Wow! That's a phillipic and a half.
Because the tend to average out to some degree. Sometimes each count over bets and under bets. The tendency may last to the end of the shoe or be corrected as things average out within the shoe. Like if in my example Hilo was underbetting because more 4 and 5 have been removed by rank than 2, 3 or 6. Then the 2, 3 and 6 start being removed. Hilo ups its bet relative to the Hiopt2 bet change for the cards removed. Hilo may even no longer be under betting the actual advantage. Hiopt2 is never that far off the actual advantage. So even though Hilo is raising its bet when advantage might even be decreasing it is getting closer to the actual advantage. Less likely the deck gets even whackier instead with more 4 and 5's removed despite the fact that they are the most depleted low card rank. So basically the odds favor a correction that may be in opposition of the actual advantage shift if the 2, 3 and 6 ranks are not being depleted at the same pace as the 4 and 5 ranks. Since the ranks that cause Hilo to stray from actual advantage by not being removed as much as the other card group in the low cards are more likely to be the next low cards removed the likelihood is a correction toward the actual advantage. We all know that the likelihood is often not what actually happens in these crazy deck compositions that have use betting more but in the long run what is likely will happen more often than what is less likely.Originally Posted by DSchles
Basically if Hilo is off the actual advantage the odds favor that Hilo and Hiopt2/ASC will likely have less similar bet changes (More 2, 3 and 6 by rank removed from the low card group than 4 or 5 by rank or 7's are out of whack). When 4 and 5's are removed the same as T's both bet the same but when 2, 3, or 6 (or 7's) are removed the same as the T's Hiopt2 is lowering it's bets while Hilo bets the same. That means 60% of the time a low card is removed Hiopt2 is affected differently than Hilo.
SCORE is the relationship between overall EV and SD. There is like a 15 to 20% difference in Hilo's SCORE compared to Hiopt2/ASC. To me that is a big difference in SCORE.
This is the way I like to analyze trips. I figure I can get in at least 3K rounds in a trip of a few days. That's about 15 hours of play when being picky about playing conditions. Stats layout is based on:
Count............SCORE........n0.................. EV.........................SD..............
--------------------------------------------------------------------------------------
Hilo full: ........62.68; 15,954 rounds; $3,782/3K rounds; $8,722/3K rounds
HO2/ASC full: 72.74; 13,749 rounds; $4,382/3K rounds; $9,381/3K rounds
SD range.... (frequency)..............Hilo results.............HO2/ASC results
-----------------------------------------------------------------------------------
Less than -3SD (0.1%).....-$22,384 or less...........-$23,761 or less
-3SD to -2.5SD (0.5%).....-$22,383 to -$18,023....-$23,760 to -$19,071
-2.5SD to -2SD (1.7%).....-$18,022 to -$13,662....-$19,070 to -$14,380
-2SD to -1.5SD (4.4%).....-$13,661 to -$9,301......-$14,379 to -$9,540
-1.5SD to -1SD (9.2%).....-$9,300 to -$4,940........-$9,539 to -$4,999
-1SD to -0.5SD (15.0%)....-$4,939 to -$579..........-$4,998 to -$309
-0.5SD to 0SD (19.1%).....-$578 to +$3,782.........-$308 to +$4,382
0SD to +0.5SD (19.1%)...+$3,783 to +$8,143......+$4,383 to +$9,072
+0.5SD to +1SD (15.0%).+$8,144 to +$12,504....+9,073 to +$13,763
+1SD to +1.5SD (9.2%)...+$12,505 to +$16,865..+$13,764 to +$18,453
+1.5SD to +2SD (4.4%)...+$16,866 to +$21,226..+$18,454 to +$23,144
+2SD to +2.5SD (1.7%)...+$21,227 to +$25,587..+$23,145 to +$27,834
+2.5SD to +3SD (0.5%)...+$25,588 to +$29,948..+$27,835 to +$32,525
More than +3SD (0.1%)....+$29,949 or more.........+$32,526 or more
The bold SD ranges are where the 2 counts are about equivalent. Below that Hiopt2 results are better (69.1% of trips), above that Hilo results are better (6.7% of trips). Trip results will fall at the frequencies listed. As you can see the most likely frequencies Hiopt2 results are better. When you get into the less likely frequencies Hiopts2 results are much better or a little worse.
Last edited by Three; 12-28-2016 at 09:45 PM.
What's the exact definition of "betting correlation"? I did a quick search, and came up with "how accurately a count system predicts you're at an advantage compared to what perfect (computer) play would do" (not verbatim). Perhaps I'm misunderstanding what BC actually is, but given HILO's BC is 0.97 (97%), I believe that'd mean it's accurate 97% of the time, or rather, is inaccurate 3% of the time.
