Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
The Almighty Buck Twitter Idle News

New Twitter-Based Hedge Fund Beats the Stock Market 209

nonprofiteer writes "Derwent Capital, a new hedge fund that makes trades and investments based on Twitter sentiment, beat the market — and other hedge funds — in its first full month of trading. From the Atlantic: 'Using an algorithm based on the social media mood that day, the hedge fund predicted the market to make the right trades. Sounds unbelievable that something cluttered with mundane musings and media links could have anything smart to say about the market. But it's working so far.' Blind luck?"
This discussion has been archived. No new comments can be posted.

New Twitter-Based Hedge Fund Beats the Stock Market

Comments Filter:
  • Cool (Score:4, Funny)

    by itchythebear ( 2198688 ) on Thursday August 18, 2011 @09:35AM (#37129588)

    Twitter may have finally found a way to make a profit!

    • by Svartalf ( 2997 )

      Heh... It's certainly nothing worse than anything else the hedge funds do to try to "make a profit".

    • In other news: Million monkeys typing on typewriters have written "War and Peace"! I suggest waiting until they are able to demonstrate consistent results as opposed to proclaiming great success after first month. I'm going with 'blind luck' on this one for now.
      • by cusco ( 717999 )
        Two decades ago the Wall Street Journal pitted the "best" (as in the most expensive) fund management experts on the NYSE against a blindfolded secretary throwing darts at the Dow Jones listing. She beat ALL of the experts every month but one over the course of a year, and her performance was about 50 percent higher than the next-best performing analyst. This news doesn't surprise me a bit.
        • Re: (Score:2, Funny)

          by TheLink ( 130905 )
          Ah but who earned more for themselves and continued earning more?
        • by Smauler ( 915644 )

          You can't make money on stocks now... the FTSE is about the same level as it was in the early 90's. No one seems to mention this - if you'd invested across the board, 20 years ago, all you'll have back is the dividends.

          Money out of nothing doesn't work.... the reason pension funds etc are doing crap is because the markets aren't growing. Pension funds did invest across the board, and because there's no real inflation in their investment, they're in trouble.

          In desperation, banks have been looking to new

  • One whole month! (Score:3, Insightful)

    by Anonymous Coward on Thursday August 18, 2011 @09:36AM (#37129614)

    It beat the market for one whole month? Wow! That puts them in the same class as 50% of high-school finance students!

    Show me the three year and I might start to be impressed. If it doesn't go broke in 10 years, then I might take it seriously. A random pick of stocks has a non-negligible chance of beating the market as a whole in a single month.

    • Re:One whole month! (Score:5, Informative)

      by TheRaven64 ( 641858 ) on Thursday August 18, 2011 @09:59AM (#37130012) Journal
      There's an old scam that works because of this. You set up a few funds, say 30, and make random trades with each one. On average, most of them will do about as well as the market as a whole. A few of them will do much worse. You close these. A few will do much better. You then get people to invest in these (with the obligatory disclaimer that past performance does not ensure future returns).
      • Re:One whole month! (Score:4, Interesting)

        by greg1104 ( 461138 ) <gsmith@gregsmith.com> on Thursday August 18, 2011 @11:38AM (#37131370) Homepage

        I always liked the related scam for selling stock selection services. You e-mail a group of marks "stock XYZ will go higher in the next month!", breaking them into (say) 4 groups. The next month, you contact everyone who got a good recommendation the last time with "stock ABC will go higher in the next month!". Repeat a few layers deep. After 3 or 4 such calls, a fraction of the people you contacted have now gotten nothing but winning picks from you; them you try to sell your picking service to.

      • Someone has been watching too much Leverage :)

    • "Even a stopped clock tells the right time twice a day"

    • by ceoyoyo ( 59147 )

      More than 50%. There's a study that shows a random stock picker will beat the average fund manager.

    • In other news: Werdent Capital, a new hedge fund that makes trades and investments based on Facebook sentiment didn't beat the market or make any headlines.

