Dev Update – 2nd Nov 2020

At last, we are going to start an Orbit test next week. It was a challenging journey for us. Creating machine-learning-based asset management technology demands sweat, tears, and enormous efforts. Sometimes challenging issues lead us to fall into a big disappointment; we did not know what to do to resolve the problems except keep working on it. It takes more than my expectation, but we are ready to move forward.

All software components necessary to run asset management, such as Athena, Spear, Orbit, Mercury, HTS Gateway, Data crawler, and others, are good to go. We will inject seed money into Athena, and start monitoring in the coming sprint.


Data Crawling

  • Data Crawling is a crucial component in the whole system because it provides data to Orbit and Spear. Without data, Orbit and Spear are useless. Due to intensive data crawling, data collision happens occasionally. Data source does not allow multiple access points at a time.
  • We revised data crawling scheduling logic to have only one access point at a time, so no more data collision.
  • We also updated task queuing logic for smart processing. So the logic can eliminate redundant tasks.
  • The performance as well as stability of the update are much better than the previous one.


Athena

Athena Daemon – Automated Workflow Manager

  • We have developed Athena Daemon, which manages the workflow of Athena itself, Spear, and Orbit.
  • When Athena Deamon receives a signal message, it will start the required processing one by one without human intervention.
  • It begins with retrieving data from database. And it ends with sending long and short messages to Mercury determined by Orbit’s analysis.


Mercury

Timeout management

  • Suppose an order’s position is too big or the target instrument’s liquidity is small. In that case, the order processing takes more time than usual.
  • Timeout management plays an essential role in this picture. The management logic will check the order’s status and decide whether to reorder, cancel, or do error handling.
  • We updated timeout management using state-machine. So timeout management is being centralized and easy to manage.

The below screenshot is the outcome of all software components’ integration(Athena, Spear, Orbit, Mercury, HTS Gateway, Database, Kafka, and Data Crawling.)

Get early access to the latest updates on the MoA

Get early access to the latest updates on the MoA Investment Solutions