We are a remote-first formation of experienced engineers from different countries, collaborating with each other and our customers on a B2B basis. Our approach is enjoy ourselves while crafting elegant functional engineering solutions for the financial and blockchain domains.

Mission and Goals

Our current mission is to simplify the process for both individual and institutional algo traders to validate or invalidate their trading algorithms and reduce the "time-to-market" metric. We are dedicated to providing a convenient toolset for backtesting strategies and accurate data sources for machine learning and AI models. In summer 2024, we are just at the beginning of our journey as a public project with the release of the KawBo website and our first public API GraphQL schemas.

By 2025, we aim to achieve a production-ready stage, and by 2026, we plan to reach commercial readiness by offering deployable clusters of crawling solutions with data platform for institutional traders, along with extended features-rich API interface for individual traders via our SaaS platform.

We are also providing outsourced engineering services for our customers on a contract basis. We are 100% bootstrapped project and trying to keep that way as long as it's possible, but "that way" also means that we need to fund ourselves by contracting and consulting.

If you wish to build any Erlang or Elixir project, we will be happy to serve your needs. We accept projects of any complexity and in any business domain.

Read more about our model and the types of engineering services we can provide in the engineering section.

Our Story

Our journey into trading began in 2013 with the development of our first profitable hobby trading bot for the cryptocurrency trading platform Mt.Gox.

This success was the culmination of approximately two months of testing ideas and development.

We were excited and happy - we didn't spend too much time on development and it felt like winning a Nerdy Game where programming skill and logic determined the victor, with coins as the prize. As for a hobby project, it felt like a good outcome, so we continued to play that Game on weekends, polishing algorithms - having fun and diverting from our full-time jobs.

Our bot was among the winners for approximately two to three months. Then, market conditions changed and our strategy stopped being profitable, so we had to rethink and implement the second version of the bot.

Initially, we conducted naive live tests, trading small amounts in the real markets to validate or invalidate our hypotheses. Gradually, as we developed our second strategy, the profits from our first bot dwindled away due to our live tests.

Just as we were nearing profitability with our second strategy, the Mt.Gox crash happened, marking what felt like the end of the Game. Our trading bot was abandoned to the archives, gathering dust for over a year.

....

In 2015, during a period of boredom, we decided to return to algo trading, this time on the BTC-E platform.

Adapting our original bot for BTC-E was not straightforward due to its originally naive design and differences in how order books on that market worked.

We opted to start from scratch, this time choosing Erlang as the programming language (the first bots were implemented in PHP).

As we mostly worked on that project on weekends, it took more than a year to get in and feel comfortable with Erlang and develop new bot engine. Finally, when we were close to our first Erlang-version release, BTC-E also went into the abyss, similar to Mt.Gox.

The cryptocurrency market was in a 'Wild West' stage, and the continuing disappearance of trading platforms led to the idea of a trading platform-agnostic solution - data platform where the integration with a new trading platforms would take only a week or two, and where strategies could be backtested across various markets and pairs simultaneously. This ended in the creation of the Maria project - our internal Erlang cluster of Market Crawlers. Now Maria operates under the hood of the KawBo platform, which is a GraphQL gateway to some of the functionalities of Maria.