Tthree, your example showed that the HO2 player and the HILO player would still be betting more into advantageous counts. Even if they are betting different amounts, I don't think that's going to throw off the EITS or the pit. What're they gonna think, "He dropped his bet from $150 to $100 and the count went up! Nope, he's not a card counter!" ???
Any way you slice it -- HO2 and HILO are both focused on the same thing -- little cards vs high cards. HO2 is simply more accurate than HILO, but it is not drastically different.
EDIT: Thought I hit 'submit' on this post several minutes ago. In the mean-time, I made an Excel spreadsheet to graph a HILO player's bets and a HO2 ASC's player's bets over the course of a "random walk" shoe. I simmed a 5/6 pen game, H17, DAS, LS, RSA for HILO and HO2 /w ASC, both full indices, 1 player at the table, $100k BR, 0.5 kelly (1.83% ROR), 2 hands play all 1-12 spread. Used the TC's and bet amounts from CVCX and put them into Excel, to have the player "wager" the respective amount according to the current TC.
Question about ASC, if I understand it correctly, you multiply the excess Aces (per deck) remaining by 2 and add that to the RC, then divide by decks remaining to get the TC...yes? IE: If there are 10 aces remaining with 2 decks remaining with an RC of +4, you determine there is 1 excess ace per deck remaining (you expect 8 aces in 2 decks, you have 10. 10-8 = 2, 2/2 = 1). Multiply 1*2 = 2, add that to the RC of 4 to get an RC of +6, then divide by decks remaining (2), 6/2 = +3 TC. And if there are negative excess Aces you subtract the multiplied number, yes? (Of course, for the following round, you revert back to the RC of +4 (or whatever it changes to, based on cards coming out in that round), and make the ace-adjustment for each round individually.)
Last edited by RS; 12-29-2016 at 04:18 AM.
"Everyone wants to be rich, but nobody wants to work for it." -Ryan Howard [The Office]
Yes you are. BC is the correlation of the count tags to the full deck betting EoR's. It says nothing about betting accuracy. Betting accuracy would be measured by the SD of the actual advantage bell curve around a betting bins advantage estimate. BC is also a relative measure not quantitative. A BC of .97 is better than .96 but you can't quantify how much better. All you have is a relative comparison.
If you are referring to my SD ranges it doesn't show anything but the result distribution for 3K round sessions. How can you get any info about when they change their bets from it. You are under the impression that you are trying to fool the catchers into thinking you aren't counting. They are going to be pretty sure both are counting. What you are trying to do is make the catcher comfortable not taking action against you. That means a tolerable spread and bet moves and plays that don't fit the Hilo template they are almost certainly using to catch you. If you are using Hilo and they don't act there is no way they can say that they had doubt since that is the count they use to catch you unless you are making lots of mistakes. You are making less as a Hilo player so can you afford the cover. Are you spreading more to make up for a lower hourly or playing longer?
I thought my SD range example showed this. You only make about $600 more per 3K round trip on average. You have larger worst case losing trips which brings you trip average down to $600 or so but almost always come out $1K ahead give or take.
That would be interesting to see the sim results for bet change per round and several rounds or advantage estimate per round for the TC bin you are in or Kelly optimal bet per round and its change. I would use players playing 2 spots. One player using Hilo and the other Hiopt2/ASC.
Last edited by Three; 12-29-2016 at 09:38 AM.
Originally Posted by RollingStoned
Any way you slice it -- HO2 and HILO are both focused on the same thing -- little cards vs high cards. HO2 is simply more accurate than HILO, but it is not drastically different.
Tthree wrote:
I thought my SD range example showed this. You only make about $600 more per 3K round trip on average. You have larger worst case losing trips which brings you trip average down to $600 or so but almost always come out $1K ahead give or take.[/quote]
Tthree, no two players play exactly the same way even if they are using the same systems. Subsequently your one SD range example provided very little factual information in my mind. Now you are adding dollar amounts for differences of gain using HO2, that is quite a stretch indeed.
The $600 is the EV difference for 3K rounds. The table shows that you are more often above that difference than below it. The thing you have to understand is there is no this would happen versus that. The trip results would be totally independent but we are talking a long term distribution of trip results. This table also shows the effect of the 16%+ increase in SCORE on the results distribution for trips. Imagine you hit a really unlikely frequency losing trip. How likely is having more losing trips around it? What is the likelihood of winning enough to recover from it and how many trips is it likely to take? The differences in the answers to these questions are what define the differences in the counts' results. The math says Hilo will wallow in losing runs for a lot longer with a slow recovery. A fast recovery relies on hitting a really unlikely winning trip. For Hiopt2 the recovery time is faster and not as reliant on the rarer results at the top tail of the results frequency bell curve.
