clarke.where YA been. (nt)
Don and I have had differences on these matters for some years but we discuss such differences civilly, just as we have discussed other differences civilly through the years.
But we both agree the effect is present, the magnitude, and that at the 1/6 level it is across the board. What you call the bow effect is present (although with play variations the bow can be straightened some) and I think it has everything to do with the difference in expectation and the changed play strategies as more cards are put in the stack dealt but I have not persuaded anyone yet of this brilliant insight so perhaps it is not so brilliant.
The second ingredient of your sausage, treating a true count as a running count below fifty-two cards, is hogwash IMHO because it has nothing to do with the floating advantage being discussed and nobody asserts it does.
Your post answering mine was insulting because it twisted what I said into an endorsement of what you were saying when that was inaccurate. You also threw in your usual insults to ones you like to insult although the rules of that board did add some to your discreteness.
The floating advantage is simple, whatever device is used to explain it. It means the expectation at -10, a zero, or a +10 count at the four deck level of six decks is very similar to the same expectation off the top of four decks and it means the same thing for two or one decks. The expectation at 0 and for all other counts keeps rising.
I was looking for some specific answers from Don about his Chapter 6 tables concerning a particular game I had discovered that offered <1 deck cut-off. Norm and ML offered extensive, helpful, and further explanatory statistics. ET Fan gave me some practical advice and some caveats about putting this strategic derivative into action.
As I have seen several times before, Clarke jumped into the discussion without anything positive to offer, and not only antagonized the 'authorities' I was seeking specific advice from, but also mucked up the whole thread with some sort of 'sausage mathematics' theory derived from a reading of Griffin that has nothing to offer over modern computational analysis. He has a confrontational bias that did nothing to help my understanding about the original thesis. In addition, his posts remain obtuse and undisciplined with a propensity towards self-gratification....
Now that he has been slapped on the wrist at the other site, he comes here to curry sympathy about his mistreatment and continue to confuse and complain. If Clarke were interested in providing anything but regurgative remarks, he would be held in higher esteem by the people who are creative, published, respected and generous to students of the game. I do not think that all people who post on the other site are clones, nor would I lump the philosophies and personalities of The Mayor, ML, ET Fan, Don S., Parker, and Norm into one category. It is only Clarke who wants to wrap a skin around these people and insult them with his 'ground meat' analogies, undifferientiated and substantially composed of by-products. All evidence points to no nutritive value in Clarke's posts, and they always give you indigestion.
My suggestion? Stay away from the other board and admit that your visa has been revoked. Post here to your heart's content, but if you indeed want to add to the knowledge base, at least proofread your posts until they make sense to someone besides yourself.
...are usually unclear and befuddled, regardless of whom they are directed towards.
Cyclops
that is wrong and is only found because of the error I have described. The bow effect has been verified in blackjack, Griffin's Woolworth Blackjack, Pell and the Over/Under bet. Just check over a listing of Gambling Conference papers. The only thing that shows up when penetration gets to the 1 deck point that raises all edges is this miss accounting of true counts.
The bow effect is the only actual floating edge. Even then it occurs from how any effects of removal analysis--or any count-- is going to miss-diagnose egde in games like blackjack. And even that missdiagnosis is well covered by the well known fact that counting has less and less efficiency.
This is an obvious elementary error. Just check how many courses in statistics have sections devoted to such data rank errors. Of course you have your simulated edge going up radically, past the bow effect even, when you are equating a 1.0 TC with a 1.33 TC and the like.
Face it!
The bow effect does change the edge a bit. But with the bow effect there is also the well-known drop in overall betting efficiency; they are linked. There would be differences between the 1.33 TCs at various penetrations, but nowhere near the claim that all counts rise.
Check the graphs that Norm posted. The spikes appear in the intervals where this partition flaw occurs.
...
The claim that all counts show a rise in edge cannot be squared with what is known about the bow effect. The rise claimed by the clones is exactly in-line with the partition error I claim is being seen. It should also be noted how this rise also, like the effects of this partition error, are claimed to have their spectacular effects at the one deck point etc.
The bow effect results in a slight rise for sure. But the rise claimed is just an elementary partition error. I still use the terms clones because none have been able to admit that they are worng.
Because the edge tops off with high counts and drops further with low counts, it must have a bow in the middle.
The bow effect is from how each possible true count, when averaged, has to come to the overall basic strategy edge. That high true counts become nothing more than push 20s and blackjacks is well known (did you read Griffin's text on this or what?). That low counts with basic strategy have you underperforming the dealers hitting is well known. Griffin just added that when alot of the distribution of a count in in these areas the middle of the curve of the count versus the actual edge is bowed. The higher counts falls in edge, the lower counts fall in edge; the middle bends. That is the bow effect.
The bow effect is clearly there. What I think is just an embarassing patition error is the claim that all counts have higher edges. The data suporting this is flawed. It is better explained by this rounding.
See my other answers. I could really have fun posting on how you define ruin and closed and open trials but I digress.
This thread and its parent on rge21.com (thanks for the censorsip Parker---) are all over clones that refuse to admit that they fell for a partition of data error just because Don made it---that two of them have used similar, in past second year statistics courses they have taught at their respective universities, see the indexes online for past courses at "surfsup U" (actually a fine school academicaly but not wanting to violate the no-outing rule) and "our lady of frozen autoworkers" (similar at a more midwest location), as an example of how erroneous conclusions can be made by grouping data wrongly in Signed Rank Statistics tests. Now to be fair this is not a signed rank issue here, but you would think that they would recognise making this type of error in this context too.
There is a bow effect that is limited and is further limited in that it occurs to the same degree that any counting system starts to have its overall efficiency drop off. Kind of hard to exploit it as it occurs just as your ruler starts to get limp: your count's betting efficiency falls off, where the bow effect gets mildly interesting.
There is no floating edge that raises the return from all true counts.
Do they realize the horror in that they have made a simple error and tried to paper it over in a way that will feed Frank's posts for years, and years and years...?
The floating edge, by measurement, tracks very closely the improvement derived when fewer decks are dealt. That is accepted because sims show that. All one has to do is look at Norm's data. While I think that nobody except myself :-) has adequately explained the reasons for the phenomonon (Perhaps I am a Cant clone instead of a Schlesinger clone because for sure I am not conveying clearly enough what I see so clearly), no question remains as to the existence and magnitude of the floating advantage. There is no room for disagreement.
An actual true count that exists has to be at these values: for any ineger n, including 0, n*52/(#cards left). For any range of true counts that are evaluated in a sim, whether rounding to the nearest TC or grouping in ranges of inteeger TCs, at the one deck point the posible TCs rise, for any such interval.
What shows up in, for example TCs that round to 1, past the one deck left point, grows, and the edge for such grows, largely because the TC for that grouping grows.
What you get from the bow effect...well see above new thread!
You just cannot admit that you clones have been equating TCs, past the one deck left point, which would be inflating, with TCs that average one, for similar TC ranges elsewhere in the penetration!
in omitting him and Dan Goren as shuffle track theorists. I just think that aiming to play the shoe as a group of sub-decks is a better approach than his, which is one of an equivalent true count. Appologies to them.
I did omit Grossjean only because his book, Beyond Counting, gives information on the power of lccation and other alternative plays, and is very good for that, but omits so, so much of the how-to.
a grouping of, or partition of, data error that two of the supporting cast involved have used as topics in courses they have taught in second year statistics courses? It will be interesting to see just how near never the time period is before the clones admit this.
