Aria Back to Aria →
ARIA ROADMAP

One match.
Four different reads.

Aria is the first engine. The longer-term plan is a family of models that read the same match at different points, each built for a specific job rather than trying to make one model do everything.

Aria is live in development. The other engines stay behind the curtain until the data says they are ready.
LATE PHASE ENGINE · 70–80 MINUTES

Isla

Isla is the later-game read. She takes over when the match has moved beyond half-time and the information state has changed materially.

  • Built from the 70-minute snapshot archive.
  • Focused on what the next phase of the match is likely to produce.
  • Designed to complement Aria, not duplicate her.
IN DEVELOPMENT
LAST 10 MINUTES · GOAL / NO GOAL

Xena

Xena is deliberately narrow: the final ten minutes, reduced to one question. Goal or no goal.

  • Binary output rather than a broad market screen.
  • Built for the 80th-minute information state.
  • Fast, late and intentionally specialised.
PLANNED
CORRECT SCORE ENGINE · HALF-TIME

Ange

Ange turns the half-time read into a correct-score view, presenting the leading second-half score outcomes and their fair prices.

  • Top score outcomes rather than a single prediction.
  • Built from the same half-time information state as Aria.
  • Designed for dutching and price comparison.
PLANNED
ARIA TOOLS

Software around the models.

The engines are only one part of the roadmap. We also want to build tools that make operating them faster, safer and less repetitive without hiding the underlying logic from the user.

Multi-game scannerRun several eligible matches through the cockpit workflow without processing them one by one.
Offline backup cockpitA local version of the core calculation so a website outage does not automatically stop the trading day.
Multi-engine bankrollOnce multiple engines are live, split one total bank across Aria, Isla and Xena before each engine applies its own staking logic.
No launch dates for the sake of having launch dates.

Each engine moves forward when its dataset and blind testing justify it. The roadmap is a direction of travel, not a promise that unfinished models will be rushed out because a date was printed on a website.