Visual Tools
Calculators
Tables
Mathematical Keyboard
Converters
Other Tools


Dice Roll






Dice Roll Probability Model


The dice roll model describes a random mechanism in which a single trial produces one outcome from a finite set of distinct possibilities.

Unlike binary models, this setting introduces randomness with multiple categories. Each outcome represents a separate case, and exactly one of them occurs on every trial.

By allowing more than two outcomes, the dice roll model provides a basic framework for studying categorical randomness, grouping of events, and the probability distributions that arise from finite multi-outcome experiments.

Key Terms

Random Experiment— rolling a die is a standard random experiment
Sample Space— Ω={1,2,3,4,5,6}\Omega = \{1,2,3,4,5,6\} for a standard die
Event— any subset of faces, e.g. rolling an even number
Equally Likely Events— each face has probability 1/61/6 for a fair die
Discrete Uniform Distribution— the distribution of a fair die roll

See All Probability Definitions →


What Is Being Modeled

A coin has two faces; a die has six. Both are a single trial that ends in exactly one result.
Rolling a die is tossing a coin with more possible results. The question is what the extra outcomes change.
Side by side, the die has more outcomes, smaller chances for each, and far more events to ask about.
fair coinfair six-sided die
possible outcomesheads, tails1, 2, 3, 4, 5, 6
chance of each outcome1/21/6
a typical eventheads: 1/2an even number, {2, 4, 6}: 3/6 = 1/2
how many different events2² = 42⁶ = 64
With six outcomes, questions like "even?", "more than 4?" or "a prime?" all make sense; the coin offers almost nothing to ask beyond which face came up.
What changes is only the number of mutually exclusive outcomes — and with it, how rich the events can be.
Each trial produces exactly one outcome, and no two outcomes can occur at the same time. The model does not depend on the physical act of rolling a die; it abstracts away all physical details and keeps only the structure of a finite multi-outcome mechanism.
What distinguishes this model from simpler ones is not the presence of numbers, but the existence of several mutually exclusive categories generated by a single random experiment.
The definition
The dice roll model is a probability model of a single trial with a fixed, finite number of mutually exclusive outcomes, each given a probability, the probabilities adding up to 11.
The dice roll model represents a single random trial with a fixed, finite number of possible outcomes.
So a fair die is a dice roll model with six outcomes of probability 16\frac{1}{6} each, and a spinner with eight equal sectors is one with eight outcomes.
It is the general finite model, and the coin toss model is its two-outcome case.
Any experiment with a list of possible results — a spinner, a draw from a bag, an answer on a survey — fits it, with the coin toss model as the smallest instance.
The outcomes need not be equally likely, and the labels need not be numbers.
A loaded die still fits the model, and so does a die with colours instead of dots. The outcomes themselves are listed in outcome space.
The rest of the page covers the outcome space, events, the probabilities, rolling more than one die, and the distributions that follow.

Outcome Space


    The outcome space of the dice roll model is a finite set containing more than two elements.

    Outcomes are often labeled using integers, such as
  • 1, 2, 3, 4, 5, 6

  • These labels serve only to distinguish outcomes.
    At this stage, they are not numerical measurements and carry no arithmetic meaning.

    The outcome space specifies what can happen in a single roll.
    Probabilities, numerical values, and interpretations are introduced only in later steps.
Sum: 2P = 1/36Sum: 3P = 1/36Sum: 4P = 1/36Sum: 5P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 3P = 1/36Sum: 4P = 1/36Sum: 5P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 4P = 1/36Sum: 5P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 9P = 1/36Sum: 5P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 9P = 1/36Sum: 10P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 9P = 1/36Sum: 10P = 1/36Sum: 11P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 9P = 1/36Sum: 10P = 1/36Sum: 11P = 1/36Sum: 12P = 1/36
Two dice: 36 ordered outcomes

The outcome space has more than two elements: 36 ordered pairs, one per cell, with the row for the first die and the column for the second. The integer labels are convenient but carry no probability by themselves. Select events and count their cells on the dice roll sample space explorer.

Pitfall: Ignoring the order of the dice
Listing two dice as 21 unordered pairs, each with probability 121\frac{1}{21}, makes "a 1 and a 2" as likely as double six. In the grid "a 1 and a 2" fills two cells and double six one, so the probabilities are 236=118\frac{2}{36} = \frac{1}{18} and 136\frac{1}{36}.

