What Question Types Do Live-Camera Betting Markets Use?
Live-camera betting markets use five question shapes: RACE, FIRST TO, COUNT, HAPPEN and STATE AT CLOSE. Lukk, by Adkuu, runs them as YES/NO or multiple-choice markets.
Live-camera betting markets on Lukk, by Adkuu, use five question shapes: RACE, FIRST TO, COUNT, HAPPEN and STATE AT CLOSE. Each shape is a template the AI fills from what the vision layer sees on the live camera, written as a natural-language question anyone can answer at a glance, and offered either as a YES/NO market or as a multiple-choice market, some of which carry a "Neither" option. Counting games such as 155.io's Rush Hour fix a single shape (how many objects cross a tracking zone, bet Over, Under, Range or Exact); Lukk's choice is to vary the shape with the scene.
The five shapes, with example questions
| Shape | What it asks | Example question on Lukk |
|---|---|---|
| RACE | Whether one thing happens before another | "Will a bus cross the junction before the light turns red?" |
| FIRST TO | Which of several named candidates does something first | "Which lands first — a cardinal, a blue jay, or neither?" |
| COUNT | Whether a count of tracked objects reaches a threshold in a window | "Will 10 or more people be on the crossing in 120 s?" |
| HAPPEN | Whether a defined event occurs at all before the window ends | "Will a gondola pass under the bridge before any motorboat?" and "Will a penguin dive into the pool in the next 60 s?" |
| STATE AT CLOSE | What the scene looks like at the moment the window closes | "Will the crane be lifting at zero?" |
The assignment of each example to a shape follows the wording of the question; the shape itself is what the frozen rule encodes. Windows are short, a minute or two, with a round lasting about 90 seconds and a new market every minute or two, so each question is sized to be answerable before the next scene.
YES/NO versus multiple choice, and what "Neither" means
Most Lukk markets are YES/NO: the player backs one side at a locked price, and payout is stake × odds, exactly, with odds up to ×20. RACE questions are naturally multiple choice, and some carry a "Neither" option for the case in which none of the named candidates does the thing before the window closes. "Neither" is a real outcome that pays whoever backed it. It is not the same as a void: a void happens only when the camera cannot tell, and a void is always a full refund of every stake on the market, never a re-grade.
For the operator, the two forms arrive the same way in the B2B feed, with fair probabilities and margined fixed odds per outcome, a frozen rule_hash, timing, status and limits, so a multiple-choice market needs no special handling beyond displaying more than two outcomes.
How each shape is settled
Every shape is settled the same way at the top level: the rule is frozen and hashed before the first bet, the deciding frames are locked, signed (ed25519) and published with the result, and an evidence replay sits behind the result. What differs is which judge applies the rule:
- A deterministic counter or tracker suits shapes that reduce to counting or ordering tracked objects. COUNT is a count against a threshold; RACE and FIRST TO are orderings of tracked objects against a line or against each other.
- A vision judge suits shapes that need a reading of the scene rather than a count. STATE AT CLOSE asks what is true in the final frame; HAPPEN asks whether a defined event, such as a dive, occurred.
In every case the judge applies the frozen rule; it does not choose it after the window closes. Settlement arrives within seconds of close, with a short verification hold on larger payouts, and the operator receives each ticket as won, lost or refunded. The full chain is described in How are live-camera betting markets settled?.
Why shape variety matters
Engagement. A live scene offers different things at different moments. A junction at rush hour supports a COUNT; a bird feeder supports a FIRST TO; a container port supports a STATE AT CLOSE on a crane. Because the AI writes the question from what it sees, the market fits the scene instead of forcing the scene into one template, and two consecutive rounds on the same channel rarely ask the same thing. This is the same shift that sports micro-betting made from match results to next-play markets, applied to a live camera rather than a sports feed.
Integrity. Variety is also a defence. A fixed market shape on a known camera gives a would-be manipulator a target to study; a question that changes shape with every rotating scene gives them nothing to plan for. Combined with bets closing before the window begins, a one-to-three-second acceptance delay with an in-flight veto from the tracker, price protection and per-bet stake caps, rigging becomes uneconomic, and every resolution is logged and replayable regardless of shape.
Pricing. Each shape is priced live from the scene, not from a payout table. A RACE with three candidates and "Neither" carries four prices that move with the birds in frame; a STATE AT CLOSE moves as the close approaches. How those prices are formed, and how the operator's margin is added, is covered in What is an AI-written betting market?.
The five shapes can be watched in rotation on the live demo at play.adkuu.com/lukk, which runs with play money on a test network, for adults (18+) only; nothing can be won or cashed out there.
Sources
- Hub88 chosen as exclusive aggregation partner for 155.io's breakthrough CCTV game Rush Hour
- Rush Hour: live traffic betting reshapes the iGaming market
- AI-driven micro-markets: Betting on moments, not matches
Last verified: October 10, 2026.