Welcome and thanks for sharing.
In the short run every method seems to be good mostly as players try to raise the probability of success in every way (progressions, bet selection widely intended, following or not trends or lucky/unlucky players, etc) but itlr every attempt will be of no avail to consistently win.
We need more than that.
For example we have been playing successfully "for long" a very simple method: we simply bet that a new Banker hand was followed by another single Banker hand (that is betting B after PB) utilizing a 1-2-4 progression.
Anytime this progression failed (meaning that a cluster of three or more B singles appeared) and whenever a new B streak trigger came out, we raised our standard unit to 2, now wagering 2-4-8, then 4-8-16 and so on until the deficit was proportionally and slowly recovered step by step.
Even if it could sound as silly, this system has a math foundation as itlr PBB>PBP, B streaks are more likely than B singles, isolated B singles are more likely than clustered B singles and so on.
In a word and even taking into account the vig burden, the probability to be ahead of something along the way is close to 100%.
Notice that patterns as BPPPPBPPPBPPBPBP....will produce "just" three losses as the betting is stopped until a new B streak comes out.
If you test your data you'll see that a two-step martingale failing won't come out so often and of course you need a kind of balancement to get a consistent long term profit.
The main problem to overcome is to get a decent distribution of winning and losing shoes, nonetheless is just a matter of time to recover any deficit.
But if you look more carefully to those shoes producing a lot of B singles clusters and few B streaks you'll see a kind of cluster-cluster effect.
The reason is because such shoes will present few asymmetrical hands, asymmetrical hands went "wrong" for B side, B drawing hands were more predominant than standing hands, fifth card was mostly belonging to the 3,4,5,6,7 category.
Easy to notice that itlr a perfect world would contain a minor whole amount of such situations.
On average asym hands impact on the whole shoe is 8.4%, on asym hands B gets a 15.86% advantage, B drawing hands are inferior to B standing hands, fifth card is more likely to be a not 3,4,5,6,7 rank category (1:1.6 ratio).
Moreover any two card point higher than the opposite side is going to win about 2/3 of the times.
Similarly to what happens in other games, we should think baccarat as a game of ranges and not in term of exact outcomes.
That's why the shuffling issue is of utmost importance as it's one trillion impossible to guess right into a random distribution.
In some way a proper shuffle judgement is even a better indicator than edge sorting as we want to beat the game legally and, more importantly, we want to be payed after our winning sessions (with all due respect to the baccarat queen Cheung Yin Sun).