With two dice, an outcome records the face of the first die and the face of the second die, in that order. A 3 on the first die with a 5 on the second sits in a different cell from a 5 on the first die with a 3 on the second. For fair dice, each of the 36 cells carries the same probability, 136\frac{1}{36}.
Events in this model are groups of these cells.

Events


Events in the dice roll model are collections of outcomes selected from the outcome space.

Because the outcome space contains several elements, events can be formed in many different ways, including:
• single-outcome events
• grouped outcomes
• category-based events such as parity or range conditions

Events describe which outcomes are considered together for a given question.
They organize the outcome space without assigning probabilities or numerical values.

This richer event structure is one of the main differences between the dice roll model and binary models.
Sum: 2P = 1/36Sum: 3P = 1/36Sum: 4P = 1/36Sum: 5P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 3P = 1/36Sum: 4P = 1/36Sum: 5P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 4P = 1/36Sum: 5P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 9P = 1/36Sum: 5P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 9P = 1/36Sum: 10P = 1/36Sum: 6P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 9P = 1/36Sum: 10P = 1/36Sum: 11P = 1/36Sum: 7P = 1/36Sum: 8P = 1/36Sum: 9P = 1/36Sum: 10P = 1/36Sum: 11P = 1/36Sum: 12P = 1/36
Sum equals 7: six outcomes on the diagonal

The highlighted diagonal is an event: the six ordered pairs whose faces add to 7, with probability 6 over 36. Doubles, even sums and every other condition on the two faces pick out other groups of cells. Highlight them and compare their sizes on the dice roll sample space explorer.

The probability assignment gives each cell its weight.

Probability Assignment


To complete the dice roll model, a probability is assigned to each outcome in the outcome space.

AXIOM
Each outcome receives a non-negative number, and the total probability across all outcomes is equal to 1.

These probabilities are part of the model definition and must be specified explicitly.

Pitfall: Weights that do not sum to 1
For a die whose six is twice as likely as each other face, taking P(6)=26P(6) = \frac{2}{6} and 16\frac{1}{6} for the rest gives a total of 56+26=76\frac{5}{6} + \frac{2}{6} = \frac{7}{6}. With weights qq and 2q2q, 5q+2q=15q + 2q = 1, so q=17q = \frac{1}{7} and P(6)=27P(6) = \frac{2}{7}.

In many introductory settings, all outcomes are assigned the same probability, producing a symmetric or "fair" model. This symmetry, however, is an assumption rather than a requirement. Unequal probabilities lead to a different, but equally valid, dice roll model.

The probability assignment determines how likely each outcome is, independently of how the outcomes are labeled.

Assumptions


The dice roll model is defined under a small set of explicit assumptions.

• exactly one outcome occurs in each roll
• all listed outcomes are mutually exclusive and exhaustive
• the probability assigned to each outcome is fixed
• rolls are independent when the experiment is repeated

These assumptions are part of the model, not conclusions drawn from it.
Changing any of them produces a different model, even if the same outcome labels are used.
DownloadPNG2×PDF
Layout
AssumptionWhat it statesWhat violating it produces
Single outcome per rolleach roll produces exactly one outcomea fundamentally different model (e.g. multi-result mechanisms)
Mutually exclusive & exhaustiveoutcomes don't overlap and cover every possibilityoverlapping outcomes or unmodeled cases
Fixed probabilitieseach outcome has a probability set in advance, unchanging between rollstime-varying or adaptive probability models
Independence across rollsoutcomes of separate rolls don't influence each otherMarkov-chain-like models with memory between trials
AssumptionsLearn Math Class
Pitfall: Expecting a missing face to be due
After ten rolls without a six, a six seems due, so the next roll is given P(6)>16P(6) > \frac{1}{6}. Rolls are independent: the next roll still has P(6)=16P(6) = \frac{1}{6}, whatever the earlier rolls showed.

Long stretches without a given face are an ordinary part of repeated rolling. In ten rolls of a fair die, no six appears at all with probability (56)10≈0.16\left(\frac{5}{6}\right)^{10} \approx 0.16, about one time in six. Under the independence assumption, each roll in such a stretch is a new trial with the same probability assignment as the first roll.

Random Variables on the Dice Roll Model


Once the dice roll model is fixed, numerical quantities can be defined on top of it.

