The Million Roll Question
Share
The Question
A statement is often made in discussions about craps strategy:
“Because of the Law of Large Numbers, you cannot determine whether a strategy is successful since it is not possible to physically roll the dice a million times.”
Is this true?
Does the inability to perform massive real-world trials prevent us from evaluating a strategy?
1. Understanding the Law of Large Numbers
The Law of Large Numbers states that as the number of trials increases, observed results will converge toward the true mathematical expectation.
In craps, each bet has a fixed expected value determined by probability. For example:
- Pass Line: approximately 1.41% house edge
- Don’t Pass: approximately 1.36% house edge
- Place 6 or 8: approximately 1.52% house edge
These numbers are not estimates from trial and error. They are derived from the combinatorial structure of the dice. Every possible outcome is known: 36 combinations per roll. From that finite structure, exact probabilities and expected values are calculated.
The Law of Large Numbers does not require physically rolling the dice a million times. It simply describes how outcomes behave over increasing repetition.
2. Expectation vs. Experience
The confusion often arises because of variance.
In the short term:
- A player may win consistently.
- A strategy may appear highly effective.
- Structural positioning may produce steady gains.
But short-term results do not override expectation. They reflect variance.
Craps contains volatility because:
- Multi-roll bets extend exposure.
- Numbers cluster naturally.
- Sevens arrive unpredictably.
- Positions interact across multiple rolls.
Variance can dominate for hundreds or even thousands of rolls. That does not invalidate the mathematics. It simply means convergence has not yet expressed itself clearly.
3. Why a Million Rolls Are Not Required
The claim that a strategy cannot be evaluated without a million physical rolls misunderstands how probability works.
A craps table does not generate infinite possibilities. It operates within a closed mathematical system:
- 36 possible dice combinations.
- Fixed payout structures.
- Known transition states between bets.
Because the system is finite, it can be modeled precisely.
Modern computation can simulate millions of rolls instantly. Even without simulation, expected value can be calculated exactly using probability trees and state transitions.
Casinos do not test strategies by physically rolling dice millions of times. They rely on combinatorics and expectation.
4. What Strategies Actually Change
While expectation remains fixed, strategy can change other important dimensions:
Volatility
Some structures reduce dramatic swings. Others increase them.
Drawdown Depth
Layering bets may cushion certain outcomes while exposing others.
Exposure Distribution
Structural positioning (for example, combining Don’t Pass with Come or Don’t Come bets) changes how rolls interact across the board.
Psychological Stability
Smoothness of results affects decision-making.
None of these alter house edge. But they alter the experience and capital behavior of play.
5. The Real Limitation
What is true is this:
In short sessions, you cannot prove a strategy’s true expectation from results alone.
A player could:
Win for months with a negative expectation system.
Lose heavily using a mathematically superior position.
Short-term outcomes are not proof of long-term edge.
But this does not mean strategies cannot be evaluated. It means they must be evaluated mathematically rather than emotionally.
6. The Deeper Question
The more meaningful inquiry is not:
“Can we roll enough times to know?”
It is:
“What is the mathematical expectation of the structure?”
And beyond that:
“What is the volatility profile of the structure?”
If a strategy uses only negative expectation wagers, it cannot overcome the house edge long-term. That is mathematically settled.
However, strategy can influence:
- Risk exposure timing
- Volatility compression
- Structural value layering
- Capital efficiency
These are strategic considerations, even within a negative expectation environment.
7. Conclusion
The inability to physically roll a million times does not prevent us from evaluating a craps strategy.
The dice operate within a finite probability space. Expected value is calculable. Variance is measurable. Risk is modelable.
The Law of Large Numbers guarantees eventual convergence toward expectation. It does not require physical demonstration to validate mathematical truth.
The real distinction is between:
- Short-term experience
- Long-term expectation
A strategy may influence the path.
It does not alter the destination.
Gus Santos