20% performance fee on any profit. WHAT da F@$K !!!! From the Bloomberg article ! Heck. There goes all your profits !!!! NEVER have heard of costs like that !

Slashdot Top Deals

He who has but four and spends five has no need for a wallet.

Close