g-tech's work on the idea that in order to move an object weighing a known amount (lb) and a known rate of increase (g's) it would take "X" amount of energy (HP) to do it. It can calculate speed by using a formula like speed = oldspeed + (g's/time or time/g's) and repeating it hundreds of times a second. if a car was going 0mph and then had a g force of .2 for 1 sec it will be moving and 4mph (numbers not real all concept for idea that has been proven by some physicist I'm sure)

The acurateness depends on how close you get the weight input, scales at the drag strip would be nice, and how close the level the unit is set up (I think body roll screws with results a little also but I don't remember reading about it). And it will vary from a dyno. the variation should be dyno hp minus the hp loss to wind drag and other things not afecting a dyno test including bearings on wheels not turning on dyno (normally rear for us and is almost no drag but some drag does exist)

It is still a good tool for a before and after impact of a mod on a car. Oddly though mine shows that I have more HP with struts set loose than I do with them set firm, hense my body roll thery mentioned above. Because I have no wheel spin and struts don't affect HP. Drives should keep eyes on road and let a friend read the g-tech.