    • The fallacy at work here is what trading system developers call "curve fitting". If you're given a set of data and claim there's a correlation between two things in it, you can always fit a curve to predict one from the other with good accuracy if you work on it a while. Clear signs of curve fitting in play are "magic numbers"--constants like the "2 to 6 days" alluded to here, where the model doesn't work unless you get them right.

      The fun thing about curve fitting is that you never know when it's going to

    • You simply have to understand what money is. Most people don't understand and are deliberately misled by those who do, so that they can fleece them... You need a bag holder, the guy who buys at the top and sells at the bottom.

      I've been beating the market for several years because I understand exactly what's going on, where as most people seem to view life as a series of completely random unconnected events and are therefore continually caught by surprise.

  • by Jordan (jman) ( 212384 ) on Thursday August 18, 2011 @09:37AM (#37129626)

    I think twitter will have the same effect as a monkey throwing darts...

    http://www.automaticfinances.com/monkey-stock-picking/ [automaticfinances.com]

    • Not really. There is no correlation between darts and the stock market. Twitter stuff is based on the hive mind of humanity. The stock market is also based on the hive mind of humanity. It's resonable to expect that there is some correlation. But the correlation is only useful if some attribute of twitter can be shown to lead the market. If there's not a predictable correlation, it's worthless. If there's a correlation, but the market reacts first, and twitter lags, then its' worthless.
      • But the correlation is only useful if some attribute of twitter can be shown to lead the market.

        And only if you can eliminate all random events from the world. eg. What if there's an earthquake tomorrow? The CEO is discovered having an affair? Some granny goes on TV saying her car accelerated suddenly...?

    • Sir, I would like to buy your monkeys.
  • One month? (Score:4, Interesting)

    by Mabbo ( 1337229 ) on Thursday August 18, 2011 @09:37AM (#37129638)
    They did better than average for one month. I could buy a random subset of stocks, and still have a 50% chance of beating the average. Call me if they can maintain this for 12 months straight. Then maybe they can see some of my money.
    • I could buy a random subset of stocks, and still have a 50% chance of beating the average.

      Heh, good luck with that! Actually, chances are lower because the increases in stock value are not homogeneous.

      • Illustration: There are 100 stocks. 1 goes way way up, like 100x. 99 go very slightly down. A random subset of 10 stocks does not have a 50% chance of beating the average, since there is not a 50% chance it will include the 1 that went up. (Yes, I know I'm agreeing with you.)
        • by sribe ( 304414 )

          Illustration: There are 100 stocks. 1 goes way way up, like 100x. 99 go very slightly down. A random subset of 10 stocks does not have a 50% chance of beating the average, since there is not a 50% chance it will include the 1 that went up. (Yes, I know I'm agreeing with you.)

          And what exactly are the odds of such an odd distribution? Rather small perhaps? Isn't it more likely that deviations from the average will be spread more evenly???

          • by Sique ( 173459 )

            Actually no. Stocks often have a geometrical distribution, with a large group slightly below the average, and a few very high.

        • by ceoyoyo ( 59147 )

          There are 100 stocks. 1 goes way down, to like 1%. 99 go very slightly up. A random subset of 10 stocks does not have a 50% chance of beating the average, since there is not a 50% chance it will include the 1 that went down. (Yes, I know I'm disagreeing with you.)

          In concrete terms, you're suggesting that the distribution of stock changes is highly skewed, and claiming that it MUST be skewed in terms of a long tail of stocks going UP. You're going to have to justify that.

        • Yes, but there's also a chance that 1 goes way down, while 99 go slightly up. With no other information we'd have to conclude that there is an equal chance of either situation occurring, thus there would indeed be a 50% chance of beating the market with your random 10 stocks.

        • by lee1026 ( 876806 )

          On the other hand, a random subset of 99 stocks have a VERY good chance of beating the market....

        • by vux984 ( 928602 )

          1 goes way way up, like 100x. 99 go very slightly down.

