Cookbook betting
RJT said:
Two points i would make however, the first is that you shouldn't use the betting system suggested in the cookbook - it doesn't work. Look up NRS betting. It's a little more involved but that will actually win you money.
RJT.
I also have a problem with the betting recommendations in the cookbook and the broader conclusions Snyder draws from them. (My opinion is that the cookbook is quite good on some methods of tracking shuffles, but not good on betting or the evaluation of profitability.) However, I want to qualify my criticism of the cookbook betting method, because I think it is a useful method in some circumstances, just not all.
In my view, the betting method in the cookbook works fine for the kind of approach it is probably intended for: namely, targeting very small, rich slugs that end up in a small section of the new shoe. Two factors make the cookbook betting method okay in this specific context.
First, the slug is handpicked. It is not taken from a predetermined point in the shoe that therefore has a random count attached to it. Rather, it is a rich slug that is selected from wherever it occurs in the previous shoe precisely because it is rich.
Second, the slug is small and may therefore be dispersed through a small section of the next shoe. As a result, the initial true count will be high and unlikely to turn neutral or negative before the final betting point of the slug.
An example may illustrate these two factors. If a player always intends to track the first 6 cards of a shoe as the slug (a predetermined slug), the count of those 6 cards can vary from -6 to +6. Sometimes the slug will be useful, sometimes it won't be. But if, instead, the player intends to find 6 cards somewhere in the shoe that is rich, the slug count will not be random. It will always be useful information (provided it is trackable and not broken up by the shuffle). This aspect of the example relates to the first factor.
In terms of the second factor, if those 6 slug cards are all big and get mixed into a half-deck section of the next shoe, they produce an initial true count of just under +12. The player won't lose much by assuming a constant true count through that half deck. Even if the running count dropped 13 over the first 13 cards of this half deck, NRS tells us that the true count would still be about +6. (I've assumed a 6-deck game, so the IRC using NRS would be about +20 and N about 1.8.) Unless the player is heads up, a bet can't be placed much deeper into the playzone than this, so the cookbook betting method performs okay.
A third factor that will sometimes further validate the cookbook betting method in cases similar to the example above is that the slug may be dispersed fairly evenly throughout the section of the shoe in which it appears (depending on how the relevant cards are shuffled). If the slug cards are virtually all big cards, there will then be potency all through the slug.
Having said all this, the cookbook downplays the value of other tracking approaches because it evaluates them under the assumption that the cookbook betting method (effectively, an assumption that the true count is constant throughout the playzone) is always used. This is very, very far from optimal if either the slug has a random count (i.e. comes from a predetermined area of the previous shoe) or the slug is dispersed throughout a large playzone. In these cases, a lot of the value of the track comes in being able to update the true count using NRS throughout the playzone.
For example, the player might track a predetermined 1.5 decks (the "slug," which may come from various parts of the previous shoe) into one half of a 6-deck shoe. Unless, by chance, the predetermined 1.5 decks is very rich, the cookbook betting method will be useless. For instance, there is not much use assuming a constant true count of zero through one half of the shoe because the predetermined slug happened to have a zero count. Accordingly, Snyder arrives at the conclusion that best-half play is worthless. However, even the zero-count situation (which is the worst-case scenario for the best-half tracker) is not worthless when using NRS. The player still benefits, even though the IRC is zero, because the shoe can be treated off the top as containing only 4.5 decks rather than 6.
Like Norm, I spent time analyzing the cookbook betting recommendation, though in my analysis I needed to resort to algebraic approximations, and so on, as I am not a programmer. (I have always been big on NRS for predetermined and/or large playzones, and the comments on NRS in the cookbook irritated me.) My findings agree with yours and Norm's
if the slug comes from a predetermined location and hence has a random count. (I am guessing this is why Norm qualifies his statement by noting that the cookbook approach is unprofitable under many, but by implication not all, circumstances.) For large playzones containing predetermined slugs, the cookbook betting approach often won't even produce a player edge (over the entire shoe, not just playzone). However, if the slug is small and rich (because it is handpicked), the method will produce a decent player edge. The shorter, richer, and less dispersed the slug, the nearer the cookbook method gets to NRS profitability.
I get the definite impression that Snyder's own approach to shuffle tracking involves handpicking slugs, thereby guaranteeing richness. On the other hand, there is an ambiguity, because the cookbook presentation of count frequencies for different slug sizes gives the impression that the slug count is random. This means that Snyder evaluates the profitability of various slugs under the assumption that they have a random count. This method of evaluation is not applicable if the slug is handpicked and hence guaranteed to be rich. The evaluation method also ignores the player disadvantage when not playing the slug.
So, my view is that if a player handpicks small, rich slugs, it is okay to use the cookbook method. However, if the slug comes from a predetermined area of the shoe and/or is considerably dispersed, NRS is indispensible.
The cookbook analysis, based on the constant true-count analysis, results in a dramatic undervaluation of useful methods such as best-half play, or the cutting out of poor slugs, because these methods acquire much of their power from true-count updates within the playzone.