Also in this playlist...
This transcript is automatically generated
Ever think -- your tweets might predict the future.
And Indiana university professors been putting your tweets into a powerful computing system to determine -- general public mood and by Halloween.
How we feel he's predicting the movement at the markets and -- shocking six fax.
Tests from his first paper showed an 86 point 7% success rate against short term ups and downs on it now.
Here to tell us more is the man himself Johann -- and welcome doctor.
Either hire a PC sends along this chart it is essentially snapshot of how -- sentiment dating you glean from Twitter.
Pretty much parallels the performance of the Dow hopefully we can pop that up for our viewers at home to see it is.
As I -- almost a direct parallel fascinating stuff.
How the idea to go to Twitter to gauge market sentiment.
While there was a little bit of lucky coincidence don't like team and I have very interest in in public sentiment and -- such focused on not collective mood states doesn't exist in the public and twitters a perfect tool for that.
-- people volunteers so much information about their personal mood states about their activities there whereabouts etc.
and they're doing it in real time -- very short saw.
-- short -- period so it was an ideal data set for us to look at.
-- give an example of some of the words or phrases you look for that would -- give you an indicator.
What does a lot of goals into the secret sauce you know we're trying to make his computer algorithms.
I don't care can visit -- not as complicated as we candidates complicated is necessary.
But it looks for things like dramatic you know grammatical -- certain words like up com sad to.
Still committee did actually recognizes certain American dialects.
-- -- record it recognizes -- indications of his old.
-- bunch of features that go into this assessment that the computer's look forward to to draw some kind of indication of the mood state of the person.
Thank you raise a great point because Twitter so widely used would be -- easy for a market manipulator someone with that goal to go -- in sort of set it up right sort of set you -- so there must be some kind of margin of error mechanism that you're working with here to sort of weed out the false sentiment claims.
-- we did it on several levels and it's absolutely clear that there's communities out there whose purposes simply too was spread misinformation.
Two distant it's sort of throw wrench into -- the -- of this algorithm.
But yes -- -- get yelled at him does actually it it doesn't filtered did tweets according to its political content but it does it's on the woke up from prefer.
And efforts to manipulator outcomes absolutely.
So this is still in its testing phase what type of investor is this tool using Twitter ideal for.
How you get it on to.
Mark well I -- Well I would say that it's it's not for some of the retail investor it's not for individuals like myself for someone else is gonna.
A couple thousand also invest with this is for major hedge funds -- are are specialized in for example algorithmic trading.
-- dental looking for an additional source of information to get double good hedge.
So they're all kinds of technical indicators out there already relative strength index determine overbought oversold levels resistance support levels for markets.
Do you think this Twitter predict or Twitter indicator didn't actually name for it -- -- think it'll.
Really kind of fall into.
You know are sort of toolbox for looking at markets in predicting.
It could it could very will be an and it stands by itself I mean as we showed in our research.
But I can definitely see it being included in some of the tool -- of all all of investors as just another data source but we give -- -- an idea of what was going on -- the public.
And none near real time and I think that doesn't that's particularly your interest in and that we -- little bit to the weather report right you've growing corn it's really useful to know -- -- there's going to be a storm coming.
Renault and absolutely -- Johann doctor Johann -- we're gonna keep tabs onion progress.
He has proliferate -- into the marketplace thanks again for your time and sharing your.
Filter by section