Ttwick is a tool for measuring, tracking, and interpreting how people feel about the things that are on their minds. It is our attempt to classify, quantify and track the formation, development and spread of ideas and concepts across social media messages.
Ttwick takes social media messages and converts them into “ttwicks”. A tick is a measure of how many stocks are rising in price versus how many are declining in price: an upticking stock is one where the most recent change in price was positive, a downticking stock is one where the most recent change was negative.
Wouldn’t it be nice if we could simultaneously read every online opinion about a particular subject, right now, and determine if they are positive or negative or sad or happy or whatever feeling summarized the context? What if instead of simply tracking “trending” terms, or the opinions of a finite number of friends (via TweetDeck, for example), we could instead track the collective opinion of social media users all over the world, and even see changing trends in those opinions every second?
That would be virtually impossible to do by hand or via existing tools, because Twitter alone generates about 200 million tweets a day on average (about 2,314 tweets every second), and that number is growing. And an equally large number of opinions get posted on Facebook, blogs, Linkedin, YouTube, Google +, etc.
Enter ttwick, the combination of a social media message and a market tick via artificial intelligence (A.I.). Ttwick can “read” and interpret as many messages we care to feed it, and doesn‘t complain. On top of that, it enhances the simple message by indexing other pieces of information to it.
Our algorithms enhance a message by determining its topic, its sentiment, where it fits in relation to other ttwicks, whether it was sent by a bot or a real person, its potential to become viral, etc. Ttwick collects data on topics that interest us: a person, a song, a movie, a company, a brand. As a body of data is amassed, patterns and trends begin to form; keywords attributed to the subject begin to emerge, from very positive to very negative, and every shade of feeling in between we trained our A.I to recognize; preference patterns related to geographical locations start to become clear.
A continuous surge of negative sentiment among social media messages in Egypt, and around the terms “Mubarak”, “government” and “repression” probably pushed collective popular sentiment into negative territory, which led people to start organizing public protests that eventually led to the fall of that government. Tracking sentiment, memes, and even identifying participants and intent via A.I. could have warned of this months in advance.
This concept can be used in many industries, for example to determine the popularity of movies and performers, planning capital allocation for advertisement campaigns, measuring the impact of news stories, forecasting the success of new product launches, etc. It could even provide some form of calibration to quantitative trading models used by hedge funds and proprietary trading desks of investment banks that trade based on the Consumer Sentiment Index, behavioral economic indicators, etc.
Having spent many years in the quantification of credit, market, operational and catastrophic risks for top-tier Wall Street firms, we found a very interesting angle for our proprietary valuation models when applying them to social media.