True Count calculation comparison

[nightspirit]

I always wanted to know how the type of true count calculation can take effect on your SCORE. So I simed three different methods of calculation with CVCX for 6D 75%. I chose the Hi-lo Count with Stanford Wongs Basic Indices.

The first method is the calculation by the exact quarter deck, for the second I choose the version with the rounded factors, because for some people it's easier to multiply than to divide. The third sim was very conservative with full deck accuracy. The table is shown below.
I was surprised when I saw the results of the sim. I thought a more accurate TC calculation would yield a bigger SCORE. But up to penetration of circa 80% it's a close call between this three. At higher pen. the quarter deck finally takes the leadership.

I only ran 500 Million rounds and I make no warranty for the accuracy of the numbers. If somebody can check the sim it would be really appriciated. I'm sorry for the bad quality of the image. How can I post it with a higher resolution?

Decks             quarter deck    rounded multiplication   conservative
remaining	     resolution              factor              denominator
			
6	                   6	                  0.2	                6
5.75	                5.75	                0.2	               6
5.5	                 5.5	                 0.2                   6
5.25	                5.25	                0.2	               6
5	                 5	                  0.2	                  5
4.75	                4.75	                0.2	               5
4.5	                  4.5	                 0.2	             5
4.25	                4.25	                0.2	              5
4	                  4	                  0.3                   4
3.75	               3.75	               0.3	              4
3.5	                3.5	                 0.3	               4
3.25	               3.25	               0.3	              4
3	                 3	                  0.3	                3
2.75	               2.75	                 0.4	              3
2.5	                2.5	                  0.4	               3
2.25	               2.25	                 0.4	               3
2	                  2	                     0.5	         2
1.75	              1.75	                 0.6	               2
1.5	                1.5	                   0.7	                2
1.25	               1.25	                 0.8	               2
1	                   1                         1	                 1
0.75	                 0.75	                 1.3	              1
0.5	                   0.5                       2	               1

 

[schismist]

I would try simming a 2 deck game if you want to see a difference. E.g. whether you divide by 5.5 or 6 makes little difference. Whether you divide by 1.5 or 2 can make a big difference.

 

[nightspirit]

I would try simming a 2 deck game if you want to see a difference. E.g. whether you divide by 5.5 or 6 makes little difference. Whether you divide by 1.5 or 2 can make a big difference.

Thanks for the advice. I simed 1 Billion rounds for each method, same strategy and True Count factors like above.
The difference is now clearer. But only for the conservative method against the quarter deck resolution. I would reach a conclusion that if you want avoid the truecount conversation by fractions multiply your RC with the rounded conversation factors or use a RC-System like KO. Otherwise you won't lose that much by using the conservative method when you divede by full decks. I also included comparisions against reKO and KO Full for 2 decks.

http://www.flickr.com/photos/9860955@N02/show/

2D S17 DAS RSA Sr 50% Spread 1:6 
Basic Hi-lo 
                                  SCORE 
1/4 deck resolution         48.54

rounded multiplication
factors                         48.50

conservative division        47.14
by full decks

KO Full Indices              48.74

reKO                         44.10

 

[sagefr0g]

very cool. i suppose the differance as penetration increases can be attributed to greater variance for the less accurate methods as opposed to the more accurate method since the more penetration represents more opportunities to increase bets. just my guess.

 

[nightspirit]

next shot

Somehow I mistrust the sims a wee bit. The above results seemed to close to me. So I ran a new simulation using the first time CVData. What should I say I’m deeply impressed by the vast of information and possibilities it provides. I used the Multi Tracking feature for 3 Players and removed the seat effect, so the players rotate every million rounds.
I ran 2 Billion rounds, rules: S17, DAS, DOA2, LS, 50%.Spread 1-4 min bet $25 Again for our 3 players with different TC methods (same as above)
Player 1 “Trueculator” divides with quarter decks accuracy
Player 2 “Smarty” multiplys by rounded factors
Player 3 “Lazy mind” divdes by full decks

The result was even more clear-cut.

