Probability Simulations — I can design and run a simulation to estimate probability and interpret the accuracy of the estimate. - Grade 9 Math

SECTION 1

Historical Context & Motivation

Humans have been grappling with chance for thousands of years, from ancient dice games to modern weather forecasts. For a long time, mathematicians could only calculate probability for simple, well-defined scenarios like coin flips and card draws. But what happens when a real-world problem is too complicated for a neat formula? That is exactly where probability simulations come in — they let us run experiments, collect data, and estimate probabilities that would otherwise be extremely difficult to compute.

The idea of using repeated random trials to solve problems grew from the work of gamblers, scientists, and eventually computer pioneers. The timeline below highlights key moments that shaped the simulation approach we use today.

1654 • The Birth of Probability Theory

Blaise Pascal and Pierre de Fermat exchanged letters about gambling problems, laying the mathematical groundwork for probability. Their correspondence established the rules that simulations would later test.

1777 • Buffon's Needle Experiment

The Comte de Buffon proposed dropping a needle onto parallel lines to estimate π. This is one of the earliest examples of using a physical random experiment — essentially a simulation — to approximate a mathematical constant.

1946 • The Monte Carlo Method

Stanislaw Ulam and John von Neumann developed the Monte Carlo method while working on nuclear physics at Los Alamos. They used random number generation on early computers to simulate neutron behavior, establishing simulation as a serious scientific tool.

2000s • Simulations Go Mainstream

With powerful personal computers and free software, simulations became accessible to students and professionals alike. Today, fields from finance to medicine rely on probability simulations to make decisions under uncertainty.

The central question this lesson addresses is: How can we design a random experiment to estimate the probability of an event, and how do we know whether our estimate is accurate? By the end, you will be able to set up a simulation, run it, and evaluate how close your experimental estimate is to the true probability.

SECTION 2

Core Principles & Definitions

Before we build a simulation, we need to understand the vocabulary and logic behind it. A probability simulation is a model that uses random processes — like flipping coins, rolling dice, or generating random numbers — to mimic a real-world event. The goal is to repeat the random process many times and track how often our target outcome occurs.

Simulation

A model that imitates a real-world process using random events. Each run of the simulation is called a trial. The random device (coin, die, random number generator) must match the probability structure of the actual event.

Theoretical vs. Experimental Probability

Theoretical probability is calculated using logic and formulas: P(event) = favorable outcomes ÷ total outcomes. Experimental probability is estimated by running trials and dividing successes by the total number of trials.

Law of Large Numbers

As the number of trials increases, the experimental probability tends to get closer to the theoretical probability. This foundational principle explains why more trials produce more reliable estimates.

Random Number Generator (RNG)

A tool — physical or digital — that produces outcomes with equal likelihood. Examples include fair coins, fair dice, spinner wheels, and computer-generated random numbers. The RNG is the engine of every simulation.

Accuracy of the Estimate

The closeness of the experimental probability to the true probability. Accuracy improves with more trials and can be assessed by comparing results across multiple sets of trials or by examining how the running estimate stabilizes.

✦

KEY TAKEAWAY

Think of a simulation like a dress rehearsal for a real event. If you wanted to know how often a basketball player makes two free throws in a row, you could watch 10 games — or you could flip a weighted coin hundreds of times (where the coin's heads-probability matches the player's shooting percentage). The more "rehearsals" you run, the more your estimate resembles reality. That is the Law of Large Numbers in action.

SECTION 3

Visual Explanation — Convergence of Experimental Probability

The diagram below illustrates one of the most important ideas in simulation: as you increase the number of trials, your experimental probability bounces around at first but gradually settles closer and closer to the theoretical probability. The graph shows a simulation of flipping a fair coin, tracking the running proportion of heads.

The purple line tracks the running proportion of heads as trials accumulate. Notice how it swings wildly with few trials (left side) but settles near the dashed cyan line (P = 0.50) as the number of trials grows (right side). This stabilization is the Law of Large Numbers at work.

The graph reveals a critical lesson: a small number of trials can produce misleading results. After just one flip you might see 100 % heads, but after 500 flips the proportion is almost exactly 0.50. When you design a simulation, plan for enough trials so the estimate has time to stabilize. In most classroom settings, 50 to 100 trials is a reasonable starting point, but professional simulations often use thousands or even millions.

