Random Number Generator Wheel: How It Works & Why It Matters
The science behind truly random numbers and how to use them effectively
Ask ten people to "pick a random number between 1 and 10" and you will watch a pattern emerge. Seven shows up far more than its fair share. One and ten barely register. Humans are terrible at being random, and that is exactly why we reach for tools when the result has to be fair. But here is the twist most people never think about. A lot of those tools are not really random either. The dice in your board game closet, the shuffle button on your phone, the "random" pick in most apps — each one hides a different story about what randomness actually means, and how much of it you are getting. This is the most technical guide on the site, so let's be honest about the details. wheel.expert's <a href="/number-wheel/">number wheel</a> pulls its numbers from the Web Crypto API, the same randomness source browsers expose for generating encryption keys and session tokens. Not a toy generator dressed up with an animation. The spin is theater; the number underneath it is built to be unpredictable. Want to know how that works and where it matters? Keep reading.
What Makes a Number 'Truly' Random?
Pseudo-random generators (PRNGs)
Most "random" numbers in software are pseudo-random. A formula takes a starting value — the seed — and grinds out a sequence that looks scrambled but is fully determined. Feed in the same seed and you get the same sequence every time, down to the last digit. That reproducibility is genuinely useful for testing and simulations. It is a disaster for anything that needs to be secret or tamper-proof.
True random generators (TRNGs)
These skip math entirely and measure physical noise instead: atmospheric static, electrical thermal jitter, the decay of radioactive atoms. The outcomes are unpredictable because the physics underneath is unpredictable. They are slower and need special hardware, which is why you do not find one in every laptop.
Cryptographically secure PRNGs (CSPRNGs)
This is the practical middle ground. Still an algorithm, but built so that seeing past outputs gives an attacker no realistic way to guess the next one. The numbers are computed, yet for any real-world purpose they behave as if they were drawn from thin air.
Why split hairs over this? Because the stakes decide the tier you need. A game of dice does not care. A raffle with a prize on the line cares a lot — a predictable generator is a generator someone can rig. wheel.expert leans on the browser's
crypto.getRandomValues(), a CSPRNG, so the number wheel sits in the same trust category as the code that protects your logins rather than the throwaway randomness inside a typical app.How wheel.expert Generates Random Numbers
Step one: ask the browser for entropy.
The Web Crypto API draws on the operating system's pool of accumulated randomness — hardware noise where the chip supports it, timing variations, and other unpredictable inputs the OS has been collecting. We do not invent randomness; we request it from the layer built to provide it.
Step two: turn entropy into uniform bytes.
crypto.getRandomValues() fills an array with values where every possible byte is equally likely. No lean toward high or low.Step three: map bytes onto your range.
This is the step amateurs get wrong. The naive approach is to take a random byte and use the remainder operator to squeeze it into your range — and that quietly biases the result toward the smaller numbers. We use rejection sampling instead: if a value falls in the leftover region that would cause bias, we throw it out and draw again. Slightly more work, but every number in your range ends up with genuinely equal odds.
Step four: animate toward the answer.
The winning number is locked in before the wheel starts moving. The spin then eases to a stop on that segment. So the physics of the animation has zero influence on the outcome — it is decoration wrapped around a decision that is already made.
That ordering matters for fairness. Nothing about how fast you click, how long the wheel turns, or which frame it lands on can nudge the result. The math finishes first, and the show comes second.
Practical Uses for a Random Number Wheel
Games & Entertainment
Education
Decision Making
Creative & Artistic
Customizing Number Ranges and Options
Set any range you like.
Minimum and maximum are yours. Dice-style 1 to 6, a percentage roll from 0 to 100, or something absurd like 1 to 1,000,000. There is no standard die shape limiting you.
Drop in your own list.
Skip the range entirely and type specific values: 3, 7, 12, 42, 99. Only those land on the wheel, which is handy when the meaningful numbers are not consecutive.
Exclude what's already gone.
Running 1 to 10 but seven is already taken from an earlier round? Remove it and spin the rest.
Stop repeats.
Turn on remove-after-selection and each number bows out once it is drawn, so nothing recurs until the pool resets. This is the bingo and raffle mode — every ball comes up exactly once.
Weight the odds on purpose.
Give a number extra weight and it shows up more often. Useful when you want high rolls to stay rare in a game, or when a class is studying loaded-dice probability and needs a visibly unfair wheel to poke at.
Spin for several at once.
Need a six-number lottery line or a batch of values for a stats exercise? Pull multiple results in a single go.
Digital Number Wheel vs. Physical Dice
A physical die gives you tactile feedback, needs no power, and fits the mood of a tabletop night in a way a screen never quite will. Pick it up, throw it, done — no animation to sit through. For the cozy stuff, dice are hard to beat.
The digital wheel earns its place the moment your needs stop being standard. Dice come in a fixed handful of shapes, so the day you need a fair roll from 1 to 73 there simply is no die for that. A wheel does it without blinking. There is also the quiet problem of the physical objects themselves — a cheap die can be subtly weighted, a worn one favors a face, and "throw it again, that one rolled off the table" introduces a person making judgment calls. Software sidesteps all of it.
So reach for digital when the range is unusual, when fairness has to be defensible because money or a prize is involved, when people are joining remotely and cannot share one physical die, or when you simply want a record of what came up. Reach for dice when you want to hold something and the exact range does not matter. Both are valid. The trick is matching the tool to the stakes.
Using Number Wheels for Probability Education
Uniform distribution.
