Definitely looks happier the closer to the lands it goes. I originally used the Hornady gauge to find the lands and didn't like that it was hard to get repeatable results. I changed my process to setting up a fire formed dummy using the same load development brass and use a caliber diameter mandrel to set interference fit @ ~.0005" seated long enough to impact the jams. I recommend using a mandrel and graphite dry lube and leave it in the neck to alleviate even more interference as to not stick the bullet in the rifling. Chamber the dummy round and force the bullet into the case while in contact with the jams. Some say to strip the bolt before doing this to better feel the pressure during the process but seems to work fine without this step.
I believe you could be in the neighborhood of .060" off the lands to start basing it off the Hornady gauge. . One or possible two reasons for this; first if you used the Hornady tapped case then you're not using fire formed brass and secondly the neck interference fit is positive on the Hornady cases by ~+.002" over caliber(.245" in your case). Each .001" of interference fit (neck tension) adds ~.015" of available seating depth or distance from the lands so these gauges are ~ .030" off the lands and with another -.002" (This is what I use some like more or less) interference fit then you would be ~.060 of the lands. Seems to show up in your results since most hybrid targets like a seating depth between .025-.045" off so your CBTO of 1.655 would put you near ~.045 off (assuming neck tension of .002") where I see POI and groups improving. I would continue closer with caution using same increments to near .025" and you should see additional improvement.