Serverless Forecasting Engine

Welcome to weatherman! A free API to forecast your time series

ForecastingAPI.com turns compact JSON payloads into production-grade forecasts, rolling backtests, SMAPE accuracy reports, and shareable run pages — all in one flow.

1) Ingest

Paste one payload with multi-series data. Keep it compact and deterministic.

2) Evaluate

Run rolling backtests with AutoARIMA + AutoETS, then compare SMAPE by model/window.

3) Ship

Get a permalinked forecast report with per-series charts and backfill diagnostics.

Enter Payload, get Forecasts

Supply your own payload, or use our M5 demo

Payload instructions

⚠️ Time series are assumed to be continuous at your time granularity. If they are not, backfill them before submitting.

Field Required Description Example
run_name_rootYesBase run name; app prepends PST timestampmy-run
start_datetimeCustomFirst timestamp in ISO format2026-01-01T00:00:00
granularityCustomStep size between points1d, 1h
horizonNoForecast periods after history7
modelNoEngine selectornixtla
seasonal_periodNoSeason length for ARIMA/ETS7
n_seriesNoSeries count hint (M5 sampling)3
backtest_windowsNoRolling holdout windows for SMAPE3
series_namesCustomSeries IDs (demo_mode_m5 reserved)["trend","seasonal","combined"]
seriesCustomSingle list or list-of-lists[[...],[...],[...]]

All Runs

Most recent forecast outputs available right now.

post-fix-check

nixtla · 1d · horizon 24

history 3938 · forecast 96

run-fix-test

nixtla · 1h · horizon 24

history 12 · forecast 48