          Hmm.. That's not the stock market i know. In the stock market i know...

          In a given month, a stocks go up (some go up a lot, some barely at all), b stocks go down some go down a lot, and some barely at all), c stocks are delisted (removed entirely - lets just consider them bankrupt although there are other situations that can cause a symbol to stop trading that aren't so dire), and d stocks are newly listed.

          A random selection of stocks is likely to do wor

      • by sribe ( 304414 )

        Heh, good luck with that! Actually, chances are lower because the increases in stock value are not homogeneous.

        Your reasoning is backward, if the increases were homogeneous, the chance of a random subset beating the market would be 0%. On the other hand, if the deviations from the average are randomly distributed the chance would be 50%, and this seems pretty likely.

        • Really? I thought it was much more likely that a few stocks will rise really high, while the rest fluctuate randomly, so, in order to beat the market, one would have to pick those few that "drive the economy".

          Nice thought, the one about being homogeneous nullifying your chances :) .

      • by ftobin ( 48814 ) *

        Due to the Small minus Large premium, randomly choosing stocks has outperformed over the long term. See equal-weighted indexes as an example of this.

    • I am reading that the average hedge fund made 0.76% during the month, while this particular fund made 1.85%. Woohoo...
      One month is a ridiculously small amount of time to judge an investment strategy.

    • You really should look up the definition of Average. You're using the definition of Median not Average. Your definition of average only applies if the stock market results are normally distributed, or in another way equal number of winners and losers and anyone who watches the stock market knows that is rare otherwise it would go up and down. What they did was they managed to get a profit of 1.85% when the average went down by 2.2%.

      A longer test would be nice but comparing them to the return rate for Jul

    • by Syberz ( 1170343 )
      In fact, wasn't there a study a few years ago where they let a bunch of preschoolers pick stocks at random and they ended up doing better than the fund managers?
      • by cusco ( 717999 )
        There are a bunch of them, actually. The Wall Street Journal used a blindfolded secretary throwing darts, someone else used monkeys throwing darts, one of the hedge funds used a chicken pecks (as a joke, and then were appalled when the chicken beat their own managers). It's pretty well established now that managed funds are just about the worst possible way to invest your money, which is probably whey most companies insist that their low-level employees invest their 401k that way but put no such restricti
  • Now a Twister-Based Hedge Fund would be a lot more fun than a Twitter-Based one...

    ttyl
              Farrell

  • It should come as no surprise that markets often function (esp on the short term) based on rumor, gossip, mood swings vice reacting to actual intrinsic value of a company or a sector. Hence focusing on a form of media that specializes in the superficial is likely a reasonable decision for someone wanting to play a short term game.

  • by rwven ( 663186 ) on Thursday August 18, 2011 @09:43AM (#37129756)

    Sounds like the theory from the movie "Limitless" put into practice.

  • How much of this is attributed to excessive tweets due to, and in conjunction with, a highly volatile market?
    • Month 2 - "Ummm... due to high volume of fake tweets from anon accounts, we made a massive loss. Net growth: 0."
  • IANAE, but maybe in the current environment of uncertainty, there's more predictability in following the psychology and state of mind of investors than following the "fundamental" indicators. But it seems that also contributes partly to the problem where we see more and more mimicry, which leads to larger cascades of buying and selling, thus even more volatility and sense of uncertainty.

  • by RobinEggs ( 1453925 ) on Thursday August 18, 2011 @09:47AM (#37129834)
    So by following Twitter trends he can make investments that beat other funds in the short run? Are we supposed to be even remotely surprised here?

    Everyone knows the stock market responds faster to fear and to delusions of sudden prosperity than to hard data; that's a large part of its problem.

    Detecting and exploiting those fears and delusions accurately is a good trick (and I'm sure it isn't easy, even with this method). But it doesn't make the man a genius by a long shot. Nor does it make him a useful investor: banking on the current "mood" means he's actually inflating the dangerous cycles of emotionally driven, short-term investment decisions rather than making any kind of long-term decisions.

