The model: gross theoretical win → minus rebate cost → net gaming revenue → net margin. Dial in your assumptions for game mix, turnover, rebate rate and session length. Figures update live.
| Game | Mix % | House edge |
|---|
The scenario
At 3× theoretical variance with 25% DOL, the casino pays out far more than it earns in a single session. The question is: how many future "normal" sessions (at 1× theoretical, no rebate triggered or rebate at lower loss) does it take to recover that deficit?
The maths has a few components:
Let me be precise. The casino receives the actual loss (3×GTW) then pays back 25% of it:
That looks profitable — and technically it is on that session. The problem inverts when you look at it from the long-run expected value perspective. The house expected to keep 1×GTW per session. It kept 2.25×GTW this time, but probability dictates a future session where the player runs hot and the casino keeps far less — or nothing, or goes negative if the player wins outright.
The real break-even question is: what if the player wins in session 1 (casino earns zero), and session 2 is 1× theoretical with rebate triggered — does the combined P&L recover? Or more practically: across a programme of visits where rebates are paid on every losing session, what visit frequency is needed to justify the acquisition cost (flights, hotel, F&B) that's bundled in?
The key insight the model reveals — which surprises most people running these programmes — is that session 1 at 3× variance is usually still net positive for the casino on the gaming line alone. The player lost 3× theoretical, the casino received all of that, then rebated 25% back. So the casino nets 2.25× GTW from gaming — better than a normal session.
The problem only materialises when you add hospitality cost. At $4,500 per trip (a standard CGB-style package), the gaming windfall easily absorbs it. At $15,000 (Tier V private jet, Royal Suite, Rolls-Royce transfers), the maths tightens considerably — at 3× variance with 35% DOL, you may be looking at a marginal session even on the bad visit.
The real break-even problem is the mirror scenario you didn’t ask about. The one that genuinely hurts operators is when session 1 is 3× variance in the other direction — the player runs hot, wins against the house, and leaves. No rebate triggers. The casino has spent $4,000–$15,000 in hospitality and has negative gaming revenue for the trip. That’s the scenario where you need 6–10 future visits to recover — and those visits never come if the player’s experience was a loss.
This is why trip-loss models anchored to minimum turnover requirements are structurally superior: the player must generate a defined volume of play before any rebate triggers, which anchors the programme to actual theoretical win rather than a single unlucky session. Get that structure right and the model is resilient across variance.