A random variable assigns a number to each possible outcome of a roll.
Different choices of this assignment lead to different interpretations and uses of the same model.

Common examples include:
• using the outcome label itself as a numerical value
• indicator variables for specific outcomes or groups of outcomes
• category-based mappings such as even vs odd
• transformed mappings such as thresholds or modular values

The dice roll model does not privilege any particular random variable.
It provides a base on which many different measurements can be defined.
Fact: Three dice beating each other in turn
Dice marked 2, 2, 4, 4, 9, 9 and 1, 1, 6, 6, 8, 8 and 3, 3, 5, 5, 7, 7 all have mean 5, yet the first beats the second, the second beats the third, and the third beats the first, each with probability 59\frac{5}{9}. A win is decided cell by cell on the grid, not by the averages.

The faces of a die need not carry the numbers 1 to 6: any six numbers can be written on them, and each choice is a random variable on the same six equally likely outcomes. A die marked 2, 2, 4, 4, 9, 9 has faces summing to 30, so its mean is 5, the same as a die marked 1, 1, 6, 6, 8, 8. Rolling two such dice and asking which shows the larger number compares the two random variables over the 36 cells of the grid, one pair of faces at a time.

Distributions Directly Induced by a Single Roll


When a random variable is defined on the dice roll model, its probability distribution is determined by the probabilities of the underlying outcomes.

If the random variable assigns a distinct value to each outcome, the resulting distribution places probability mass on a finite set of points. In the symmetric case, this produces a discrete uniform distribution.

Other choices of random variables may group several outcomes together or assign the same value to multiple faces. These mappings lead to categorical or degenerate distributions, depending on how outcomes are combined.

At this level, the distributions reflect the finite, multi-outcome structure of a single roll.
0.0000.0400.0800.1200.1600.200123456Value (k)Probability P(X = k)
Discrete uniform PMF, a = 1 to b = 6

The face value of a single fair die is a random variable with the discrete uniform distribution: six bars of equal height 1/6 at the values 1 to 6. Any other function of the roll, such as whether the face is even, induces a different distribution on the same model. Change the range and see the bars adjust on the PMF visualizer.

Repeated rolls build richer distributions.

Distributions Built from Repeated Dice Rolls


Richer distributions appear when the dice roll model is repeated and outcomes from multiple rolls are combined.

Each roll adds another categorical outcome, and sequences of rolls can be aggregated in different ways to define new random variables. The resulting distribution depends on how the outcomes are combined, not on the die itself.

Common constructions include:
• sums of multiple rolls
• averages of rolls
• counts of specific faces or groups of faces
• maximum or minimum over several rolls

These constructions produce a wide range of discrete distributions and provide natural links to topics such as convolution, multinomial models, and aggregation effects.
DownloadPNG2×PDF
Layout
ConstructionWhat it producesResulting shape or distribution
Sum of n rollstotal points obtained across n rollstriangular for n = 2; bell-shaped as n grows (route to the normal limit)
Count of a target facehow many times a specific face (or face group) appearsbinomial(n, p) with p equal to that face's probability
Maximum or minimum of n rollsextreme value across a sequence of rollsan order-statistic distribution skewed toward the upper (or lower) faces
Average of n rollsmean points per rollconcentrates around the expected value as n grows (law of large numbers)
Distributions Built from Repeated Dice RollsLearn Math Class
Two dice: maximum at most 3 is a 3 by 3 block of 9 cells; maximum exactly 4 is the L-shaped band of 7 cells around itsecond diefirst die1122334455661,11,21,31,41,51,62,12,22,32,42,52,63,13,23,33,43,53,64,14,24,34,44,54,65,15,25,35,45,55,66,16,26,36,46,56,6max ≤ 39 of 36 outcomesmax = 47 of 36 outcomesP(max ≤ 4) = 16/36P(max ≤ 3) = 9/36P(max = 4) = 7/36in general (2k − 1)/36The maximum of two rolls: square blocks and L-shaped bands
The maximum of two rolls on the grid

Roll two dice and record the larger face. The maximum is at most 3 exactly when both dice are at most 3: the blue 3 by 3 block, 9 cells. It is at most 4 on the 4 by 4 block, 16 cells, so it equals 4 on the amber L-shaped band between them: 16 − 9 = 7 cells, P(max = 4) = 7/36. The same subtraction gives P(max = k) = (k² − (k − 1)²)/36 = (2k − 1)/36.