Spin a 1-to-6 wheel many times and tally the results on the board. Each face should land somewhere near one-sixth of the spins. The gap between "should" and what actually happened is the opening for a talk about expected value and why small samples wobble.
Weighted probability.
Build a wheel where 1 has weight 1, 2 has weight 2, on up the line. Have the class predict which numbers dominate before a single spin, then run a batch and check the guesses. Seeing their prediction confirmed lands harder than a formula on a slide.
Independence of trials.
The wheel just landed on 3. Ask whether 3 is now more or less likely on the next spin. The instinct says "less" — and the wheel proves the instinct wrong. Each spin forgets the last one entirely.
Law of large numbers.
With ten spins the tally looks lopsided and "unfair." Push to fifty, then a hundred, and the bars flatten toward even. Watching that convergence in real time is the lesson.
The gambler's fallacy.
If 6 has not shown in a dozen spins, half the class will swear it is "due." It is not. The wheel has no memory and no debt to pay, and a few more spins will usually demonstrate that the universe does not owe anyone a six. For broader classroom routines, the classroom activities picker guide bundles spin-based exercises like these into full lesson plans.
Technical Deep Dive: Web Crypto API
The call itself.
crypto.getRandomValues() is a method on the Web Crypto API. You hand it a typed array — say a Uint8Array — and it fills every slot with cryptographically strong random values, in place. That is the entire interface, and it is supported across every current major browser.Where the entropy comes from.
The browser does not generate randomness on its own. It asks the operating system, which maintains an entropy pool fed by hardware random generators when the CPU has one, plus timing jitter and other physical noise the system collects. The browser is a courier passing along randomness sourced lower in the stack.
The algorithm in the middle.
Between the OS entropy and your array sits a CSPRNG, often built on a stream cipher such as ChaCha20 or an AES-based construction, depending on the platform. Its job is to stretch a pool of true entropy into a long run of output that stays unpredictable.
What "secure" buys you.
The security property that matters here is forward prediction resistance: an observer who has watched every number the generator has ever produced still has no practical way to compute the next one. Combine that with uniform output and you have a generator suitable for keys, tokens, and yes, a fair wheel spin.
And why not just use Math.random()?
Because
Math.random() is a plain PRNG with no security guarantee at all. The spec does not even require a particular algorithm, several engines have used predictable ones, and its internal state can be reconstructed from a handful of outputs. Fast and fine for shuffling a playlist. Wrong for anything where someone has a reason to cheat. That single distinction — non-cryptographic PRNG versus CSPRNG — is the line wheel.expert refuses to cross, which is why every spin routes through Web Crypto instead.Conclusion
So circle back to that party trick from the top — the room full of people who all "randomly" picked seven. The lesson was never that humans are bad at math. It is that real randomness is a specific, buildable thing, and most of the tools waving the word around are not delivering it. A number wheel that draws from <code>crypto.getRandomValues()</code> does deliver it. Same source the browser uses to keep your encrypted connections honest, wrapped in an animation that makes the result feel like an event instead of a function call. The rejection sampling keeps the odds flat, the pre-decided outcome keeps the spin from being riggable, and the whole thing runs free in any browser tab. Whether you are settling a turn order, running a fair giveaway draw, or showing a classroom why "it's due" is a myth, the underlying number was built to be unguessable. Pick your range, hit spin, and trust the number that comes back. What will you let it decide first?
Ready to generate truly random numbers? wheel.expert's Number Wheel is 100% free!
Spin the Number WheelFrequently Asked Questions
Is the number wheel truly random?
It uses the Web Crypto API's crypto.getRandomValues(), which is a cryptographically secure pseudo-random generator (CSPRNG). That is the same class of randomness browsers use for encryption keys. It is not a true hardware random source, but for fairness and unpredictability in any practical setting it is the strongest tier of randomness available in a browser.
What's the difference between this and Math.random()?
Math.random() is a non-cryptographic PRNG. It is fast, but its output can be predicted from previous values and its algorithm is not even fixed by the standard. crypto.getRandomValues() is cryptographically secure, meaning past outputs give no practical way to guess future ones. We use the latter so nobody can game a spin.
What number ranges can I use?
Any range at all. 1 to 6 for dice, 1 to 100 for percentages, 1 to 1,000,000 for big draws. You can also skip ranges entirely and type a custom list of specific numbers so only those appear on the wheel.
Can I prevent repeat numbers?
Yes. Switch on remove-after-selection and each number drops out of the pool once it is drawn, so nothing repeats until you reset. This is the mode for bingo, raffles, and anything that needs every value to come up exactly once.
Why use rejection sampling instead of a simple remainder?
Mapping random bytes onto a range with the modulo operator quietly biases the result toward smaller numbers, because the byte range rarely divides evenly into your target range. Rejection sampling discards the values that would cause that skew and redraws, so every number ends up with genuinely equal odds.
Does the spin animation affect the result?
No. The number is generated and locked in before the wheel starts moving, then the animation eases to a stop on that segment. Click speed, spin duration, and which frame it lands on have zero influence. The math finishes first and the animation is purely for show.
How is this better than physical dice?
Dice are limited to a few standard shapes, so unusual ranges are impossible, and physical dice can be subtly weighted or worn. The wheel handles any range, supports custom weights and exclusions on purpose, works for remote participants, and relies on documented cryptographic randomness rather than how a cube happens to tumble.
Can I make some numbers more likely than others?
Yes, with the weight setting. Give a number a weight of 2 and it lands roughly twice as often as a weight-1 number. It is useful for game balance where high rolls should stay rare, and for probability lessons where you want a deliberately unfair wheel to study.
