Blackjack is a fascinating game, a steadfast mainstay of the gaming industry, and is also the subject of what might be called "duelling analytics." The seminal discourse, Thorp's "Beat the Dealer," explained how computer simulation provided a means for establishing a basic strategy for playing blackjack to significantly reduce the casino's edge and enable an (almost) level playing field for the player.
However, in the intervening years, the gaming industry has taken evasive action, with institution of rules that limit the ability to exploit the basic strategy - increasing the number of decks in play, modifying the payouts, only dealing out 50% of the cards in play, etc. These innovations are likely the result of the casinos' own analytics - determining how slight variations to the rule sets adjusts the casino edge, then implementing those rules.
Of course, some advanced knowledge in statistics and probably can tip the edge back to the players, as long as they tread carefully. For example, take a look at Mezrich's book on the MIT blackjack team.
Analytics is key for a game like blackjack, in whcih the rules are clearly defined, dealer actions are prescripted, and except for the few actions of placing a bet or collecting your winnings, is largely automatic. However, poker, while being another popular casino attraction, is considered more of a game of skill, not luck, and while analysis does provide insight into how to play certain dealt hands, the skill lies in the abilitiy of the player to transform limited information into decisive action. Watching poker tournaments on TV, which are enhanced by showing the viewer each player's hands, demonstrates how thought processes drive the activity. So while analysis is important, success is a combination of information processing and operational intelligence.
So, here is the question: can we abstract the difference between blackjack analytics and poker analytics and apply these within a business intelligence program?