Larger faces are more likely as the maximum: P(max = 6) = 11/36, P(max = 1) = 1/36.
Pitfall: Adding the chances for at least one six
Adding 16\frac{1}{6} for each die gives P(at least one six)=16+16=1236P(\text{at least one six}) = \frac{1}{6} + \frac{1}{6} = \frac{12}{36}, counting the cell (6, 6) twice. The event has 11 cells, so its probability is 1136\frac{11}{36}.

The maximum also answers questions about a single face. The larger of the two faces is 6 exactly when at least one die shows a 6, so this event is the row of the grid where the first die shows 6 together with the column where the second die shows 6. Its probability is the number of cells in that row and column, divided by 36.

Approximating Continuous Distributions


Repeated dice rolls can also be combined to produce behavior that resembles continuous randomness.

By aggregating many rolls and rescaling the results, the discrete nature of individual outcomes becomes less prominent. As the number of rolls increases, the distribution of the aggregated values smooths out and begins to resemble a continuous shape.

In this way, continuous distributions can be approached as limits of experiments built from finite, multi-outcome randomness. The dice roll model therefore serves not only as a discrete model, but also as a practical foundation for simulation and approximation of continuous probabilistic behavior.

Why the Dice Roll Model Matters


The dice roll model is the simplest probability model that captures randomness with more than two possible outcomes.

It introduces genuinely categorical behavior, where outcomes are not ordered by default and where events naturally involve grouping and comparison rather than simple success–failure logic.

Fact: No fixed toss count gives a fair die
Each sequence of kk fair coin tosses has probability 12k\frac{1}{2^k}, so any rule that turns kk tosses into a face gives it a probability m2k\frac{m}{2^k}, never 16\frac{1}{6}. Tossing three coins and starting again on two of the eight results works, but with no upper limit on the number of tosses.

Because of this, the dice roll model forms a bridge between binary models and more general discrete models. It provides a foundation for understanding categorical data, aggregation of outcomes, and the construction of complex probability distributions from finite sources of randomness.

Variations of the Dice Roll Model


The basic dice roll model can be modified in several ways while preserving its finite multi-outcome structure.

Common variations include:
• assigning unequal probabilities to outcomes (biased dice)
• changing the number of possible outcomes (dice with different numbers of faces)
• conditioning outcomes on previous rolls
• introducing stopping rules based on observed results

Each variation defines a new probability model.
Although the outcome labels may look similar, the underlying assumptions and probability assignments are different.

What This Model Is Used For


The dice roll model is used whenever a situation involves a finite set of distinct categories produced by a single random mechanism.

Typical uses include:
• modeling categorical outcomes
• sampling from finite populations
• constructing and combining discrete distributions
• studying aggregation effects across repeated trials
• providing a foundation for simulation and Monte Carlo methods

Because it captures finite multi-outcome randomness in its simplest form, the dice roll model appears widely in probability theory, statistics, and applied modeling.

Dice Roll Model at a Glance

The page has built up the dice roll model layer by layer — from its finite outcome space and event structure through the probability assignment and modeling assumptions, then onto random variables and the distributions that single and repeated rolls induce. The table below collects these threads into a single reference card, pairing each aspect of the model with its concise statement and a concrete example.
DownloadPNG2×PDF
Layout
AspectStatementExample
What it representsa single random trial with a fixed, finite set of mutually exclusive outcomesa standard six-sided die
Outcome spacea finite set of labels distinguishing outcomes; no arithmetic meaning at this stageΩ = {1, 2, 3, 4, 5, 6}
Eventssubsets of the outcome space — single outcomes, groupings, or category-based selections“even result” = {2, 4, 6}
Probability assignmentnon-negative weights on each outcome that sum to 1fair die: P(k) = 1/6 for every face
Modeling assumptionsone outcome per roll, mutually exclusive & exhaustive, fixed probabilities, independent rolls“a fair die rolled repeatedly under identical conditions”
Random variablesnumerical mappings on outcomes; different choices yield different measurementsX = face value; Y = 1even_{{even}}; Z = face mod 3
Distribution from one rollfinite-support discrete law; uniform when the model is symmetricdiscrete uniform on {1, …, 6}
Distributions from repeated rollssums, counts, extrema, averages — aggregation produces richer shapessum of 2 dice → triangular on {2, 3, …, 12}
Dice Roll Model at a GlanceLearn Math Class

