Thanks for your continued support CODORÇO.
I see the bug. The polling rate (Hz = per second) is calculated like this: samples / (time - nextSecond) / 1 second. So, it needs a full second to pass before the calculation is accurate. I'll fix it somehow - good catch!
Total sample changer = very good idea. 1000Hz gamepad = wow! I hadn't considered this; my gamepads are all very low Hz.
Analogue resolution is also a very good idea. I'm wondering how to calculate this. Currently values come in from 0 to 32767. Do you think recording the minimum distance between current and previous readouts would be enough? I'll take a look at Stick Analyzer.
I'll get to work on this right now, you should see these things in the next update in a few days.