The app runs the test. You pull.
A guided, timed protocol streamed live over Bluetooth, so the number means the same thing every time you take it. The app counts you in, holds the clock, and records what the gauge actually saw rather than what you remember it saying.
The protocol is the point
Twenty millimetre edge, half crimp, single arm, five seconds, straight arm. A finger strength number without the protocol it was taken under is not comparable to anything, including your own previous number, which is why the app runs the test rather than asking you to.
The guided flow is a primer on one warm hand at a time, then a ladder at forty, sixty and eighty percent of your last known best, both hands at each step. Hands are never run together.
What it computes from the stream
Peak force is the maximum of a rolling median, so a single noise spike cannot inflate it. Rate of force development is how fast you get there, and it needs a gauge sampling at forty hertz or better; the Progressor runs at about eighty. Mean force from onset is the headline for anything judged on load held rather than load hit. Endurance decline finds the rep where the drop-off starts and how long load was actually held.
It says nothing rather than guessing
A percentile is a claim about a population. Published norms are measured under one specific protocol, so every curve in the app carries the protocol it came from and is ignored unless it matches yours. Where there is no applicable curve you get no percentile at all, because a confident, specific, wrong number is worse than a blank.
Every result stores the protocol version it was taken under, so a test from this month and a test from next year are still the same measurement.
The hardware, stated plainly
This is the one feature that does not work with the phone on its own. It needs a Tindeq Progressor, connected over Bluetooth. That is currently the only supported gauge.