What formula builds each value?

Each point is base plus trend times the index, plus a seasonal sine term of amplitude times sin(2 pi index over period), plus Gaussian noise. Combining these gives a realistic signal with a long-term direction, a repeating cycle, and random scatter.

How is the random noise generated?

The noise uses the Box-Muller transform to draw values from a normal distribution, multiplied by your chosen standard deviation. This produces bell-curve scatter that looks far more natural than uniform random jitter.

How are the timestamps assigned?

The final point is stamped at the current time, and earlier points step backward by the interval you choose, so the newest reading is last. Timestamps are written in ISO 8601 UTC format.

What is the difference between JSON and NDJSON output?

JSON output is a single array of objects, ideal for loading wholesale into a chart. NDJSON writes one JSON object per line, which is the format streaming pipelines and log shippers expect to consume incrementally.

Can I turn off seasonality?

Yes. Set the seasonal amplitude to zero, or set the seasonal period to zero, and the sine component drops out entirely, leaving just the trend and noise.

Fake Time-Series Data Generator

This tool generates synthetic time-series data that looks like real metrics — with a trend, a repeating seasonal cycle, and random noise — so you can populate charts and dashboards without waiting for real data. Export it as CSV, a JSON array, or NDJSON to drop straight into Chart.js, D3, Recharts, or a streaming test harness.

How it works

Each data point at index i is the sum of three components plus randomness:

value(i) = base
         + trend * i
         + amplitude * sin(2 * pi * i / period)
         + noise * gaussian()

The trend term gives a long-term upward or downward drift. The sine term adds a repeating cycle — set the period to 24 for an hourly-with-daily-rhythm dataset, for example. The noise term uses the Box-Muller transform to draw from a normal distribution, so scatter follows a natural bell curve rather than flat uniform jitter.

Timestamps are assigned by placing the last point at the current time and stepping backward by your chosen interval, then written in ISO 8601 UTC.

Tips and notes

For a clean trending line, set noise to 0 and amplitude to 0.
A seasonal period equal to the number of points per natural cycle (24 for hourly daily, 7 for daily weekly) produces the most convincing rhythm.
Use NDJSON when testing log-ingestion or streaming code that reads one record per line.