How Many Times Should You Shuffle a Deck of Cards?

The card that started giving itself away

We have had the same deck at family game nights for a couple of years. Somewhere around month four, the queen of spades picked up a bent corner. Nobody planted it. It wore in from being handled the same way every time. Once you know a mark like that is there you cannot unsee it, and you start wondering how many hands got played slightly differently because someone else at the table spotted it first.

That sent me looking up how shuffling actually works, on the assumption that this was a solved question with a boring answer. It is a solved question. The answer changed two years ago, and most people are still quoting the old one.

The short answer: seven, if you shuffle like a magician

The number people repeat is seven. It comes from a 1992 paper by Dave Bayer and Persi Diaconis, and it is correct: after seven riffle shuffles a 52-card deck is close enough to random that no practical test tells the difference. Six is measurably not enough. Eight buys you almost nothing.

The catch is in the fine print. That proof assumes you cut the deck very close to the middle every single time and interleave the halves cleanly. It describes a shuffle performed the way a card mechanic performs one, not the way anybody deals at a kitchen table.

The longer answer: about fourteen, if you shuffle like a person

In 2024, Mark Sellke, Jialu Shi and Jiamin Wang published a proof covering the sloppy version: cuts landing wherever they land, a different place each time. Their number is roughly fourteen.

So the honest figure for a real human at a real table is double the famous one. Sellke's stated motivation was refreshingly ordinary. He plays poker with friends and wanted to know how long he should be shuffling.

Fourteen riffles is a lot. Time yourself doing it once and you will understand immediately why nobody does it.

Why casinos shuffle three times and get away with it

A Vegas dealer does not shuffle seven times, let alone fourteen. The house standard runs closer to riffle, riffle, box, riffle, cut, which is about three. Shuffling properly would stall the table, and a stalled table stops making money.

Casinos close that gap with machines instead, and the machines have their own history. When Diaconis was hired to check one, he found the shuffled decks still carried rising and falling sequences: enough structure left over that the order was partly predictable. Automatic shufflers are consistent, which is not the same thing as random, and a few of them have been consistent in ways a patient person could read.

None of this is a scandal for a family game night. It is just worth knowing that "we shuffled it" covers a very wide range of outcomes.

So is an online card shuffler actually random?

This is where the comparison gets interesting, because the answer depends entirely on how the thing was built, and most sites will not tell you.

A shuffle written on Math.random() is not cryptographically sound. The JavaScript specification itself says not to depend on it where the sequence matters. Plenty of free card tools use it anyway, and you cannot tell by looking.

PickerSpin's card deck simulator uses a Fisher-Yates shuffle drawing every swap from crypto.getRandomValues(), the browser's cryptographically secure generator, the same one behind the dice roller and the wheel. Every one of the 52 factorial possible orderings is equally likely, on the first shuffle, with nothing surviving from the hand before. If you would rather test that than believe it, the randomness page runs a chi-squared test against the same generator.

Put plainly: a properly built digital shuffle clears the bar on the first try that a real deck needs fourteen passes to reach. That is not marketing. It is just what happens when you can pick a uniform random index instead of throwing cardboard at other cardboard.

What a real deck still does better

Anything involving your hands

Sleight of hand is muscle memory built on weight, edge friction and the give of a card under your thumb. Palming, false cuts and forces do not transfer to a screen. A simulator is useful for rehearsing the logic of a trick, which is what the guide to drawing a card online covers, and useless for the mechanics. You need cardboard for those.

Dealing hands to a full table

Worth stating plainly rather than burying: PickerSpin's deck draws one card at a time from a single pile, for whoever is looking at the screen. It does not deal four private poker hands. If "simulator" means "replace the whole table", real cards still win, or a purpose-built multiplayer app does.

Working with no battery and no signal

Obvious, but real. A physical deck runs in a car, a tent, or a cabin with no bars, and it never has a notification slide over the top of it.

What a digital deck does better

It has no history

The bent queen is the whole argument. Real decks accumulate wear: soft corners, a proud edge on a card that gets cut to constantly, a scuff you only notice at the wrong moment. A digital deck opens identical every time. Fifty-two cards, no biography.

Nothing goes missing

No three of clubs discovered absent halfway through a game, no deck left behind at someone's house.

You can watch the odds move

The deck shows what is left, which turns "the odds are shifting" from a claim into something a classroom can point at. Pair it with the dice versus RNG comparison and you have a clean unit on independent versus dependent events: dice never remember, cards always do.

Free, instant, shareable

No signup, no ads between draws, and it behaves the same whether everyone is around one table or spread across a video call.

Which should you actually use?

Real cards for a real game with private hands and chips on the table. Real cards for practising anything you do with your fingers.

The digital deck for settling something, running a probability lesson, rehearsing the talking half of a trick, or drawing a card when there is no deck in the room. On the specific question of whether the shuffle is random, it is not a compromise. It is the stricter of the two.

And if you are sticking with cardboard, the practical takeaway is not fourteen. It is that whatever you are doing now is probably half of what you need, so add a few more.

What it will not do

One standard 52-card deck. No jokers, no second deck, no dealing to multiple players at once. For a real card game around a table this is a practice and teaching tool, not a substitute for the table.