Getting that close to the lands may be preventing still tighter groups. I've measured bullet ogive variation before and it is enough that you may be making the amount of gas bypassing the bullet before it jumps to obturate (block off) the bore inconsistent. This gas bypass is what physicist Dr. Lloyd Brownell believed is responsible for pressure variation with seating depth.
I would look at either developing the load jammed into the lands about 0.010" or, start about 0.020" off the lands. I've seen one video by a fellow varying the jump from zero to 0.015" and feeling like he picked out an improved spot in between, but his groups didn't appear to me to provide a statistically significant difference in that range. They were all fairly small. But they weren't bugholes.
The late Dan Hackett wrote that he had a 40X in 220 Swift that he had trouble getting to shoot 5-shot groups under half an inch, and they were sometimes 3/4", like you had. He was also shooting bullets in the 50 grain range. At the time he was loading with about 0.020" bullet jump to the lands, trying out different bullet makes, but not getting the groups any smaller. (At one time, 0.020" had been frequently held to be a universal best jump distance in the benchrest community.) During a loading session, in switching to loading a short Nosler bullet that needed to be seated out further than the previous bullet Hackett was loading in order to make 0.020" jump, he accidentally turned the micrometer adjustment thimble on his seating die the wrong way, seating the bullet deeper, so it had 0.050" jump. He had 20 rounds loaded before he noticed the error. He said he considered pulling the bullets and reseating them, but decided just to shoot them for practice instead. To his astonishment, he got two 0.25" groups and two true bugholes in the ones (the center-to-center distances that were 0.1-something, extreme spread) using that big jump. Far better than the gun had ever shot before.
So much for 0.020" being the best jump distance. So much, in fact, for any particular distance being "best". You have to test for best jump with your bullet and load in your chamber. Berger used to recommend jamming their secant ogive VLD designs 0.010" into the lands, and expected shooters would just accept that they would shoot well in some rifles and not others. But they got so many reports from people getting them to shoot better with some jump that they experimented and found that while the jam method worked in some guns, they had to be as much as 0.150" off the lands before they shot well in others.
You can read
Berger's method of finding the best jump distance. Keep in mind that it is for the long nose secant ogive. A short radius tangent ogive, as you like will be shooting, makes a bigger change in the gas bypass gap annulus for a smaller change in seating depth, so I would use test spacing increments about half of what Berger calls for with secant ogive VLD's in that article with your bullets. But in some guns it can still have to jump long for best effect.
I'll add that my Remington 600 in .222 Remington (also a 14" twist) shot many a cloverleaf with flat base bullets, but never liked boattail match bullets. It would outshoot the Sierra 52 grain BT MatchKing with a flat base 50 grain Hornady SP consistently. So you might want to consider some flat base bullets, such as Berger offers in 52 and 55 grains, and Sierra offers in their 53 grain MatchKing. Bryan Litz has pointed out that it is intrinsically easier to make a perfectly symmetrical flat base than a perfectly symmetrical boat tail. I'll add that the boat tail takes time to clear the muzzle, during which it is subjected to the accelerating blast of muzzle gasses which must deflect off of it with perfect symmetry for best effect.
I think you may have found your powder, though I note you are running below Hodgdon's listed starting load for a 52 grain A-max bullet, so you may do better with a warmer charge.
Chris Long's OBT calculator suggests, via QuickLOAD, that 33.1 grains will produce one optimum barrel time point of 1.106 ms, where QuickLOAD thinks you have 1.109 ms now. Not enough difference to matter. QuickLOAD thinks about 35 grains of Accurate 4064 (a middling load in their data) would give you the same barrel time, but about 120 fps more velocity, and that because it will run at closer to normal pressure and burn more completely in the barrel, it may be worth a try (if you can find any). It also thinks a slightly compressed load of 36.5 grains of IMR 4064 will hit another barrel time sweet spot node (0.966 ms, from Long's calculator), but this load is about maximum using a Winchester case and WLR primer, so work up to it slowly, especially if those are not your components.