Dan (OhioCoinHunter) has it right. I definitely use slow and overlapping as my standard practice, however, when I get an iffy signal, I use both slow sweeps and fast sweeps (and the pinpoint function as well as quickmask) to analyze the target before I dig. I do it so quickly and without thinking, I don't even consider it really. One thing to be aware of in general is that faster sweeps combined with high sensitivity settings can increase falsing.
Falsing can also happen with targets on the surface. The side of your coil will pick up the surface target and give a ghost signal about 5-6" to the side of the surface item at a decent depth. This is where using the pinpoint helps isolate the surface target (and you may be able to tell if there are two targets or just one). However, it is NOT always a ghost! Several times I have dug the surface item to get it out of the way and the "ghost" signal turned out to be a good target. I have gotten better at telling these particular ghosts from co-located targets, but it is a skill that comes with many many hours on the machine and a lot of times, you just have to dig that junk out of the way.
If you pinpoint a good signal and it's not a ghost, and there's nothing in the hole using the SunRay probe you are probably dealing with iron, and you have disturbed the clump of rust so that it now reads correctly. By using the SunRay in pinpoint (all metal) mode you should be able to locate the item, whatever it is - iron or not (And you'll be surprised at the little tiny scraps of metal, buttons, etc. the machine will detect as you chase them through the dirt pile with the probe.). But I have also had occasions where the item is just deeper than I expected and the probe cannot pick it up. Here again, all metal on the probe may help (as may hitting the right arrow to bump up your sensitivity while using the probe), but anymore if I think it might be good, I remove more dirt from the hole to make sure it's not just something good that's out of range of the probe. I know for sure I have left good deep targets in holes because my probe did not pick them up and I thought "no way" they could be deeper. I was wrong. I know better now. The E-Trac at auto-sensitivity +3 goes plenty deep.
Another note about high sensitivity, you will miss targets due to overload if you run the machine too hot. Once again, I've learned my lesson here. I was used to running the Explorer a little hot all the time and I carried that over with the E-Trac. I started messing with things a bit, and especially in trashy areas, running the machine hot means it's also hot on the trash and your target separation (the strength of the machine) suffers. In a park or school hunt, I will never run high manual sensitivity again. Auto +3 is my new rule.
Not all good hits will repeat at 90 degrees. There is no surefire, perfect method to not missing anything. I have had good targets where if I can X them at 45 degrees right and left of the original scan, I'll still dig. It's a judgement call based on experience. I don't think it's reasonable or possible to dig everything, but you learn to tilt the probability in your favor by taking in all the info the machine can give you and knowing your site. I'm generally hunting the same old parks I and others have been hitting for decades and still pulling out keepers. Just resign yourself to the fact that you can never get it all and focus on what you can get. If you try to get everything all in one trip you'll just make yourself nuts.
I have not found any significant differences in performance with fast on/off or deep on/off.