SECTION 4

Mathematical Framework

Even though simulations are hands-on experiments, a bit of mathematical notation helps us describe what is happening precisely. Two key formulas anchor this lesson: one for experimental probability and one for measuring how far our estimate might be from the true value.

EXPERIMENTAL PROBABILITY

P̂(event) = number of successes ÷ number of trials

P̂ (read "P-hat") denotes the experimental (or empirical) probability. "Successes" are the trials where the event of interest occurred. As the number of trials increases, P̂ approaches the theoretical probability P.

THEORETICAL PROBABILITY (EQUALLY LIKELY OUTCOMES)

P(event) = favorable outcomes ÷ total possible outcomes

This formula applies when every outcome in the sample space is equally likely, such as rolling a fair die or drawing from a well-shuffled deck. When the situation is too complex for this formula, we turn to simulation.

RELATIVE ERROR

Relative Error = |P̂ − P| ÷ P × 100%

This measures the percent deviation of the experimental estimate from the theoretical value. A smaller relative error means a more accurate simulation. For example, if P = 0.50 and P̂ = 0.48, the relative error is |0.48 − 0.50| ÷ 0.50 × 100% = 4%.

💡

When Do We Use Simulation Instead of a Formula?

Simulations are most useful when the theoretical probability is hard to calculate — for example, "What is the probability that at least 3 out of 5 randomly chosen students share the same birth month?" Rather than computing a complicated combinatorics problem, you can simulate it hundreds of times and count.

SECTION 5

Designing a Simulation — Step by Step

Designing a simulation requires careful planning. You need to choose a random device that matches the probability structure of the real event, define what counts as a trial, decide how many trials to run, and interpret the results. The flowchart below walks through the design process.

Follow these five steps every time you build a simulation. The most critical step is Step 3: the random device must faithfully represent the probabilities of the real event. For instance, use a coin for 50/50 events but a die for events with six equally likely outcomes.

Let's look at a concrete example of how to match a random device to a problem. Suppose a baseball player has a batting average of .300, meaning he gets a hit about 30% of the time. To simulate one at-bat, you could use a random number generator producing integers 1 through 10. If the number is 1, 2, or 3, it represents a hit (3 out of 10 = 30%). Numbers 4 through 10 represent outs. Each at-bat is one trial, and a "success" is a hit.

Real-World Event	Good Random Device	How to Assign Outcomes
50% chance of rain	Fair coin	Heads = rain, Tails = no rain
1 in 6 chance of winning a prize	Standard six-sided die	Roll a 1 = win, Roll 2–6 = no win
70% free-throw success rate	RNG: integers 1–10	1–7 = make, 8–10 = miss
25% chance of selecting type A	Deck of cards (one suit)	Spades = type A, other suits = not A

Matching random devices to real-world probabilities

SECTION 6

Worked Example — Simulating Free Throws

A basketball player makes 60% of her free throws. Her coach wants to estimate the probability that she makes both shots in a two-shot foul situation. Let's design and run a simulation to estimate this probability, then compare it to the theoretical answer.

Estimating the Probability of Making Both Free Throws

1
Step 1 — State the Problem

What is the probability that the player makes both free throws, given that each shot independently has a 60% chance of going in?
2
Step 2 — Define a Trial and Success

One trial consists of simulating two consecutive free throws. A "success" for the trial is both shots going in.
3
Step 3 — Choose a Random Device

Use a random number generator producing integers 1 through 10. Numbers 1–6 represent a made shot (60%), and numbers 7–10 represent a miss (40%). For each trial, generate two random numbers.
4
Step 4 — Run 20 Trials (sample data)

Here is a sample of 20 trials. Each pair of random numbers represents two free throws: (3, 5) ✓ | (8, 2) ✗ | (1, 4) ✓ | (6, 9) ✗ | (2, 3) ✓ | (7, 1) ✗ | (4, 6) ✓ | (5, 5) ✓ | (10, 3) ✗ | (2, 8) ✗ | (1, 1) ✓ | (3, 7) ✗ | (6, 6) ✓ | (9, 4) ✗ | (4, 2) ✓ | (5, 10) ✗ | (3, 3) ✓ | (7, 8) ✗ | (2, 6) ✓ | (1, 5) ✓ ✓ means both shots made (both numbers ≤ 6). Count of successes = 11 out of 20 trials.

