Strictly Increasing Dice Rolls

Probability · Easy · Free problem

You roll three fair six-sided dice one at a time. What is the probability that the three results come out in strictly increasing order -- that is, the first roll is less than the second, which is less than the third?

Hints

  1. Think about splitting the event into two parts: first, that all three dice show different values, and second, that those values happen to be in the right order.
  2. Given three distinct values, how many orderings are there, and how many are strictly increasing? Use symmetry.
  3. Compute $P(\text{all distinct}) = 1 \cdot \frac{5}{6} \cdot \frac{4}{6}$, then multiply by $\frac{1}{3!}$ since only one of the $3!$ orderings of distinct values is strictly increasing.

Worked Solution

How to Think About It: There are $6^3 = 216$ equally likely ordered outcomes. The question is: out of all those triples, how many satisfy $d_1 < d_2 < d_3$? The slick move is to notice that any set of three distinct values can be arranged in exactly $3! = 6$ orderings, and exactly one of those is strictly increasing. So the problem reduces to: how often do we get three distinct values? Multiply that by

/6$ and you are done.

Quick Estimate: Roughly, two out of six faces collide on the second roll ($P(\text{distinct so far}) \approx 5/6$), and about two more on the third ($4/6$). So the fraction with all distinct values is around $(5/6)(4/6) \approx 56\%$. One-sixth of those are in increasing order: $0.56/6 \approx 0.093$. The exact answer should be close to $5/54 \approx 0.093$.

Approach: Decompose via $P(\text{strictly increasing}) = P(\text{all distinct}) \times P(\text{increasing} \mid \text{all distinct})$.

Formal Solution:

Let $A$ be the event that all three dice show different values, and $B$ the event that they appear in strictly increasing order. Note that $B \subseteq A$ (you cannot have a strict increase with a repeated value), so:

$P(B) = P(A) \cdot P(B \mid A)$

Computing $P(A)$: Roll sequentially. The first die can be anything. The second must differ from the first -- probability $5/6$. The third must differ from both -- probability $4/6$. So:

$P(A) = 1 \cdot \frac{5}{6} \cdot \frac{4}{6} = \frac{20}{36} = \frac{5}{9}$

Computing $P(B \mid A)$: Given all three values are distinct, there are $3! = 6$ equally likely orderings of those values (by symmetry of the dice). Exactly one ordering is strictly increasing. So:

$P(B \mid A) = \frac{1}{6}$

Combining:

$P(B) = \frac{5}{9} \cdot \frac{1}{6} = \frac{5}{54}$

Answer: $\dfrac{5}{54} \approx 9.26\%$

Intuition

The key principle here is symmetry over orderings. Whenever you have $k$ distinct values drawn from any symmetric distribution (dice, uniform draws, etc.), all $k!$ orderings are equally likely. So the probability of any specific ordering -- strictly increasing, strictly decreasing, alternating -- is exactly

/k!$. This lets you avoid counting directly: just find $P(\text{all distinct})$ and multiply by the fraction of orderings that satisfy your condition.

This shows up constantly in quant interviews and in real work. Order statistics problems -- 'what is the probability the minimum exceeds $x$?', 'what is the distribution of the median?' -- often reduce to exactly this symmetry argument. Any time you see 'strictly ordered' combined with 'independent symmetric random variables,' your first move should be: find the probability of no ties, then use

/k!$. The complement approach (subtracting out ties) is messier and error-prone under time pressure.

Open the full interactive solver →