"Truculator" Score 43.72 Winrate $/hour 38.46 NO 22874 (red line)
"Smarty" Score 42.98 Winrate $/hour 38.43 NO 23266 (green line)
"Lazy Mind" Score 36.97 Winrate $/hour 31.46 NO 27048 (blue line)

(Dead link: http://www.flickr.com/photos/9860955@N02/sets/72157600748391247/show/)

conclusion: Don`t censore your SCORE! For pitch games the difference is clear so work with shortcuts or use powerful unbalanced counts.

 

[sagefr0g]

Somehow I mistrust the sims a wee bit. The above results seemed to close to me. So I ran a new simulation using the first time CVData. What should I say I’m deeply impressed by the vast of information and possibilities it provides. I used the Multi Tracking feature for 3 Players and removed the seat effect, so the players rotate every million rounds.
I ran 2 Billion rounds, rules: S17, DAS, DOA2, LS, 50%.Spread 1-4 min bet $25 Again for our 3 players with different TC methods (same as above)
Player 1 “Trueculator” divides with quarter decks accuracy
Player 2 “Smarty” multiplys by rounded factors
Player 3 “Lazy mind” divdes by full decks

The result was even more clear-cut.

"Truculator" Score 43.72 Winrate $/hour 38.46 NO 22874 (red line)
"Smarty" Score 42.98 Winrate $/hour 38.43 NO 23266 (green line)
"Lazy Mind" Score 36.97 Winrate $/hour 31.46 NO 27048 (blue line)

(Dead link: http://www.flickr.com/photos/9860955@N02/sets/72157600748391247/show/)

conclusion: Don`t censore your SCORE! For pitch games the difference is clear so work with shortcuts or use powerful unbalanced counts.

so what is the prime reason for the differant scores? i'm guessing that Truculator and Smarty are finding more opportunities to bet up.

 

[zengrifter]

I don't see a meaningful difference. And, it is a given that at the deep-pene point in the pack we must calc TC to a 1/3 or 1/4 deck, whereas when multiple decks remain a rough per whole deck fraction is quite sufficient.

Notwithstanding, the real issue that you can sim is the effects of using 52-card based TC vs. 26-card based TC vs. 104-card based TC vs. 13-card based TC, each with appropriately calibrated indices.

The extra granularity of the 52 & 104-card TC calibrations enhance the betting. This is where Snyder's 1/4D ZEN (as currently published) breaks down, and to a lesser extent so do Snyder's HiLoLite, Revere's RPC and Uston's APM (all 1/2D-based) suffer from non-optimal betting.

Conversely, Renzey's Mentor with 2D-based TC is exceedingly accurate for betting. If I wasn't so comfortable with 1D TC ZEN I would switch to a 2D ZEN. zg

 

[nightspirit]

so what is the prime reason for the differant scores? i'm guessing that Truculator and Smarty are finding more opportunities to bet up.

Sorry sagefr0g, was too busy the last days. I share my opinion with you. I think the more accurate you calculate your TC the merrier you and exacter you can place your bets. You can finer localize those advantageous opportunities, especially at deep penetrations.

 

[nightspirit]

I don't see a meaningful difference. And, it is a given that at the deep-pene point in the pack we must calc TC to a 1/3 or 1/4 deck, whereas when multiple decks remain a rough per whole deck fraction is quite sufficient.

Lazy Mind waste 18% off his hourly/winrate. I think thats one reason to calculate your TC more accurately. I was simply running the sim to see the effect of the different calculating approaches, nothing more, just out of curiosity.

Notwithstanding, the real issue that you can sim is the effects of using 52-card based TC vs. 26-card based TC vs. 104-card based TC vs. 13-card based TC, each with appropriately calibrated indices.

That would be an interesting challenge. Maybe I do it next month when I got more time by the hand.

The extra granularity of the 52 & 104-card TC calibrations enhance the betting. This is where Snyder's 1/4D ZEN (as currently published) breaks down, and to a lesser extent so do Snyder's HiLoLite, Revere's RPC and Uston's APM (all 1/2D-based) suffer from non-optimal betting.

Yes I thought the HiloLite would have a better performance and was very astonished as I saw that this assumption was false. I think he devolped his counts for primarily the ease of use and not for optimal betting.

Thanks you both for your responses!