    I've been ripped before for criticizing short term trading, including HFT trading, but I still think the people who keep the market even remotely stable and the people who make the market useful for it's true purpose (giving corporations a bond market and investors a place for potentially stable returns) are long term investors who follow the data.

    And following twitter isn't what I mean by data-driven decisions.
    • by Ibag ( 101144 )

      ...but I still think the people who keep the market even remotely stable and the people who make the market useful for it's true purpose (giving corporations a bond market and investors a place for potentially stable returns) are long term investors who follow the data.

      The problem is that without the people who are speculating and taking bets, you wouldn't have nearly as many people that could make serious investments in companies. People invest in companies (in the sense that you want them to) because they expect them to grow. This is all well and good until an investor decides that he's done investing in the company. Maybe he's less certain about the future growth of the company. Maybe he just needs to free up capital for something else. Whatever the reason, he nee

    • Nor does it make him a useful investor: banking on the current "mood" means he's actually inflating the dangerous cycles of emotionally driven, short-term investment decisions rather than making any kind of long-term decisions.

      Actually, it's just the opposite. To the extent that the "mood" is irrational, and does not reflect long-term value, the best way to profit is to invest against the mood, buying from those cashing out due to irrational fears and selling to those irrationally willing to pay a premium. This tends to dampen the cycles, not inflate them.

      Simply copying irrational behaviors would tend to amplify the instability—but that wouldn't help them make a profit. The only way to profit, long-term, is to bring the pri

  • Harry Browne wrote a book about investing called "Failsafe Investing". In he he makes a pretty good statement. You can't beat the market long term. Any investment you make based on past performance is not logical. He proposed a thought experiment. Take a room of 100 people. Ask them to individually pick heads or tails. Flip a coin 6 times. You will likely have at least one person that picked all 6 right. Are they psychic? Are they the best coin flip picker? Nope they were lucky. The same with investors that

    • I don't agree with you.

      There are managers who've become famous for "beating the market" simply because they rationally examine what other people refuse to see and leave a field once they realize the ratings are all shit or they realize that data-driven investing has yielded to emotional positive-feedback loops.

      You don't beat the market with bold declarations about a coin flip, you beat it with cynicism and independent thought; few if any people can predict the economy or the performance of a particular
      • by iONiUM ( 530420 ) *

        I kind of agree with you. In reality, nobody can predict the market. But in practice, you don't need to. It's pretty easy to see a downtrend in the broad market, and when that happens you move from equities into bonds. And then when the market begins to trend upwards again, you do the opposite.

        All you really do is follow the trend, without making any risky moves, and you will always make money both when the market goes up, and when it goes down. It's not rocket science, but it does require moving money ever

        • You are forgetting the rebalancing. When gold goes up and stocks go down enough you sell gold to buy stocks. When it reverses you do the same. He recommends rebalancing when one goes more than 10% from 25%.

          • by iONiUM ( 530420 ) *

            Actually he said re-balance "once a year", which means to me he just has an arbitrary day he re balances, which (if he did it today, for example), could turn out bad. He's not reacting to the market, he's just doing things "because."

            • Read again. He said re-balance once a year and if you are watching the market and there are some big swings check your portfolio to see if any go below 15% or above 35% or the total value. I check it once a month and it rarely requires re-balancing.

      • No one can always time the market perfectly but a good planner/adviser should know through careful analysis when you are nearing the top of a peak or bottom of a trough. Granted you won't gain as much as you won't be getting in at the absolute bottom or out at the absolute top, but you will be able to catch most of the upswing and avoid all of the down swing (and no I don't mean daily market variations) so it becomes easy to beat the market when doing that. This is also why casinos won't let you skip more t
    • You can't beat the market long term.

      Is like saying that playing poker is all luck. Yeah, luck plays a part in it, maybe even a big part, but it's still the same group of 20 top players that finds themselves at the final table at tournament after tournament. People like Warren Buffet didn't become ludicrously rich just because they got lucky.

