Atama.AI

We are kind of factory that can produce AI-based trading algorithms that are able to predict the next market move with reasonable probability.

For making predictions of market entry points (when to buy or sell), we train neural networks by using historical market data with trading indicators for this.

Since neural networks are not designed to improve profitability, we have built a software ecosystem around Tensorflow that allows finding effective trading models. We are currently able to run 3 000 000 experiments a day.

We have been researching and developing for almost two years, during this time we have improved the training and model selection process.
We have also cooperated with the Dutch Rockstart accelerator, which made it possible to develop our unique technology further.

100T is the total amount of assets under management, globally.
From our point the market is basically unlimited, we are only limited to amount of trading capital and ability to scale the software.
For the same reason competition at this point is also not a restraint on our performance.
We have amazing growth potential.

The average daily trading volume of financial assets is 190 billion. And each time the deal is made someone buys believing the price will go up and someone sells because he believes the price will go down…
The problem is no trader can see the future.

You always have to make decision without knowing what happens next.
For example, do you think the Tesla stock price will go up and you should buy now or you think coronavirus will cause long term economic recession and stock will go down?
It’s impossible to know for sure.
But to trade profitably you only need predicting right more often than not.

And that's exactly what we are doing.
We are kind of factory that can produce AI-based trading algorithms that are able to predict the next market move with reasonable probability.
Put simply, we are a factory producing money-making machines.
The process looks like this: first, we get and prepare data, and configure experiment.
Then we train AI and run millions of simulations and then we get algorithms. This process takes now about a week for a particular asset.
It works for all assets. It’s highly scalable and can be customized for bear and bull markets.

-

The unique technology of prediction whether the asset price will go up or down with some reasonable probability.

-

-

Therefore our nearest goal is to build a proof that will be convincing to everybody, which is to make a profit from trading, and to patent our technology. To do this we need to build the trading software, trade for at least few months and file for a patent on our methodology.
To accomplish this we look for $200 000.
This money will be spent on developing trading software and our first trading capital.
We will also continuously work on our software and research and file patent application.

We are actively working on our product, but somewhere in the other corner of the world, they (newcomers) are also working on it. So, we want to be faster.

Participation in the Rockstart Artificial Intelligence Program

The winner of A Fast Track To Join The Final AI Selection Process, Lviv IT Arena