I’ve found probing to be much more accurate the slower the probe cycle. I get very repeatable results with a fast probe (now you’ve got a good idea where the probe actually is so the probe distance can be short) and then a very slow probe again. Homing by default works on a “seek” then “fine” method, you can do the same on a probe. I haven’t found a need to average, multiple slow probes get me the same answer (well sub .001")