Dragon Bonus Push Rule: How Is Push Decided?
The push decision on a Dragon Bonus wager is not a courtesy call from the dealer — it is a hard-coded branch inside the hand evaluator that fires the instant the player and banker totals land on the same natural value. Strip away the felt and the chips, and what you are auditing is a tie-resolution subroutine that compares two integers and routes the wager to a neutral outcome instead of a loss. That single comparison is where most players get ambushed, because they assume a tie means a loss. It does not. It means the round never resolved.
Here is the mechanism in cold terms. The Dragon Bonus side bet pays on a natural win or a win by a margin of four or more points, with the payout scaling off the margin — nine points pays the maximum, four points pays the floor. The push condition sits outside that payout ladder entirely. When the player hand and banker hand finish with identical totals, the margin is zero, the payout multiplier table has no entry for zero, and the settlement engine returns the stake untouched. No commission, no rake, no house edge bite on that specific round. 🎯
The Zero-Margin Exception Nobody Prices In
Auditors who actually crack open the math find the same thing every time: the push is a structural exception, not a rule variant. The margin computation is total difference, absolute value, and the payout matrix is indexed from four to nine. Zero margin falls through every index. A competent Dragon Bonus push rule implementation must intercept that fall-through before the loss handler grabs it, otherwise the engine silently converts a legitimate push into a player loss. I have seen this bug in cheap firmware more times than I care to count — the tie lands, the margin is zero, and the loss branch fires because nobody wrote the guard clause.
The correct decision logic reads like this: evaluate both hands, compare totals, if equal then return stake, else compute absolute margin, if margin below four then lose, else pay according to the ladder. Order matters. The equality check sits first in the chain. Any implementation that computes margin before checking equality is fundamentally broken and will misgrade every tie as a sub-four loss. That is not a rounding error — that is a systematic bleed on the player side that compiles into a fat house edge over volume.
Why the Push Fires Before the Margin Ladder
Think about the state machine. A Dragon Bonus wager has exactly three terminal states: win, lose, push. The margin ladder only governs the win state and the loss state. The push state is triggered by a separate predicate — total equality — and it must be evaluated on the raw totals, not on the difference between them. The reason is brutal and simple: a difference of zero is ambiguous. It could mean a tie, or it could mean a data fault where both totals failed to populate. The equality predicate disambiguates. 🃏
In rigged or sloppy hardware, the tie is sometimes resolved by a coin-flip generator instead of a deterministic equality check. That is the tell. A legitimate push resolution is deterministic — same totals in, same push out, every single time. If you ever see variance on identical total pairs across shoes, the RNG has leaked into the settlement layer and the game is no longer auditable. Determinism is the whole ballgame here.
The Tie Tax and the Margin Floor
The margin floor of four points is the other lever that interacts with the push. A margin of one, two, or three points is a loss — the stake is swept. A margin of zero is a push — the stake is returned. So the boundary between loss and push is razor thin and entirely dependent on which predicate fires first. Get the ordering wrong and you have built a machine that taxes ties. Get it right and the tie is neutral, which is exactly the mathematically fair treatment given the payout structure.
What the sharp money watches for is the frequency of pushes against expected tie probability. In a fair eight-deck shoe the tie rate hovers near nine and a half percent, and every one of those rounds should register as a push on the Dragon Bonus line. If the observed push rate comes in materially below that band, the settlement code is leaking ties into the loss column. That gap is the audit signature — the difference between the theoretical tie probability and the logged push frequency is your smoking gun, and it should sit under a percentage point on any honest build.
The push is decided by a single equality predicate evaluated on raw totals before the margin ladder ever runs, and the entire integrity of the side bet collapses the moment that ordering inverts. ⚙️