Brock Windsor said:
Straight counting in a traditional casino banked bacarrat game is not profitable in any practical sense. I believe Thorpe provided the initial proof and Grosjean has done his own research that reached a similar finding. There are ways to exploit the game and there have been countable sidebets on it in the past but these opportunities differ from casino to casino.
-BW
Thorp's (not Thorpe) primary research into baccarat was conducted in the sixties (A favorable side bet in Nevada baccarat EO Thorp, WE Walden - Journal of the American Statistical Association, 1966), and involved analysis of no more than six card subsets using valve-era computer technology. It was meant to be a preliminary study, obviously.
The Grosjean study, (Stewart N. Ethier and William R. Eadington, eds. Optimal Play: Mathematical Studies of Games and Gambling. Reno: Institute for the Study of Gambling and Commercial Gaming, 2008. ) is very amateurish. For example it cites the aforementioned Thorp study as providing proof that the tie bet cannot be counted, when, in fact the Thorp study predated the existence of the tie bet. There are only two references cited at all in fact despite numerous allusions to uncredited sources.
There are myriad methodological problems with the Grosjean study which would take me a month to cover in detail, but one of the biggest problems is that, as Grosjean admits, he has no idea at all how to improve the returns from his card-counting simulations by selectively leaving packs. Imagine someone programmed a series of computer simulations at poker or blackjack which totally ignored the advantages of game selection: they would conclude both games were a waste of time for the professional.
There are a few dozen people I know of around the world who have made a professional salary counting baccarat in one form or another. Essentially what are you looking for is very extreme concentrations of cards at the end of the pack on the last hand or the penultimate hand (the all-tens remainder being the classic example). These can produce very large advantages.
Those situations are very rare, so in order to make the thing practical you have to constantly make calculations as to whether a shoe is likely to produce a favourable subset as it is being dealt out, and very often leave for a fresh pack (or actually request a shuffle). Particulary when you add in the value of rebates, which are very easy to obtain at baccarat, you can make a good living doing this.