Two of these casinos, Slots Plus and Old Havana, came up as "non-random" for blackjack in my testing. In fact, they were so far off from the casinos that did test out for randomness that I shortened their testing sessions. Rather than just analyzing individual and conglomerate playing hand results, I test based on statistical games (the number of hands played per game varies widely), which "end" by either reaching a profit target or by losing a specified maximum bet. I then compare the average dollars won per "game" vs. the average dollars lost, which tells me what percentage of games I have to win to profit. In casinos that test out for randomness, these values are exceptionally close, even when using different software. In casinos that have been adjusted to play in the casino's favor, the results are glaringly obvious. I wouldn't recommend playing at this group's casinos.
As described, this is not scientifically valid.
You cannot stop collecting data and use the early results to decide that a longer set of results would prove a specific theorem statistically.
Random blackjack can produce a HUGE swing early on, yet longer term still end up being shown to be random on a statistical test.
One has to take a pretty large sample in Blackjack to prove that a game is cheating (in other words, is not random).
If anything, you should concentrate your efforts towards beating the games you believe are not random, as these games are the "cheaters", and if the cheat mechanism is non adaptive, a system should be able to beat them.
Surely, if RTG had a cheating Blackjack game, it would have been statistically busted long ago.
What we DO know about RTG is that the games can be reset to different RTPs by an operator, and with Blackjack, this is done by increasing the number of decks used, up to 255 I believe. The number of decks is not specified in the rules, but does affect both the RTP and behaviour of the game. The effect is greatest when fewest decks are used. 255 decks is effectively an "infinite shoe", and no card counting system could ever beat an "infinite shoe" setup.
Far better would be to look at single deck or two deck games with RTPs very close to 100%, such as Microgaming's Classic Blackjack. Here, a very small element of non randomness could yield a strategy that might beat the house, and card counting as the hand plays out may also gain a small extra edge.
If the casino also awards comps on such a game, this too could be treated as an additional contribution to RTP, maybe enough for a +EV system.
It was possible to get a +EV advantage for Classic Blackjack at the old Jackpot Factory VIP casino (it isn't now, you have missed the boat). The comps were a return of 0.2%, which was greater than the house edge for Classic Blackjack, and JF forgot to use a weighted earning system, instead using the 1 point per £10 wagered original MGS variant, which for VIPs in VIP Lounge was doubled, thus producing 0.2% return from comps, and more than 99.8% return from the Classic Blackjack game. This was very slightly +EV, but they also did wager challenges where the more you wagered, the more points you scored.
The bad news, it isn't worth it unless you use a bot, which is against the terms and conditions, and if caught, your RTP drops to 0%, although you will find your deposit thrown out after you as you are booted out of the door.
It also wouldn't work with a bonus, as the play would be more thoroughly audited, and they would spot the bot straight away (there is no Autoplay on Classic Blackjack, so claiming "it wasn't a bot, just autoplay" would not work.