Common Pitfalls

Each of the four mistakes below is marked where it happens in the sections above; here each is worked through in full, with the wrong and the right working side by side.
DownloadPNG2×PDF
1Ignoring the order of the dice
Two dice are rolled, and the outcomes are listed as unordered pairs of faces, so that a 1 with a 2 is one outcome however the dice fall. Each of these pairs is then given the same probability.
Wrong turn
Error here
There are 66 doubles and 1515 pairs of different faces, 2121 in all, so
P(a 1 and a 2)=121,P(\text{a 1 and a 2}) = \frac{1}{21},

P(double six)=121P(\text{double six}) = \frac{1}{21}
Right path
The equally likely outcomes are the 3636 ordered cells. "A 1 and a 2" is the two cells (1, 2) and (2, 1); double six is the single cell (6, 6).
P(a 1 and a 2)=236=118,P(\text{a 1 and a 2}) = \frac{2}{36} = \frac{1}{18},

P(double six)=136P(\text{double six}) = \frac{1}{36}
Why it happensThe two dice look alike, so (1, 2) and (2, 1) seem to be the same result. Physically they are still two dice, and each way of landing is a separate outcome.
How to checkColour the dice, one red and one blue. Every unordered pair of different faces then splits into two distinguishable results, giving 6+2⋅15=366 + 2 \cdot 15 = 36 outcomes, and only these 3636 are equally likely.
2Weights that do not sum to 1
A biased die is described in words, for example "a six is twice as likely as each other face", and the probabilities are written down by adjusting the fair values 16\frac{1}{6} without checking the total.
Wrong turn
Error here
P(6)=2⋅16=26,P(6) = 2 \cdot \frac{1}{6} = \frac{2}{6},

P(1)=⋯=P(5)=16P(1) = \dots = P(5) = \frac{1}{6}
total=5⋅16+26=76\text{total} = 5 \cdot \frac{1}{6} + \frac{2}{6} = \frac{7}{6}
Right path
Give each of the faces 1 to 5 the weight qq and the six the weight 2q2q, and require the total to be 11.
5q+2q=15q + 2q = 1
  ⟹  q=17\implies q = \frac{1}{7}
P(1)=⋯=P(5)=17,P(1) = \dots = P(5) = \frac{1}{7},

P(6)=27P(6) = \frac{2}{7}
Why it happensThe fair value 16\frac{1}{6} is so familiar that it is kept for the unchanged faces. Making one face more likely must make the others less likely, because the total is fixed at 11.
How to checkAdd all six probabilities before using them. Here 5⋅17+27=15 \cdot \frac{1}{7} + \frac{2}{7} = 1, and the ratio P(6):P(1)=2:1P(6) : P(1) = 2 : 1 is the one described.
3Expecting a missing face to be due
A face has not appeared for many rolls, and its probability on the next roll is believed to have risen.
Wrong turn
Error here
After ten rolls without a six,
P(6 on roll 11)>16P(6 \text{ on roll 11}) > \frac{1}{6}
Right path
Rolls are independent, so roll 11 has the same assignment as every other roll.
P(6 on roll 11)=16P(6 \text{ on roll 11}) = \frac{1}{6}
A stretch of ten rolls without a six is not rare: its probability is (56)10≈0.16\left(\frac{5}{6}\right)^{10} \approx 0.16.
Why it happensOver many rolls each face appears about one time in six, and this is read as the die making up for a shortage. The proportions settle because later rolls far outnumber the stretch, not because the die compensates for it.
How to checkCompare two sequences of eleven rolls that agree in the first ten and differ only in the last roll. Each has probability (16)11\left(\frac{1}{6}\right)^{11} for a fair die, so the earlier rolls give no reason to prefer any face next.
4Adding the chances for at least one six
The probability that at least one of two dice shows a six is found by adding the probability of a six on each die.
Wrong turn
Error here
P(at least one six)=16+16P(\text{at least one six}) = \frac{1}{6} + \frac{1}{6}

