Is Twitter The End-Game For Street Traders?
In 2013 social media is apart of everything we do from checking–in at our favorite pizza place to tweeting about the latest sports news. But how does WS trading fit-in or does it? To me, it seems each successive major technological transition has led trading further and further away from the street and the latest one comes from Paul Hawtin, CEO of DCM Capital who’s launching his platform tomorrow called, DCM Dealer that utilizes a social media algorithm for sentiment-based investment analysis. Read on to find out about how it works and what it means for WS trading…
TECHNICALS
In a February 2011 article in the Journal of Computational Science titled, Twitter Mood Predicts The Stock Market, Bollan, Mao, and Zeng of the Universities of Manchester and Indiana predicted closing DJIA prices 87.6% of the time. They used Opinion Finder and Google-Profile of Mood States on 10 months of Twitter data from 2008 and cross-validated it with the election and Thanksgiving results from the same year. Then they analyzed these results with a Granger causality analysis, a statistical test of significant forecasting ability of a particular time series of another. Subsequently, they used a self-organizing fuzzy neural network, a type of self-leaning AI that I believed mimicked Twitter activity, with the inclusion of specific mood variables that yielded their results.
Will this be quickly priced into securities or spawn a new market from social media?
PERFORMANCE
London based Derwent Capital Markets then re-created this algorithm based off of this academic research and formed a ₤25M HF subsidiary in 2011 that garnered a return of 1.85%. As such they received the award for Excellence in Trading & Technology Europe in that year. Despite the modest returns, the fund was quietly shutdown due to the hostile investment environment which was a bad year for equities even at IB's, see the FT League Tables.
Do you incorporate social media into your trading analysis at all, why?
RELEASE
Here’s what to expect tomorrow when DCM Capital launches its platform…
• Crawls 350M teewt per day and 900M Facebook users for real-time sentiment data
• 250 UK equities, major FX pairs, and commodities but purports the ability to trade 1000’s
• Offers CFD trading or spread-betting
• Metrics range from 0-100 with 0 being very negative and 100 very positive
• Charting with 26 different tools, customized layout, sentiment alerts, set limits, and portfolios
• End of January plans to be in EURO equities and into U.S. and Asian markets next month
Is this another nail-in-the-coffin as trading continues to de-centralize from WS?
SO LET'S HEAR IT MONKEY'S, SHARE YOUR THOUGHTS WITH US?
No.
This. Two other funds have tried to run off of Twitter-driven algorithms and failed.
I feel like this subject has been beaten to death
[quote=GMngmt]₤25M HF subsidiary in 2011 that garnered a return of 1.85%. quote] does a 1.85% return prove anything?
last year I started a dummy account and bought stocks on Mondays that started with the letter A. On Tuesdays I bought stocks that start with B, and so on. I made a 9% profit! My genius strategy must work.
correlation doesnt means causation. E.g. No. of priates out in sea have been decreasing over the last 300 centuries while global temperature has been increasing. Does this means that the number of pirates affect global temparture?
Thanks for the pirate example. I didn't fully understand the concept before, and now I do.
Thank you everyone for your comments on this subject! I’d like to address a number of them…
So, I’ll take your word for two Twitter-based algorithm failures. And how many non-Twitter algorithm failures have there been that we don’t know of like the Knight Capital incidence? I’m not defending DCM Capital but let its performance speak for itself and not cite the Law of Small Numbers bias.
At the surface it would seem yes since in 2011 the HFRX Global Hedge Fund Index was -9% as of 12/27. But as I wrote, they were only vested for a month, not 12 so you’re right and in my opinion, it doesn’t mean much.
That’s quite a bite simpler than the Twitter-based algorithm but point taken. I’m sure the market will shake-out the less efficient indicators, if this be one of them.
This is true and that’s why the Granger Causality Test was applied. If you read the study, the authors explicitly state that when reporting their findings. In this case, we’re all on the same page.
Ad molestias voluptatum vel aspernatur. Officia iste odio voluptate sit maxime aut. Quidem consectetur quaerat beatae doloremque qui ad.
Laboriosam est eligendi rerum velit quis fugiat. Itaque in voluptatem nemo illo enim aut. Alias in odit aut autem.
Nam et et inventore qui molestias. Cupiditate qui aut vero consequatur illum qui iste sunt.
Rem eaque quia doloribus voluptatem rerum. Omnis aut id accusantium necessitatibus ducimus minus quia. Itaque sed aut adipisci laboriosam. Minima ut odit eveniet qui aut harum tempore.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...