There is some variation in the blackjack statistics found on the web and in the various books published on the subject. Part of this is certainly due to the variations in the rules and number of decks used, but it seems that there should be agreement on the flat basics of the game, and on the details upon which the statistic is based.
The calculation of house edge is a case in point. In many cases, the particular discussion neglects to mention the number of decks used for the result, in other instances, the author will mention the decks, but omit as to whether splitting is to 2 or 3 or 4 hands total, or whether the figure was arrived at independently, or copied from others, or based on a simulation or on probability calculations.
I know that the differences between �about -.5%� and �-.4408%� and �-.48%� may be of interest only to those of us with an obsession for this kind of detail and may be a source of eye rolling among most, but I am hooked on finding that answer, and also the answer to a hard standard deviation figure.
I am running my own blackjack simulation program, using the Generic Basic Strategy published by Stanford Wong in Basic Blackjack, on page 30, with the adjustments for �double allowed after split� shown on page 31.
My simulation uses 8 decks, dealer stands on soft 17, double on any 2 cards, aces split once, other splits allowed to re-split up to 3 hands, naturals pay 1.5 to 1 and no one wins tie hands.
The house edge calculation is based upon the initial bet (1 unit), and that initial bet constitutes 1 hand, regardless of whether the player subsequently splits or doubles, or both.
My simulation program is presently at 400,000,000 hands and shows a house edge of .4804% (-.4804 % for the player), playing 1 on 1 with the dealer, and a standard deviation of 1.1494. I do not know how many hands I have to run in order for the 4 digit house edge result to lock in, but it looks like at least twice as many as I have now.
Of course, there can be errors in my program, so I would welcome any input on these topics.