Pro Models

Pro Models turns the raw model output into four disciplined strategies. Each profile has its own odds window, probability threshold, edge requirement, and staking guidance, and each one publishes at most a handful of selections a day.

The four profiles

Acca Hunter rejects obvious correlation between legs and publishes nothing at all when the target cannot be reached honestly. Silence is a valid output for every profile.

Reading a profile card

Verified win rate and stake ROI

Performance across the profile’s settled bets, not a projection.

Settled bets

The sample the record is built from. A small sample makes the other figures provisional, and the page says so rather than hiding it.

Avg edge

The mean gap between model probability and implied probability at signal time.

Avg CLV

How the signalled price compared with the closing price. This is the early evidence that the selection process is finding real mispricing.

Max drawdown

The worst peak-to-trough run the profile has been through.

Bets per day target

The intended volume, alongside the rate actually observed.

Active and paused markets

Markets are gated on rolling production evidence rather than being permanently on. A market is paused when its sample is too small, its recent ROI is not positive, its closing-line coverage is too low, its closing line value is negative or not statistically positive, or its evidence has gone stale. Both lists are published, with the reason attached to each paused market.

The automation rules

• Only consider approved markets that pass the rolling production-evidence gate.
• Require odds inside the profile window, model probability above the threshold, and a positive expected-value edge.
• Publish zero to three singles per profile, and never lower a threshold to fill a quota.
• Pause new signals when the rolling sample turns negative, closing-line coverage drops, or a market falls below the minimum sample rule.