Does the Craps 5-Count Strategy Work? 1 Million Rolls Simulated
Question
Does waiting through a shooter's early rolls before betting, the "5-Count," actually improve the odds, or does it just mean betting less often?
Rules & assumptions
The 5-Count is a shooter-vetting method: a player watches a shooter's hand and doesn't bet until the shooter has survived several rolls, the idea being to skip shooters who seven-out immediately. This simulation uses one common formalization of the count: a come-out roll that doesn't establish a point (a natural 7/11 or a craps 2/3/12) doesn't advance the count; the roll that finally establishes a point counts as roll 3; every roll after that, of any kind, adds one to the count. A player is "in" once the count reaches 5, and stays in for the remainder of that shooter's hand. The count resets to zero when the shooter sevens out and a new shooter begins.
Both strategies place $6 on 6 and $6 on 8 (7:6 payout) using the identical wager sizing and resolution rules as Place 6 & 8, Simulated. Seed 43, both strategies watching the same sequence of rolls.
Strategy A waits for the 5-Count before placing 6 and 8, then rides the bet for the rest of that shooter's hand.
Strategy B places 6 and 8 the instant every point is established, with no waiting.
The experiment
166,652 shooter hands simulated, totaling 1,000,000 rolls while Strategy B's bets were active (Strategy A's bets were active for a smaller number of those rolls, since it skips the early ones).
Results
| Metric | Strategy A (5-Count) | Strategy B (bets immediately) |
|---|---|---|
| Hands where the strategy ever bet | 117,575 of 166,652 (70.6%) | 166,652 of 166,652 (100%) |
| Total wagered | $2,584,014 | $3,668,604 |
| Net result | -$42,241 | -$52,888 |
| ROI on money wagered | -1.635% | -1.442% |
Strategy A wagered 29.6% less money than Strategy B over the identical stretch of dice, and lost 20.1% less in raw dollars as a direct result of exposing less money. But its ROI, the cost per dollar actually wagered, was not better than Strategy B's. Both numbers sit close to the same 1.52% theoretical edge on Place 6 and Place 8; the small gap between -1.635% and -1.442% here is sampling variance from Strategy A's smaller, filtered sample, not a real edge in either direction.
Why it happens
Every dice roll is independent of every roll before it. A shooter who has already rolled four times without sevening out is not more or less likely to seven out on the fifth roll than a shooter who just started; the dice have no memory of the hand's history. The 5-Count doesn't change that fact, and this simulation shows exactly why: Strategy A's per-dollar cost is the same 1.52%-ish edge as Strategy B's, because skipping the early rolls of a hand doesn't change the probability of what any individual future roll will be. What it does change is how much action gets exposed to that edge in total, since a hand that sevens out before reaching the 5-Count is a hand Strategy A never bets on at all.
Verdict
Waiting can reduce the amount of money exposed to the house edge. It does not create a mathematical advantage. Betting less often is a real, legitimate way to manage bankroll and variance, and the 5-Count does that. It is not a way to beat the 1.52% edge on Place 6 and Place 8, because no amount of watching past rolls changes what an independent future roll of the dice is going to do.
Simulation and analysis by 0Stakes. Full methodology at /methodology/.