      • by sjames ( 1099 )

        There actually appears to be a huge factor of luck involved, bolstered by "the club". Make enough and then it doesn't matter what happens. You can loose it all when statistics catch up with you and other investors will throw enough money at you to keep you going.

        For every Warren Buffet out there, there's a few thousand others just like him who weren't lucky enough to get the needed windfall to get started. That's not to say he isn't smart, he is. That's a prerequisite. Alas, it is necessary but not sufficie

    • I'd lump gold and bonds into the same bucked and use what Buffet suggested one time and keep 1% in Bonds for each year you are. So if you're 35 then you keep 35% in the safe bond bucket. That way by the time you retire you already have most of your money in the safe buckets for your retirement.
    • That actually doesn't look too different from what I have for my 401k and other investments. I have less in money markets but generally it is similar. A more general form is label the categories would be stocks, commodities, bonds, and cash.

      Now It is possible to get better than market returns fairly consistently but it requires that one studies and analyzes the data for long term gains and not chase fads, or has a planner who's job it is to do that. Over the last 7 years I have had an average return clos
  • The article didn't say HOW WELL it "beat the market" (that is, what percentage return), nor does it say how it did on a day-to-day basis. So I'm treating "beat the market in its first month" as a single data point.

    We'll see how it does next month, and the month after that, and the month after that... [wikipedia.org]

    • From the source of the source...

      [Twitter based trading] made 1.85 percent in its first month of trading, ending in July. This not only beat the S&P, which fell 2.2 percent that month, but it also beat out the average of other hedge funds, at 0.76 percent.

      So it's pretty significant. It's all based on a paper which showed that there's a 5-8 day lag in the correlation between Twitter sentiment and stock price. If something is getting negative attention on Twitter, there is a nearly 90% chance that it's stock price will drop ~1 week later with a similar relationship for positive attention. I imagine people hear something on twitter, make an appointment with their financial adviser or make a note, then a few days later actually

  • ... but the idea of gauging a companies prospects by watching customer sentiments is quite valid. There have been a number of anecdotal cases (sorry, can't recall any links) about people (sometimes just kids) who have done well in stock market games just by walking through the shopping mall, looking for popular stores.

    Sure, there's got to be some due diligence. One has to weed out the outfits with great product ideas but crappy business plans. But everything boils down to customers and market. Find happy

  • Sudden crash? (Score:4, Insightful)

    by djlemma ( 1053860 ) on Thursday August 18, 2011 @09:53AM (#37129920)
    I wonder how the twitter fund is doing with the sudden 500 point drop in the Dow this morning...
    • The wild swings of late have more to do with computers being setup to do trading. They get new input then they start doing rapid wild trades which results in it down 500 one day then up 500 the next.
  • Would love to see someone figure out how it works and get it to buy some crappy penny stock in order to dump it. What a great use of twitter spam.
  • ..we can influence their predictions by coordinated postings of a large number of targeted "mood" tweets

    and here goes your tweet-based prediction out of the window
    • by dkf ( 304284 )

      ..we can influence their predictions by coordinated postings of a large number of targeted "mood" tweets

      and here goes your tweet-based prediction out of the window

      Except you'll likely persuade lots of real traders that they've got to change their positions too, at which point you'll end up trampled by the stampede of mooing morons. That's the point when the rest of us will really laugh.

  • "A change in emotions expressed online would be followed between two and six days later by a move in the index, the researchers said, and this information let them predict its movements with 87.6 percent accuracy"

    Say What?:
    - "A change in emotions expressed online" (50/50)
    - "would be followed between two and six days later" (2-6 attempts)
    - " by a move in the index" (50/50)

    Rephrased:
    - Flip 2 coins, you'll get the same face 87.6 percent of the time, if you keep trying up to 6 times.

    I know i'm stepping into t

    • by ceoyoyo ( 59147 )

      You'd be correct, except that longitudinal data is positively correlated. Checking the market six times is not equivalent to flipping a coin six times. Also, presumably by "move" they mean a move relative to the original price, not the last time they checked.

