How to Teach Probability With a Virtual Deck of Cards

The line on the page with nothing behind it

The card deck page has a short "Perfect for" list near the bottom. One of the lines is "Probability experiments". I wrote it, it has been sitting there for months, and there was nothing on the site explaining how you would actually do that. This is that explanation.

It matters more than it sounds, because a deck of cards is one of the few objects that teaches probability honestly. Dice and spinners hand you independent events: every roll starts clean. A deck does not. Every card you draw changes what is left, and that difference is most of a probability unit.

Why a deck teaches this better than dice

Ask a class the odds of rolling a six, then rolling another six. Two independent events, one in six each time, and the first roll tells you nothing about the second. That is the dice roller and the number wheel: every result stands alone.

Now ask the odds of drawing an ace, and then a second ace. The first draw is 4 in 52. If it was an ace, the second is 3 in 51, because there are genuinely fewer aces and fewer cards. That is a dependent event, and a deck is the cleanest way to show one, because the class can watch the deck shrink while it happens. The dice versus generator comparison is the same idea for older students: independent tools for independent events, a deck for dependent ones.

It is also the distinction most syllabuses ask for by name. Independent and dependent compound events sit squarely in middle-school probability, and the standards usually list simulation alongside lists, tables and tree diagrams as a way to get at them. A deck on a projector is that simulation.

The theoretical probabilities a 52-card deck gives you

A standard deck has no jokers, so the fractions stay clean. From a full 52:

• One particular rank, say a king: 4 in 52, which is 1 in 13, about 7.7 per cent.
• One suit, say hearts: 13 in 52, which is 1 in 4, exactly 25 per cent.
• A face card, jack queen or king: 12 in 52, which is 3 in 13, about 23 per cent.
• A red card: 26 in 52, exactly half.
• One exact card, the queen of spades and nothing else: 1 in 52, about 1.9 per cent.
• Two aces in a row, without putting the first back: 4 in 52 times 3 in 51, which comes to 1 in 221, about 0.45 per cent. Put the first card back and reshuffle and it rises to 1 in 169.

Those last two lines are the entire lesson in miniature. Whether the card goes back changes the answer, and by a factor of not quite one and a third.

Running an experimental probability activity on one screen

Theoretical probability is what should happen. Experimental probability is what did. Holding the two against each other is where it lands.

1. Open the card deck on a projector. It loads pre-shuffled.
2. Pick an event, for instance "the card is a heart". Theoretical probability: 25 per cent.
3. Tap Draw Card from Deck and have the class tally hit or miss. The Recent Cards from Deck row keeps the last eight draws on screen, so nobody has to take your word for what came up, and the Drawn badge counts every draw.
4. After an agreed number of draws, work out the experimental fraction and compare it with 25 per cent.
5. Hit New Deck to reset and run it again, or pool every group's results into one larger fraction.

What students take away is that a short run wanders, three hearts in the first four draws and then none for ages, while a long run settles, and pooling the whole class settles it faster still. That is the law of large numbers without the phrase getting in the way first.

The catch worth showing them

Put this in front of the class rather than hiding it. If you draw all 52 cards and tally the suits you get exactly 13 of each, every time, because nothing goes back and the deck cannot do anything else. So for a run that is meant to drift around the theoretical value, draw only part of the deck and hit New Deck between trials.

For a genuine "draw, record, replace, repeat fifty times" experiment, where every trial has to be independent, the number wheel set to 1 to 52 or the dice roller is the right instrument. Repeats are allowed there, which is the point. The deck is for when depletion is the lesson.

Is the shuffle random enough to grade?

A fair question to ask of anything you are marking. The deck runs a Fisher-Yates shuffle drawing every swap from crypto.getRandomValues(), the browser's cryptographically secure generator, rather than Math.random(). It uses rejection sampling too, so no card position is quietly favoured by the arithmetic. Every one of the 52 factorial orderings is equally likely on the first shuffle.

None of that has to be taken on faith. The randomness page runs a live chi-squared test against the same generator, which is a defensible thing to put in front of a class that has just been taught what "fair" means. It is arguably more honest than the deck in the classroom cupboard, which has had two rushed riffles and is nowhere near random. The shuffling guide covers why that takes about fourteen realistic riffles to fix.

What it will not do

Stated plainly, because the gaps shape the lesson plan.

• No draw-with-replacement button. To simulate replacement you tap Shuffle Deck or New Deck between draws, which is a second tap on every trial. For a long independent-trials experiment a wheel or dice is less fiddly.
• No built-in tally or chart. Students record results on paper or in a spreadsheet. For a hands-on lesson that is arguably the point, but the tool will not graph it for you.
• The visible history is the last eight cards. The Drawn counter keeps the full count, but you cannot scroll back through all thirty draws of a long run.
• One shared deck per screen. Ideal for a projector or a group working together. Thirty students each running their own experiment each open it on their own device, free and without signing in, but they are not sharing one synchronised deck.
• One card per tap. To deal a five-card hand for a compound-probability question you tap five times; the dealing guide covers that pattern.

A twenty-minute lesson you can run tomorrow

1. Minutes 0 to 5. Full deck on the projector. Ask for the theoretical probability of a heart, a face card, the ace of spades. Write the fractions up.
2. Minutes 5 to 12. Draw twenty cards while the class tallies hearts. Compare the experimental fraction with 25 per cent, then ask why it is not exact.
3. Minutes 12 to 17. New Deck. Draw until you turn up an ace, then ask for the probability that the next card is also an ace. Draw it. This is where dependent events land.
4. Minutes 17 to 20. Pool every group's twenty-draw tally into one fraction on the board. It sits closer to 25 per cent than any single group's did. Now name it: the law of large numbers.

There are more classroom tools on the teacher tools page, and the single-card guide is the plain walkthrough if you want something to hand students directly.

Start a demo now

1. Open the card deck.
2. Pick an event and state its theoretical probability.
3. Draw, tally, compare, then New Deck and go again.

One standard 52-card deck, no jokers, no signup, drawing without replacement so the odds actually move as you go. For teaching the difference between what should happen and what did, that is the whole toolkit.