=1236=13= \frac{12}{36} = \frac{1}{3}
Right path
The only way to have no six is that neither die shows one, with probability (56)2=2536\left(\frac{5}{6}\right)^2 = \frac{25}{36}.
P(at least one six)=1−2536=1136P(\text{at least one six}) = 1 - \frac{25}{36} = \frac{11}{36}
On the grid this is the row of the first die's 6 and the column of the second die's 6: 6+6−1=116 + 6 - 1 = 11 cells.
Why it happensProbabilities of events that cannot happen together do add, and "six on the first die" and "six on the second die" look like separate events. They can happen together, in the cell (6, 6), which the sum counts twice.
How to checkTry more dice. With seven dice the same addition gives 76\frac{7}{6}, more than 11, which no probability can be; the complement gives 1−(56)7≈0.721 - \left(\frac{5}{6}\right)^7 \approx 0.72.
Common pitfalls with dice rolls·/probability/models/dice-rollLearn Math Class
Each slip builds the model of the dice wrongly. Two dice give ordered pairs, 3636 of them; the weights of a biased die must add up to 11; a face that has not appeared is not due; and at least one six has probability 1−(56)21 - \left(\frac{5}{6}\right)^2, not 16+16\frac{1}{6} + \frac{1}{6}. Colouring the dice, adding the weights, or trying more dice catches each slip.

Skills You Should Learn

Each skill opens its step-by-step method page. Skills marked "On this page" are taught above.
Taught on this page
More probability methods

Dice Roll Model FAQ

How is the dice roll model different from the coin toss model?

+
More than two outcomes, and that changes the kind of reasoning available. With two outcomes everything reduces to success-or-failure; with six, events involve grouping and comparison — “even”, “greater than four”, “not a one”. The outcomes are also categories rather than an ordered scale by default. It is the bridge from binary models to general discrete ones.Read more →

Why isn't the sum of two dice uniform when each die is?

+
Because the sum can be reached in different numbers of ways. There is one way to make 2 and one to make 12, but six ways to make 7, so 7 is six times as likely. The uniformity belongs to the individual die; the distribution of the sum depends on how outcomes are combined, not on the die itself.Read more →

Are the numbers on a die ordered, or just labels?

+
By default they are labels. The model treats the six outcomes as mutually exclusive categories, and nothing in it requires 3 to sit between 2 and 4. Order enters only when you define a random variable that uses it — a maximum, a sum, a comparison. The faces could be colors and the model would be unchanged.Read more →

How do you assign probabilities to a loaded die?

+
Turn the description into weights and make them add up to 11. For a die whose six is twice as likely as each other face, give the faces 11 to 55 the weight qq and the six 2q2q; then 5q+2q=15q + 2q = 1, so q=17q = \frac{1}{7} and P(6)=27P(6) = \frac{2}{7}. Keeping 16\frac{1}{6} for the other faces and doubling the six instead gives a total of 76\frac{7}{6}, which is not a valid model.Read more →

Can three dice each beat the next one in turn?

+
Yes. Dice marked 2, 2, 4, 4, 9, 9 and 1, 1, 6, 6, 8, 8 and 3, 3, 5, 5, 7, 7 all have mean 55, yet the first beats the second, the second beats the third, and the third beats the first, each with probability 59\frac{5}{9}. A win is decided pair by pair over the 3636 cells of the grid, and 2020 of them favour the winner each time, so the averages say nothing about who wins.Read more →

Can you simulate a fair die with coin tosses?

+
Yes, but not with a fixed number of tosses. Every sequence of kk fair tosses has probability 12k\frac{1}{2^k}, so any rule turning kk tosses into a face gives it a probability m2k\frac{m}{2^k}, never 16\frac{1}{6}. Tossing three coins, matching six of the eight results to the faces and starting again on the other two does work, at 44 tosses on average, but no upper limit can be put on the number of tosses.Read more →

What is the distribution of the larger of two dice?

+
P(max⁡=k)=2k−136P(\max = k) = \frac{2k - 1}{36} for k=1,…,6k = 1, \ldots, 6. The maximum is at most kk exactly when both dice are, which is a kk by kk block of k2k^2 cells; removing the (k−1)2(k - 1)^2 cells where it is at most k−1k - 1 leaves 2k−12k - 1. Larger faces are therefore more likely: P(max⁡=1)=136P(\max = 1) = \frac{1}{36}, while P(max⁡=6)=1136P(\max = 6) = \frac{11}{36}, the same as the chance of at least one six.Read more →