A risk-and-profit routing engine decides, for every bet, how much a casino keeps on its own book and how much it hedges on an external market. i-play.io's Decision Engine keeps the small, safe, profitable flow in-house and automatically hedges the bets that could threaten the bankroll — protecting solvency while preserving most of the operator's margin.
Keep every bet on your own book and you earn the highest margin — but a few large wins can wipe out an under-capitalised operator. Hedge every bet externally and your risk is near zero — but you pay a fee on every dollar and give away the natural house edge. Most operators are forced to pick a side.
Keep the small, safe, profitable flow in-house. Hedge only the flow that genuinely threatens the bankroll — large bets, longshots, sharp players and bets near settlement. Per bet, automatically, in real time.
A casino's payouts are lumpy — an ordinary bad streak can exceed a small bankroll and end the business. The engine hedges the bets big enough to hurt, so a lucky player is paid by the market, not by you.
It isn't only a safety device. Among every strategy tested that never went bust, the engine was the most profitable — keeping the fee-free house edge on safe flow, and still earning a margin on hedged flow with no outcome risk.
Validated against twelve months of real prediction-market history. For a newly capitalised casino, the simulated annual probability of going bankrupt fell from 2% to 0%.
The engine weighs four signals — using only what is known before the outcome.
The larger a bet is relative to your bankroll, the more of it is hedged.
Longshots carry outsized liability and are treated more cautiously.
Bets close to settlement are hedged more tightly, countering late information.
Consistently sharp players are hedged heavily — their edge transfers to the market, not your book.
The engine was backtested on twelve months of real prediction-market history — over 23,000 markets and 12 million actual trades with their true final outcomes — across more than 200,000 Monte Carlo simulated casino-years, three bankroll sizes and six adversarial stress scenarios. The comparison covered fully hedged, fully unhedged and engine-routed strategies on identical betting flow.
0% bankruptcies held even under deliberately hostile conditions — including a tripled population of skilled bettors and volume compressed into the worst days.
In a B-book, the operator keeps a bet on its own books: it earns the full margin and the house edge, but pays winners from its own bankroll. In an A-book, the operator hedges the bet on an external market: it still earns a margin, but carries no outcome risk because a customer's win is paid by the market.
Holding everything (pure B-book) is the most profitable until a big win isn't — it can bankrupt a small operator. Hedging everything (pure A-book) is safe but leaks fees and gives away the house edge. The Decision Engine blends the two per bet, keeping what is safe and hedging what is dangerous, so the operator gets most of the B-book's profit without its risk of ruin.
For each prediction-market bet it routes part to the house book and part to a hedge on a leading external exchange, keeping safe flow in-house and offloading what threatens the bankroll — automatically, with nothing to manage per bet.
That's the failure mode it's built to prevent. Bets large enough to dent the bankroll are hedged on an external market, so a big win is paid by the market, not the operator. Across 200,000+ simulated casino-years, including adversarial stress tests, the engine-protected book never went insolvent.
Not materially. In simulation the engine kept roughly 93–98% of the profit of running fully unhedged while removing the bankruptcy risk, and earned up to about 4% more than hedging every bet — the most profitable strategy tested that never went bust.
It was backtested on twelve months of real prediction-market history — over 23,000 markets and 12 million trades with their true outcomes — across 200,000+ Monte Carlo simulated casino-years and several adversarial stress scenarios.
We'll walk qualified operators through the model and the results in a live demo.
