Symbiotic intelligence learns to earn on the exchange

Original article:

Fintech startup Cindicator predicts changes in the value of various financial tools based on the wisdom of the crowd and artificial intelligence. In the last six months, the value of the company share portfolio increased by 168%, and mobile app users received $24,000 as dividends.

From a user’s perspective, Cindicator is a mobile app, which can serve to answer questions regarding the price of company shares or other financial tools in the nearest future.

“On Friday, March 3, Bank of America’s shares were listed at the price of $25.44. In your opinion, what will be the maximum and minimum prices of Bank of America’s share on Monday, March 6?”

On average, some 500 to 800 people answer similar questions. Subsequently, all the answers are aggregated and processed by artificial intelligence, which determines the value of each answer and creates a single forecast, for example the range of Bank of America’s share price for the next day. Based on this forecast, the company’s robot trader makes deals on the exchange, with the subsequent distribution of part of trade gain among users.
“At the end of each month, part of our trade gain is distributed among users as dividends in proportion to the quality of their forecasts,” project co-founder and CTO Yury Lobyntsev told Hightech. “In the last two months we paid a total of $5000 to 30 best forecasters [those who make forecasts — editor’s note]. The maximum forecaster payment was $120.”

The wisdom of crowds
James Surowiecki in his “Wisdom of the Crowds,” published in early 20th century, went into detail describing the idea that quite a large and diverse group of people, which answers a certain question independently of each other, often finds the correct answer. The history of Cindicator started from this very book.

“We need to review our vision of time.”
In the summer of 2015, project co-founder and CEO Mikhail Brusov read this book and decided to put the theory into practice: to create an application which would allow people to forecast the outcomes of various events in a format of a game. He contacted the Octabrian neuro-laboratory, engaged in neuro sciences and custom innovation application development, headed by Yury Lobyntsev.

“That was a mobile game, The Vote, which published news related to finance, fashion, politics, sports, technology and art, as well as five or six outcomes regarding these events,” Lobyntsev says. “This is how we started to explore the wisdom of crowds.”
Lobyntsev says this was quite a successful experience from the research perspective, but they also had to think of the commercial aspect. He quickly developed partnership relations with Brusov, and in December 2015 they jointly registered a new company, Cindicator, focusing mostly on financial markets.

Artificial intelligence and superforecasters
The role of artificial intelligence in the entire process can be limited to clusterisation mechanisms: grouping people for various types of questions. When a new user enters a system, machine learning algorithms do not introduce his/her answers in the overall forecast in the first days, but study the accuracy of the answers. After that, they “distribute the weights of confidence,” which finally influences the end forecast (user experience of interaction with the application remains unchanged).

Some time later, project founders found that there was a certain group of people, who could be named superforecasters, as in most cases they provided more accurate forecasts than the others. According to Lobyntsev, these people account for approximately 2% of all users.
“This is a specific group of people, which is capable of more accurate forecasting. ‘Why’ is a question which has no unambiguous answer. Some rely on their competence based on experience, some are simply talented by nature,” Lobyntsev says. “Later we found that some people have the ability to forecast events merely of a certain type. It may be said that this is not a common ability, when a person is capable of forecasting everything, but a very complex thing.”

Trading at the stock exchange is not the single, nor the basic source of company income. Cindicator is currently using the B2B business model.
The company provides hedge funds, banks and private investors with access to its API trade signals in exchange for a part of income received by them.
Based on partners’ requests (currently, there are no more than 10), some questions to users are prepared and then asked via the application.
Yury Lobyntsev says that Cindicator is unparalleled on the forecast market so far. “Our activity is the hybrid, or symbiotic intelligence. This is a new phenomenon,” Yury Lobyntsev says. “This system trains both people and machines. As a result, all participants become contributors rather than merely users, while the inner mental work becomes a capital which can be estimated.”

In mid-February, Cindicator completed a joint experiment with Moscow Exchange, in the course of which the company’s robot trader generated an investment portfolio with a return of 2.8% (47% p.a.) within three weeks, based on forecasts of 863 project participants. The second stage of the experiment, when users collectively managed a portfolio of 3,000,000 roubles, was completed on March 14. An increase in profitability based on 30 deals amounted to 0.19% during three weeks.