I think 220k hands is good enough to start looking at winrates:
220k hands @ 7.37bb/100:
70% confidence true winrate was between 5.45 bb and 9.29 bb
95% confidence true winrate was between 3.53 bb and 11.21 bb
compared to 500k hand sample which would put you at:
70% confidence true winrate was between 6.1 bb and 8.64 bb
95% confidence true winrate was between 4.82 bb and 9.92 bb
Of course the bigger the sample the better but the difference is not drastic, certainly a 220k hand sample qualifies as "real"
I see where you're coming from though since you put in massive amounts of hands at a lower winrate you probably have 200k hand breakeven or even losing stretches all the time even though you know you're a winning player
220k hands @ 1.5bb/100:
you'll be down 21% of the time (vs 0.0006% if your winrate was 7.37bb)
and as the confidence intervals suggest, this player's true winrate could easily be <0bb/100
--
link to variance calculator:
http://pokerdope.com/poker-variance-calculator/