11 successes out of 20 trials
5
Step 5 — Calculate Experimental Probability

P̂(both made) = 11 ÷ 20 = 0.55, or 55%.

P̂ = 0.55 (55%)
6
Step 6 — Compare to Theoretical Probability

Since each shot is independent, the theoretical probability is P(both) = 0.60 × 0.60 = 0.36 (36%). Our estimate of 55% from only 20 trials is not very close. The relative error is |0.55 − 0.36| ÷ 0.36 × 100% ≈ 52.8%. This large error tells us that 20 trials is not enough for a reliable estimate.

Theoretical P = 0.36 (36%). Relative error ≈ 52.8% — need more trials!
7
Step 7 — Scale Up (500 trials with a computer)

Running the same simulation with 500 trials on a computer yields 187 successes: P̂ = 187 ÷ 500 = 0.374. Now the relative error is |0.374 − 0.36| ÷ 0.36 × 100% ≈ 3.9%. This is much more accurate, illustrating why more trials matter.

500 trials → P̂ ≈ 0.374, relative error ≈ 3.9%

SECTION 7

Strengths and Limitations of Simulations

Probability simulations are powerful tools, but they are not perfect. Understanding when to use them — and when a formula is better — is an important part of statistical thinking. The table below contrasts the main strengths and limitations.

Strengths	Limitations
Can estimate probabilities for complex problems where formulas are impractical.	Results are estimates, not exact answers; they vary each time you run the simulation.
Easy to understand — anyone can flip a coin or use a random number generator.	Requires a large number of trials for accurate estimates, which can be time-consuming by hand.
Flexible — can model multi-step events, dependent events, and unusual distributions.	A poorly designed simulation (wrong random device, too few trials) produces misleading results.
Builds intuition about randomness and variability in probability.	Does not explain why a probability has a certain value — only estimates what it is.

Comparing strengths and limitations of probability simulations

✦

KEY TAKEAWAY

A simulation is like taking a poll: the more people you ask, the closer the result is to the whole population's opinion. But no poll of 20 people can perfectly represent millions. Similarly, a 20-trial simulation gives a rough idea, while a 10,000-trial simulation gives a very precise estimate. Always consider whether your sample size (number of trials) is large enough for your purpose.

SECTION 8

Connection to Advanced Probability & Statistics

The simulation techniques you learn now are the foundation for much of advanced statistics. In AP Statistics and college courses, you will encounter formal tools for quantifying how accurate a simulation estimate is, including confidence intervals and margin of error. You will also study the Central Limit Theorem, which explains the mathematical reason why estimates improve with more trials.

This Lesson (Grade 9)	Advanced (AP / College)
Estimate probability using experimental proportion P̂	Build confidence intervals around P̂ with specified margins of error
Observe that more trials give better estimates (Law of Large Numbers)	Prove the Law of Large Numbers formally and use it with the Central Limit Theorem
Use coins, dice, and simple RNGs	Use Monte Carlo simulations with thousands or millions of computer-generated trials
Judge accuracy informally by comparing to theoretical value	Calculate standard error and use hypothesis testing to evaluate accuracy

How probability simulation concepts scale from Grade 9 to advanced study

The key idea to carry forward is this: every simulation is an experiment, and every experiment comes with uncertainty. The tools for measuring that uncertainty become more precise as you advance in your math education, but the underlying philosophy — use repeated random trials to approximate truth — remains the same from Grade 9 through graduate school.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

Explain in your own words why running only 5 trials of a simulation is likely to give a poor estimate of a probability, while running 500 trials is much more reliable. What principle justifies this?

Show answer

With only 5 trials, a few unusual results can drastically change the experimental proportion. For instance, if you flip a fair coin 5 times, getting 4 heads (P̂ = 0.80) is quite plausible, even though the true probability is 0.50. With 500 trials, the Law of Large Numbers tells us the experimental proportion will be much closer to 0.50 because the effect of any single unusual outcome is diluted across hundreds of trials. The larger sample "averages out" the randomness.

Question 2: BASIC CALCULATION

You run a simulation of rolling a standard die and checking whether it lands on 6. In 60 trials, you roll a 6 exactly 13 times. (a) Calculate the experimental probability. (b) The theoretical probability of rolling a 6 is 1/6 ≈ 0.1667. Find the relative error of your estimate.

