BJSurvey analyzes two main aspects of a player, which are
his betting pattern to determine if he is card counting,
his basic strategy to rate his decision accuracy.
Both are part of the casino's long-term advantage or disadvantage. Since the two calculations are quite different, each requires a different amount of data to read the player accurately.
A minimum of 75 rounds are required to determine if a player is card counting, but the recommendation is 100-120 rounds. Entering more than 300 rounds will increase the calculations' accuracy only by a minuscule amount and is generally not necessary. This recommendation assumes that
there were sufficient fluctuations of the count. For example, if the count was never positive, BJSurvey can't know how much the player will bet when it is.
the player isn't "borderline." Players that use a winning progression betting strategy can look like card counters in the short-run. When the count is positive, it is more likely to win a hand. Therefore, this type of player will sometimes bet higher when the count is positive - similar to a card counter. However, there will be sufficient high bets when the count is negative to distinguish a winning progression player from a card counter in the long-run. At the same time, some card counters will use a winning progression for camouflage. This type of player makes it necessary to enter more rounds than recommended above to analyze this properly.
When you need to know a player's decision skills to a high degree of accuracy, at least 270 hands - not rounds - are required.
This number is based on the fact that
there are 350 different basic strategy decisions (16 hard totals, 9 soft totals, and 10 pairs against the dealer's 10 different cards) That number does not even consider that a 2-card soft 18 is not the same as a 3-card soft 18 since you can double on one, but not the other. BJSurvey does treat those as two different decisions.
a player has to make about 130 decisions per 100 hands.
Therefore, you need about 270 hands (350/1.3) to get at least one of each decision - that's assuming the decisions are perfectly distributed (which they never really are).
Since the player will make more decisions when occupying multiple spots, use the total hands played rather than rounds (dealer hands).
For the unencountered Basic strategy decisions, BJSurvey has to assume that the player did them correctly. Therefore, a player would appear to be more skilled than he is when the sample size is too small as he never had the opportunity to make those mistakes. This small sample size causes a slightly lower-than-actual long-term "Total Casino Advantage."