I find laying it out like that helps you deal with the varying trip results. My play differs a lot from that Hiopt2 layout. Of course I have juiced up Hiopt2 to perform tighter around a higher EV. My betting count is not the same as Hiopt2/ASC. The point was to show there isn't a lot of difference in results but the difference adds up quick for both swings and positive results. While the Hiopt2 swings are more extreme they favor larger wins. The Hilo swings are less severe but are slanted more toward negative swings than Hiopt2's profile. You have to imagine the way frequencies are distributed within each range to fit the bell curve. Toward the peak and tails frequencies are somewhat evenly distribute within a SD range. In between there is a large bias in frequency toward the less magnitude end of the results range. Anyway the point of that SD range layout is to understand the frequency of results to expect for trips. You will feel a human bias toward looking at BR swings from trip to trip. This is a flawed way of looking at things as the trip end is an artificial boundary within one long session of lifetime duration. However being human it is difficult not to see things as trip to trip.
The layout helps you deal with this human failing by understanding how trip results are likely to fall and how they are likely to add together over time creating swings and flat times for BR growth. Recovery time from losing trips as you pull back and see the trips before and after and the bigger picture recovery likelihood becomes an important aspect of the understanding swings. It can also help spot playing bias that may be introduced by how you manage sessions and game selection. These are human failings. Many feel it is better to play rather than wait for better conditions. This can actually result in playing a lot fewer rounds in the same duration of time by the clock. Others feel a need to try and force wins and play when they should have left already. Both of these things cause a negative playing bias. If you leave after a good run at the tables but stay to try and win back losses after a good opportunity that produced poor results, you will both overexpose your play and create a session bias toward larger losses. The way you end sessions should be symmetric or favor winning sessions. like wonging out of bad counts.
Put that human failing aside a BR are enough to withstand whatever swings are expected with your count and the intestinal fortitude to be unphased by them is all you need. Over the long term of massive number of trips the differences tend to average out and the difference in EV is all that is left.
You wrote: "Question about ASC, if I understand it correctly, you multiply the excess Aces (per deck) remaining by 2 and add that to the RC, then divide by decks remaining to get the TC...yes? IE: If there are 10 aces remaining with 2 decks remaining with an RC of +4, you determine there is 1 excess ace per deck remaining (you expect 8 aces in 2 decks, you have 10. 10-8 = 2, 2/2 = 1). Multiply 1*2 = 2, add that to the RC of 4 to get an RC of +6, then divide by decks remaining (2), 6/2 = +3 TC. And if there are negative excess Aces you subtract the multiplied number, yes? (Of course, for the following round, you revert back to the RC of +4 (or whatever it changes to, based on cards coming out in that round), and make the ace-adjustment for each round individually.)"
No, the math is wrong. If RC = +4, and you have two more aces remaining than you ought to have, you add 2 x 2 = 4 to the RC, to get +8. Then you recalculate the TC, based on the adjusted RC, to get TC = +8/2 = +4.
Don
Can someone answer this question? I've always understood it to be as described above but without the "(per deck)" in bold above. That is, one multiplies the raw number of surplus/deficit aces by 2 and adds it to the running count. Otherwise you actually end up dividing the number of surplus/deficit aces by the number of decks remaining twice.
The answer to this question could help to solve an issue I am currently struggling with.
The thread is hopelessly, and uselessly, complex, to no avail. NONE whatsoever. It is utterly useless to talk about the s.d.s of true count bins, the likelihood of errors, the outcomes of trips, etc. What does any of it possibly matter? Standard deviation of outcomes and expected returns, globally, are all that matter. Both are completely captured by SCORE. Blackjack is a game where the third and fourth moments of distributions (skew and kurtosis) are negligible and needn't be discussed. So, when you enunciate the e.v. and the s.d. of a particular game, you have described all that is necessary to make an evaluation. All the rest of the verbiage is useless.
Don
I am trying to understand this. In the Theory of Blackjack the betting correlation is calculated by taking the sum of the product of the effect of removal which will come out to be 8.12 divided by the square root of the sum of square of the card counting system times the sum of squares of the effect of removal to get the betting correlation. The sum of the squares for the effect of removal would be 2.84.
But I am not sure how you would calculate betting accuracy as Tthree described.
Posting Permissions
Reply With Quote
Bookmarks