Random series generator

How it works

You choose a length n, a domain — usually integers between two ends — and a rule about repeats. The tool draws n times and prints the values in the order they were drawn. That ordered list is the series. It is not a permutation of the whole domain unless n equals the size of the domain and you forbade duplicates. It is not a model of a market, a fever, or a plot. It is n samples in a row.

Output shape is part of the tool. Some pages print one number per line. Some print a comma-separated row you can paste into a spreadsheet. Some wrap the list as JSON. The numbers do not change because the punctuation changed. Pick the shape that matches the next program, then copy once. Re-generating because you needed commas instead of spaces is how a saved series silently becomes a different series.

Most on-screen versions are pseudorandom: an algorithm that is hard to predict by eye. That is enough for dummy chart data, a teaching sample, a test fixture, or a game event log. It is the wrong instrument for a cryptographic nonce, a prize with an audit trail, or a scientific simulation that needs a documented generator and a seed you can cite. A pretty list is still a toy if the page will not tell you how it drew.

Independent positions allow the same value twice. A series that secretly avoids recent values is no longer a sequence of independent draws. If you needed unique entries, say so in the control, and keep n no larger than the domain. If the page sorts after it draws, you asked for a sample and then dressed it as a ranking. Sorting is a later step. It is not extra randomness.

Length, separators, and a file you can reopen

Length is the first setting that changes the job. Three values are a handful. Three thousand values are a file. A screen that truncates a long series without warning is a poor record. Download or copy the whole list if the next step is a chart. A glance at the first screenful is not the series.

Length: how many values, stated before you draw. Domain: inclusive ends, and integer versus decimal. Repeats: allowed on a stream; forbidden on a unique list. Shape: lines, commas, or a structured blob the next tool can read.

Save the settings with the file: n, ends, repeats, and whether you sorted. A naked column of numbers is a souvenir. The next person cannot tell whether 0 was allowed, whether duplicates were a bug, or whether someone ordered the list by hand and called the order “random.” The filename can hold the settings if you are kind. A sidecar note is kinder.

If you need the same series again, you need a seed and a documented algorithm, or you need the saved file. Most page widgets will not replay yesterday. That is normal. It is also why you do not argue from memory about what the series was. The file is the series. The memory is not.

A series is not a trend

Random draws wiggle. Runs of similar values happen. A short climb can appear without a cause. If you plot the series and then narrate a story about growth, you invented the story. Dummy data for a chart library should look like dummy data, or you will debug a “trend” that was never in the world. If you needed a rising line, say you want an increasing sequence, which is a different generator: a rule, not a shuffle of chance.

A random series will wiggle. That wiggle is not a market, a fever, or a plot. It is noise you asked for. Do not paste an unlabeled series into a report as measurements. Do not paste it into a medical note. Do not paste it into a financial slide. A teaching slide that says “synthetic sample” is the honest job. A caption under a live dashboard is not, unless the dashboard is a demo and the caption says so.

The generator will not repair a biased domain. Drawing from 1 through 10 and then dropping anything above 7 is not a fair 1-through-7 series. Set the ends you meant. Live with the list, or draw a new list under a new label. Editing a few cells by hand after you disliked them is how a fixture stops being random and starts being a wish.