Our AI Method

How we engineer the model edge.

Data lake

We ingest domestic and international football data: xG, travel, rest, weather, officiating profiles and player news.

Model stack

Gradient boosted trees, neural nets and Bayesian sims battle-test each edge. We only publish consensus plays above a 3% margin.

Human guardrails

Our analysts audit every pick for lineup volatility, market liquidity and responsible staking alignment.

Automation workflow

Daily publishing starts with a Slack approval flow before syncing to the member dashboard and Telegram bot.

Daily publishing routine

  1. Morning: models retrain with the latest lines and injury updates.
  2. Analyst review: plays must clear our risk checklist before posting.
  3. Member post: tips go live on My Account with push alerts for Pro members.
  4. Nightly: results logged with ROI, CLV and variance notes.