CVData - Deck Estimation for True Count Calculation

**Bigdaddy** · 07-20-2016, 07:03 AM

Running some DD sims in CVData and I'm getting a slightly higher SCORE when I use half-deck estimation vs. quarter-deck estimation for purposes of the true-count calculation (using TKO). How could that be? Isn't quarter-deck estimation more precise?

**Norm** · 07-20-2016, 07:10 AM

Generally speaking, yes. But, there exist a huge number of quirks in blackjack due to non-linearity and borderline conditions since humans don't play with exact TCs. Just as one example, switching between half- and quarter-deck estimation might happen to cause the indices to perform in a more or less risk averse manner, affecting SCORE.

**Bigdaddy** · 07-20-2016, 07:31 AM

Originally Posted by Norm

Generally speaking, yes. But, there exist a huge number of quirks in blackjack due to non-linearity and borderline conditions since humans don't play with exact TCs. Just as one example, switching between half- and quarter-deck estimation might happen to cause the indices to perform in a more or less risk averse manner, affecting SCORE.

Ok - thanks. I use risk-averse indices in the sims. I bet that has something to do with it. I might try generating a set of non-risk-averse indices and see how that affects things.

**Three** · 07-20-2016, 07:32 AM

So you are saying using half deck estimate based indices with 1/4 deck estimates could or maybe should have worse performance than using the deck estimation procedure the indices where derived with? This effect obviously would be greater in pitch games.

**Freightman** · 07-20-2016, 07:40 AM

Originally Posted by Tthree

So you are saying using half deck estimate based indices with 1/4 deck estimates could or maybe should have worse performance than using the deck estimation procedure the indices where derived with? This effect obviously would be greater in pitch games.

I don't think he is referencing performance. What he is suggesting is that an easy to make deck estimation error could inadvertently change some indices from EV maximizing to Risk Averse etc.

**Three** · 07-20-2016, 07:46 AM

Originally Posted by Freightman

I don't think he is referencing performance. What he is suggesting is that an easy to make deck estimation error could inadvertently change some indices from EV maximizing to Risk Averse etc.

They are talking about sim results. I hope simulators aren't programmed to make errors.

**Bigdaddy** · 07-20-2016, 07:49 AM

Originally Posted by Tthree

I am not hands on familiar with unbalanced counts (I have never used one). Did you use the same indices for both sims or generate a separate set generated for 1/4 deck estimate when using that approach? (Hopefully that is not a dumb question. I sense it might be.)

You got it - thanks! I just realized that my indices were generated using half-deck estimation and I was using those same indices in both half-deck and quarter-deck sims. So it makes sense that the quarter-deck sims were not coming out as good, because I was using incompatible indices. Time to create and learn some new indices...

**Norm** · 07-20-2016, 08:12 AM

Originally Posted by Bigdaddy

You got it - thanks! I just realized that my indices were generated using half-deck estimation and I was using those same indices in both half-deck and quarter-deck sims. So it makes sense that the quarter-deck sims were not coming out as good, because I was using incompatible indices. Time to create and learn some new indices...

Hah. Had to go out, and thought of that while I wasn't here.

Yes, that would have such an effect.

**Bigdaddy** · 07-20-2016, 08:27 AM

Originally Posted by Tthree

Unbalanced counts would use a more complicated approach to true counting but I assume the effect would still be there.

Actually, TKO lends itself very nicely to quarter deck estimation in a DD game. That's because the RC has to increase by 1 for each quarter deck played in order to stay at a TC of zero.

So for instance, if the running count has increased by 2 after a quarter deck, then I know I'm one point over and I use my quarter-deck multiplier to determine the TC: (2-1) x .57 (I round to .6 for simplicity), so the TC is +.6. I have a multiplier for each quarter deck played: .6, .7, .8, 1, 1.3, 2.

So another example - If the running count has increased by 10 after 1.25 decks have been played (5 quarter decks), the TC is (10 - 5) x 1.3 = 6.5.

Pretty easy, No?

**Three** · 07-20-2016, 09:45 AM

Originally Posted by Bigdaddy

Pretty easy, No?

Like anything, once you get used to it. The decimal multiplication seems intimidating but like I said once you get used to the few multipliers you have they should get pretty easy. I think, to me, .57 would be easier as 4/7ths or multiply the index by 7/4 and subtract 1 and compare to the RC as a RC index barrier. The others look easy enough to use as multipliers.

**Freightman** · 07-20-2016, 09:55 AM

Originally Posted by Tthree

Like anything, once you get used to it. The decimal multiplication seems intimidating but like I said once you get used to the few multipliers you have they should get pretty easy. I think, to me, .57 would be easier as 4/7ths or multiply the index by 7/4 and subtract 1 and compare to the RC as a RC index barrier. The others look easy enough to use as multipliers.

To me, it seems intuitively easier to use .5 and then add 12%

**Three** · 07-20-2016, 10:11 AM

Originally Posted by Freightman

To me, it seems intuitively easier to use .5 and then add 12%

That's the cool thing about math. There are so many ways to work a problem and get the right answer. Each person will have what is easiest for them and each can develop new strengths to do what was not easy before.

**Goodboy** · 07-20-2016, 03:07 PM

Originally Posted by Bigdaddy

Actually, TKO lends itself very nicely to quarter deck estimation in a DD game. That's because the RC has to increase by 1 for each quarter deck played in order to stay at a TC of zero.

So for instance, if the running count has increased by 2 after a quarter deck, then I know I'm one point over and I use my quarter-deck multiplier to determine the TC: (2-1) x .57 (I round to .6 for simplicity), so the TC is +.6. I have a multiplier for each quarter deck played: .6, .7, .8, 1, 1.3, 2.

So another example - If the running count has increased by 10 after 1.25 decks have been played (5 quarter decks), the TC is (10 - 5) x 1.3 = 6.5.

Pretty easy, No?

Just like Freightman and T3 point out, math has a different approach for everyone tackling problems like this. For example I handle TKO by performing the math described by Bigdaddy before even playing, constructing a table and memorizing it. So I know the RC and the deck we are playing and therefore I know the TC.

Easier for me, already used to it after many years, but not everyone's cup of coffee I am sure.