Technology
From data to the closing auction.
Athena runs on technology built in-house. Data, models, execution and surveillance form one pipeline, automated and supervised by the partners.
The cycle
Four phases, ending at the close.
The New York exchanges close at 16:00 Eastern Time. The closing auction sets the official closing price of each stock, and it is where Athena's orders are executed.
-
Monthly preparation
A monthly cycle prepares the models for the sessions ahead.
-
Before the open
The previous session is closed with official data, and the measures for each asset are updated.
-
Portfolio decision
The models produce their forecasts and the model portfolio for the day is set.
-
Execution and control
Orders are executed at the closing auction. Results are checked, and the closing report follows on the next business day.
What the models read
Every session, a set of measures is computed for each of about 180 US-listed assets.
- Volume and price, read together (volume-price analysis)
- Relative strength (RSI)
- Volatility
- Correlations between assets
How forecasts are made
Two families of machine-learning models, neural networks and random forests, produce a forecast for each asset. The forecasts feed the model portfolio.
AI agents automate the routine steps of the process and explain what the model did. They do not make investment decisions.
Supervision
Automated, and watched.
Human in control
The partners define the operating mode and follow every order submission. A general brake stops all orders from being sent, at any moment.
Automatic surveillance
The system runs under automatic surveillance, alongside the partners.
Security
Dedicated infrastructure, access restricted to the partners, and credentials separated by function.
See it applied.
Athena RCR is the first model built on this pipeline. Ask for access and a partner will walk you through it.