Your Watch Is Wrong. That Is Not the Same as Useless.

Consumer devices are poor at estimating energy expenditure, and this is not a close call. The most-cited head-to-head test of seven wrist-worn devices found that not one estimated energy expenditure to within 20% error in any setting tested u2014 the same devices whose heart-rate estimates were mostly within 5%. The pattern repeats in meta-analysis: no consistent accuracy, and enormous variation between devices measuring the same activity. So the honest question is not whether your watch's calorie figure is right. It is not. The question is whether anything useful survives that.

Something does, and it is the part most apps throw away. A given device tends to be wrong in a *similar* way from day to day, because the same model is making the same assumptions about the same body. That consistent portion of the error is exactly what cancels when you compare a day against your own recent days measured by that same device. Kriterion never adds a day's reported burn to your target. It adds only the amount today exceeded your own typical day, so most of the device's standing bias subtracts itself out before the number reaches your food budget.

Whatever is left over is caught by the adaptive engine within a fortnight, because that engine reads your actual weight change rather than anything your watch claims. One popular belief deserves correcting, because we checked it rather than repeating it. Wearables are often said to simply read high. Against gold-standard measurement the direction is not that tidy: in a comparison against doubly labelled water, most devices tested *under*-estimated total daily expenditure and a minority over-estimated it, with error running to enormous proportions on some models. Reading high is common during exercise specifically, but as a universal rule it is folklore.

This matters for how the feature is built u2014 subtracting your own baseline is direction-agnostic, and works whether your particular device runs hot or cold, whereas a fixed haircut assumes a direction the evidence does not support. What none of this fixes is a device that is inconsistent rather than merely biased, or a genuinely unusual day whose reading is unusual for the wrong reason. That is why the credit is capped at 1,000 kcal, why it is off until you switch it on, and why every calorie it adds is itemised on your target breakdown with the baseline it was measured against.

We are not claiming the watch is right. We are telling you where it is wrong and letting you decide anyway.