Blackjack q - continuous shufflers - can they be beaten?

[denks]

Hi All, first post so be kind please!

I have a q that needs some maths analysis on it (Im in the process of running several mill simulations but don't have the mathematical background to also run a statistical analysis as well).

To provide some history - over several years I have been toying with the concept that blackjack was perhaps beatable using BS, with emphasis on bet size management rather than using single bets where obviously the house will win in the long run. After much playing with ideas I was actually playing around writing up an AI engine for predictive purposes for another interest when I came across a few concepts which I believed could have been fitted back into BJ. To cut a long story short, with a decent sized roll (roughly 300x the min table bet) I hit the casino and made a bit over $10k in about a month - without counting - on $25 tables. What happened was I would lose most tables, but have a huge win on roughly 1 in 3 tables which more than overcame the losses on the previous tables. Yes the variance was high and rather scary at times but overall everything appeared to be working. Due to other circumstances in my private life I have had to unfortunately use most of my winnings elsewhere and have thus reduced my roll to a level which cannot handle any variance and so have had to stop for the moment.

Now for the maths whizzes - is it possible that the 400 bet win (winning over 50% of the days I played) was just a statistical outlier and cannot continue in the long term? Keep in mind that out of the 30 days of playing I won over 20 days.

I am keen to listen to other peoples findings here based on either the statistical maths, computer simulations or hundreds of hours play (last two I have done).

Here are some basic findings that I have observed in a live casino environment with a continuous shuffler:
Winning and losing streaks tend to happen more often and go for longer than what would be statistically expected in a truly random game
Small and large cards tend to be clumped together - ie it is very common for a stream of picture cards to come out followed by a stream of bad cards - giving almost the entire table 14s and 15s against a dealer 10, or 20s and blackjacks all round against a dealer 8
It is disproportionately common for a dealer to keep busting repeatedly
It is disproportionately common for the dealer to hit high totals on bad cards (eg 4 or 5) repeatedly in a short space of time
I also am finding that I have a greater chance of winning if there are more players on the table than if it is myself heads up against the dealer - I have never won heads up and only once with a single other player. I have won many times on a near full table. I believe this may be due to moving through the cards at a greater rate so as not to get stuck on a batch of 'bad' cards. I know this goes in direct contradiction to the above statement regarding rolls going for longer but I am only going by observations.

Overall what I am getting at - rolls of wins or losses appear to be much more common using continuous shufflers than I would have expected.

Keen to hear any thoughts on this - about to complete a simulator which I believe most closely resembles what a truly random continuous shuffler would deal (not a random card each deal - 50 cards in the discard tray with 10 - 20 cards in a buffer about to be dealt, rest of cards randomly shuffled). Will post the results if anyone is interested?

Would be keen to know what betting strategies would be best suited if a long run were more likely than a purely random game (no doubling every win is not usable, you need a massive roll to overcome the variance).