Browse By

IDEA #31 – Stock Market Social Media Tracker

From Eric Nagel, friend and colleague (from an IM conversation, so apologize for the brevity and not backing up the sources of these numbers):

I heard an unbelievable fact… 50% of all US stocks are held by 1% of all the shareholders in the world and 90% of all stock is held by like 10% of the shareholders of the world. So 90% of the shareholders in the world only own 10% of the stock out there.

The Idea: Use social media to provide some additional data/feedback regarding stocks. How many times is a company’s name [or ticker] mentioned in the press? Mentioned on blogs? (Possibly create a Techmeme site for stocks). Does this at all correlate to the actual stock price of a stock? When news hits from one of the primary press sources (CNBC, MarketWatch, WSJ, etc) — what typically happens to the stock? Who are the competitors of a stock — what’s their P/E compared to the others and are they all trending in any direction? How does one sector influence another (are any dependent on another — or do any shift when another sector shifts)?

The application would likely need to know how to parse out key information from an article — such as information on an earnings announcement, or whether bloggers are hyped (or loathing) about something.

What other data / analysis could be found (or derived) from the web or blogosphere?

  • Wayne

    I once thought about making something that took SEC filings and allowed me to compare certain expenses over time and against competitors. For example, is the selling/general expenses going up or down and how does it compare with other companies their size or in their industry over time.

    What got me thinking about the SEC filings was that I once wanted to break down costs, revenue and expenses by segment of a company; if the data was available. Unfortunately a lot of them don’t break this out in public filings.

    There is a lot of structured data from the SEC filings which could be used for this.

    Partnerships via news releases might be useful, there is always a section at the bottom where they talk about each company. This would make it easy to extract what organizations are involved.


  • Eric Nagel

    Check out:

    Investors rate stocks… the better they do, their picks are highlighted.

    I just had some software written to help me screen for stock picks… I have the data mining part done, now I need to write an algorithm to use the data to rank stocks. In my spare time :)

    Somehow tie this in with perhaps Google Alerts, to see who’s talking about these companies?

    example data:
    Data on grmn
    tickerobj Object
    [fLastTrade] => 54.48
    [fRecThisWeek] => 2.7
    [fRecLastWeek] => 2.7
    [fMeanTarget] => 60.60
    [nNoBrokers] => 15
    [aSurprises] => Array
    [“Mar-06”] => 26.5
    [“Jun-06”] => 14.6
    [“Sep-06”] => 0.0
    [“Dec-06”] => 50.0

    [nCaps] => 4
    [nTotalOutperforms] => 1646
    [nTotalUnderperforms] => 58
    [nAllStarOutperforms] => 421
    [nAllStarUnderperforms] => 13
    [nWallStOutperforms] => 7
    [nWallStUnderperforms] => 1

    Compared with, say Ford:
    Data on f
    tickerobj Object
    [fLastTrade] => 7.90
    [fRecThisWeek] => 3.5
    [fRecLastWeek] => 3.4
    [fMeanTarget] => 8.19
    [nNoBrokers] => 8
    [aSurprises] => rror
    [nCaps] => 1
    [nTotalOutperforms] => 1085
    [nTotalUnderperforms] => 928
    [nAllStarOutperforms] => 157
    [nAllStarUnderperforms] => 164
    [nWallStOutperforms] => 3
    [nWallStUnderperforms] => 9

    I can quickly see which is the better pick, according to analysts, investors, etc.

    However, in the end, the only rule that applies is:
    “If there are more buyers than sellers, the stock price goes up; if there are more sellers than buyers, the stock price goes down”

  • Kevin

    I’ve tried something like this in the past, more around pulling in web content for a given company. It’s a hard problem, definitely opened my eyes to the whole “semantic web” problem. Still, it’s one of those problems where if you can even get a 1% improvement over existing algorithms, you’ve suddenly got something very valuable in your hands.

    My favorite theory is this one from the Univerity of Michigan (go blue!) b-school:
    Basically, they found a correlation between lunar cycles and worldwide markets – returns are up on new moons and down on full moons. I’m not a hippy or anything, but thought it was an interesting concept. :)

  • rulepark

    do we have live shareholder talking on a page while, looking at the live data display on the page?

  • newsmgr

    Check out a company called Collective Intellect ( They search “new media” sources for clients (companies, investors, etc.) and then provide analytics on what they find. Pretty sophisticated process to analyze the data and pick out relevant stuff. It may not provide all the features mentioned in the post, but they are in that space.

    Disclaimer: I’m not affiliated with the company in any way, just happen to read the CTO’s blog.

  • Steve Poland
  • Commodity Market

    I checked out your blog. Really cool stuff. I also spent a few minutes on their blog.
    Thanks for sharing.

    Commodity Market

  • Yuriy

    Send me email (yuriless [at] gmail [dot] com) if you are interested in collaboration regarding this subject.