  • The mathematical models need access to a large number of independent human minds to effectively control the level of uncertainty exhibited by the stock market. Formal (and hence finitary) mathematical methods just cannot cope properly and reliance on them is usually the cause of stock market bubbles and crashes.

  • I was reminded of Bayesian Inference [wikipedia.org], where experts make their best guesses along with probability limits. Twitter isn't exactly like that, but the stock market is driven by sentiment. People should buy low and sell high but they tend to buy high and sell low. ("Stocks are crashing! SELL! SELL SELL! Stocks are going up! BUY! BUY! BUY!") Measuring the mood of the crowd might be a good way to figure out the herd mentality and try to get some money out of it.

    Not saying it works, but that's probably the theory

  • ... tweet me a new minivan!

  • Something that takes as it's input random psychological output from many sources does well at predicting the performance of a system that takes as it's input random psychology? Makes sense, actually. The stock market isn't always governed by rational decision making. What they seem to be attempting is to cut out trying to evaluate the rationality of people's though processes, but just observing them. The danger here is that if enough people behave stupidly, this thing might enhance that through it's tra
  • Not blind luck: Insider Trading. If Twitter's not a news source and is instead a "social medium" then you're getting hot tips from insiders somewhere in that mess.
    • Does it count as "insider" trading if the same information is available to everyone?

      What would be interesting and extremely valuable (and open to manipulation, even by the tweeter if they became aware they were an indicator) is to come up with a "hot list" of people who's tweets had some sort of correlation with market movements. Whether you'd have to go further and demonstrate direct causality (maybe the CEO's children: we're going on a long cruise / no ski-ing holiday this year) would be an interesting

    • It's not XKCD but just as appropriate and obligatory! Dilbert, algorithms and insider trading [wordpress.com]
  • If an algorithm based on "mood swings", coupled to the utter crap that comes spewing out of Twitter 99% of the time, is being used to determine market investments, I don't know how that could speak any louder as to just exactly how ridiculous investment planning has gotten.

    Oh, and all you Wall Street professionals, you might want to start looking for a new job. Sounds like you were just replaced with Twitter. Yeah, I know, I'm as shocked as you are, your new boss is a hash tag. Don't worry though, you ca

  • While the market in the long-run does follow "Data", as a famous Economic Nobel Laureate said "in the long-run we're all dead".

    The long-run is, of course, made up of lots and lots of short runs. On a time-scale of days/hours/minutes/seconds (even micro-seconds now) the market looks brownian, or fractal, or like noise. As any serious trader knows it is emotions not logic that drive it in the (very) short-run. (Excepting purely quantitative imbalance corrections, sorry I forget the term).

    Consequently this

  • unbelievable that something cluttered with mundane musings and media links could have anything smart to say about the market

    The "market" is not some sort of magic Oracle. It is the result of the cluttered mundane musings and uninformed opinions of a teeming mass of people who are basically GAMBLING.

    I really wish people would get off this fetishism, near-worship of "THE MARKET." It is not infallible, and it is sure as hell not the answer to all of our society's problems.

  • FTA:

    A change in emotions expressed online would be followed between two and six days later by a move in the index, the researchers said, and this information let them predict its movements with 87.6 percent accuracy

    So they make a prediction and then wait for up to 6 days for the market to go the way they predicted, and they only ran it for a month. Seems like random fluctuation to me.

  • The market is stupid easy to beat, and there are lots of well established methods to beat it. One of the easy ways to do it is to buy a a distribution of index based mutual funds and bonds, and readjust every year to maintain that distribution.

    That will not only beat the market, but also beat 90% of all the other mutual funds.

  • something cluttered with mundane musings and media links

    You described the pyramid scheme that is the stock market. Were you trying to describe twitter? Oh. I see. The ARE the same thing. My bad.

By working faithfully eight hours a day, you may eventually get to be boss and work twelve. -- Robert Frost

Working...