Show answer

(a) P̂ = 13 ÷ 60 ≈ 0.2167 or about 21.7%. (b) Relative error = |0.2167 − 0.1667| ÷ 0.1667 × 100% = 0.0500 ÷ 0.1667 × 100% ≈ 30.0%. The estimate overshoots the theoretical value by about 30%, suggesting that 60 trials may not be enough to pin down a probability as small as 1/6 very accurately.

Question 3: INTERMEDIATE

A weather forecast says there is a 40% chance of rain each day. You want to simulate the probability of rain on at least 2 out of 3 consecutive days. (a) Describe a random device and outcome assignment for simulating one day. (b) Define what one trial looks like. (c) Suppose you run 100 trials and get 35 successes. What is your estimate?

Show answer

(a) Use a random number generator producing integers 1–10. Assign 1–4 as "rain" (40%) and 5–10 as "no rain" (60%). (b) One trial generates three random numbers (representing three consecutive days). A success occurs when at least two of the three numbers are in the range 1–4. (c) P̂ = 35 ÷ 100 = 0.35, or 35%. This means the simulation estimates a 35% chance of rain on at least 2 out of 3 days. The theoretical answer (using binomial probability) is about 0.352, so this estimate with 100 trials is very close.

Question 4: APPLIED

A bakery sells mystery boxes containing one of four flavors: chocolate, vanilla, strawberry, or lemon, each equally likely. A customer wants to know the probability of buying exactly 3 boxes and getting all three different flavors (no repeats among the three, out of four possible flavors). Design a simulation using a standard die (ignore rolls of 5 and 6 by re-rolling) and describe how to run 50 trials. Then, if 14 of your 50 trials are successes, estimate the probability and calculate the relative error, given that the theoretical probability is 3/8 = 0.375.

Show answer

Assign die faces 1 = chocolate, 2 = vanilla, 3 = strawberry, 4 = lemon. If you roll a 5 or 6, re-roll until you get 1–4. One trial consists of rolling the die three times (one roll per box). A success is when all three rolls produce different numbers. Run 50 such trials and tally successes. With 14 successes: P̂ = 14 ÷ 50 = 0.28 (28%). Relative error = |0.28 − 0.375| ÷ 0.375 × 100% = 0.095 ÷ 0.375 × 100% ≈ 25.3%. The estimate is off by about 25%, which highlights that 50 trials is a modest sample — running 200+ trials would likely bring the estimate much closer to 0.375.

Question 5: CRITICAL THINKING

A student claims: "I ran a simulation with 200 trials and got P̂ = 0.48, while the theoretical probability is 0.50. Therefore, 200 trials is always enough for any simulation to be accurate." Critique this reasoning. Under what conditions might 200 trials still produce a poor estimate? How would you advise the student to evaluate accuracy more rigorously?

Show answer

The student's reasoning is flawed because getting a good result in one instance does not mean 200 trials always suffices. Accuracy depends on the probability being estimated: rare events (like P = 0.01) require far more trials because you might see only 0–4 successes in 200 trials, making the proportion very unstable. Also, a single simulation run can be lucky — the student happened to land near 0.50 this time but might get 0.42 or 0.55 the next time. To evaluate accuracy more rigorously, the student should (1) run multiple sets of 200 trials and compare the spread of estimates, (2) gradually increase the number of trials (200, 500, 1000) and observe whether the estimate stabilizes, and (3) compute relative error when the theoretical value is known. A more reliable standard is to repeat the simulation several times and check that results cluster tightly around the theoretical value.

SUMMARY

Lesson Summary

Summary

A probability simulation uses a random device — such as a coin, die, or random number generator — to model a real-world event and estimate the probability of a specific outcome. The five essential design steps are: state the problem, define a trial and success, choose an appropriate random device, run many trials, and calculate the experimental probability (P̂ = successes ÷ trials).

The Law of Large Numbers guarantees that as the number of trials increases, the experimental probability converges toward the theoretical probability. Accuracy is assessed by comparing P̂ to the known theoretical value, computing relative error, or running multiple sets of trials to check consistency. Always remember: a well-designed simulation with a large number of trials is a powerful, flexible tool for estimating probabilities that are otherwise difficult to calculate.