Online FDR control is a way to bound the fraction of false alarms across many streams that you are testing continuously over time.
Also known as: false discovery rate control
Online FDR control is a way of bounding false alarms across many monitors at once: it gives you a single budget for false discoveries and spends it across every stream and every look, so the overall rate of crying wolf stays bounded. Watching one metric is hard enough; a real system has hundreds. If each has its own alarm, false positives pile up fast — even a tiny per-alarm error rate becomes a flood of noise across the fleet.
This is what turns "we have monitors" into "we have alarms we trust." Instead of tuning each threshold and hoping, you get a fleet-wide guarantee: of everything that fires, only a controlled fraction are false.
The rigor comes from online multiple-testing procedures (the LORD / SAFFRON family) adapted to anytime-valid e-values. It is the piece most dashboards simply do not have — they alert per-metric, with no global bound.
ValidAnytime turns these ideas into a live alarm you can trust — valid no matter how often you look. Prove it on your own data, free.