Norm Wattenberger: Effect of Color Dependent Tag Values

[Norm Wattenberger]

There have been a few comments lately about the effect of counting Red7's as +1 instead of counting all Sevens as +.5. Clearly there must be some difference in accuracy. But the comments have characterized the difference from insignificant to large. So, I though I'd run some sims. I ran two sims each for three sets of circumstances as follows:

[*]6D, S17, DAS, 1-15, 4.5/6, Accurate catch 20 indexes[*]6D, S17, DAS, 1-15, 4.5/6, Compromise indexes from BiB second edition[*]SD, H17, nDAS, 1-4, 58% pen, Accurate catch 20 indexes[/list]

Below are the resulting SCOREs:



Red7 Level II (counting all Sevens as +0.5) won in all three case; which is hardly a surprise. The damage caused by using the Red Sevens only for the three cases was:

6D accurate: 2.6%
6D compromise: 3.2%
SD accurate: 1.7%

I'm a bit surprised that the compromise indexes were more impacted than the accurate indexes. I really didn't know how the SD vs. 6D numbers would come out.

As to the question of whether these are significant differences; that's up to the reader. My own personal opinion is that these are not big differences.

[Don Schlesinger]

> I'm a bit surprised that the compromise indexes were
> more impacted than the accurate indexes.

You heap one sloppiness on top of another, and you compound the problem.

>I really didn't know how the SD vs. 6D numbers would come out.

Clear to me that SD should be the least affected. You can't be "wrong" by too much, with only four sevens in the deck. With 24 of them, there is more room to get out of whack.

Don

[Cacarulo]

> There have been a few comments lately about the effect
> of counting Red7's as +1 instead of counting all
> Sevens as +.5. Clearly there must be some difference
> in accuracy. But the comments have characterized the
> difference from insignificant to large. So, I though
> I'd run some sims. I ran two sims each for three sets
> of circumstances as follows:

> 6D, S17, DAS, 1-15, 4.5/6, Accurate catch 20 indexes
> 6D, S17, DAS, 1-15, 4.5/6, Compromise indexes from
> BiB second edition
> SD, H17, nDAS, 1-4, 58% pen, Accurate catch 20
> indexes Below are the resulting SCOREs:

> Red7 Level II (counting all Sevens as +0.5) won
> in all three case; which is hardly a surprise. The
> damage caused by using the Red Sevens only for the
> three cases was:

> 6D accurate: 2.6%
> 6D compromise: 3.2%
> SD accurate: 1.7%

> I'm a bit surprised that the compromise indexes were
> more impacted than the accurate indexes. I really
> didn't know how the SD vs. 6D numbers would come out.

> As to the question of whether these are significant
> differences; that's up to the reader. My own personal
> opinion is that these are not big differences.

I thought the differences would be larger but I think it depends on other factors like pen, spread and # of indices.
May I bother you with a couple of sims? Pen = 5/6 and C22 for example. Also, it would be good to check 1-20 and see what the difference is. Maybe you can just post the raw data.
We should don't forget what the correlations parameters are in both cases:

DRED-7
BC = 0.9800
IC = 0.7852
RED-7
BC = 0.9683
IC = 0.7757

These differences should be best seen with a 1-20 spread.

Thanks again for the interesting sims.

Sincerely,
Cac

[Cacarulo]

I've made some changes into my simulator and tonight I'll post my own sim comparison. I'll use the following rules just in case you wanted to check the SCOREs against yours: 6D,S17,DOA,DAS,SPA1,SPL3,NS,5/6,C22,5000 million rounds and heads-up. The systems to be compared are:

1) Red-7

 A = -2 

 2 =  2 

 3 =  2 

 4 =  2 

 5 =  2 

 6 =  2 

R7 =  2 

B7 =  0 

 8 =  0 

 9 =  0 

 T = -2

2) Double Red-7

 A = -2 

 2 =  2 

 3 =  2 

 4 =  2 

 5 =  2 

 6 =  2 

 7 =  1 

 8 =  0 

 9 =  0 

 T = -2

Both will use the same set of indices which I calculated for Double Red-7.
Also, I will post the differences between these two systems but simulated in TC mode. Again, the indices used will be the ones calculated for Double Red-7 in TC mode.

Sincerely,
Cac

[Norm Wattenberger]

I'm running this for 1-20 now. Looks like it will come out with greater than a 4% difference. But it's only 40% done.

> I've made some changes into my simulator and tonight
> I'll post my own sim comparison. I'll use the
> following rules just in case you wanted to check the
> SCOREs against yours:
> 6D,S17,DOA,DAS,SPA1,SPL3,NS,5/6,C22,5000 million
> rounds and heads-up. The systems to be compared are:
> 1) Red-7
> A = -2
> 2 = 2
> 3 = 2
> 4 = 2
> 5 = 2
> 6 = 2
> R7 = 2
> B7 = 0
> 8 = 0
> 9 = 0
> T = -2
> 2) Double Red-7
> A = -2
> 2 = 2
> 3 = 2
> 4 = 2
> 5 = 2
> 6 = 2
> 7 = 1
> 8 = 0
> 9 = 0
> T = -2
> Both will use the same set of indices which I
> calculated for Double Red-7.
> Also, I will post the differences between these two
> systems but simulated in TC mode. Again, the indices
> used will be the ones calculated for Double Red-7 in
> TC mode.
> Sincerely,
> Cac

[Cacarulo]

> I'm running this for 1-20 now. Looks like it will come
> out with greater than a 4% difference. But it's only
> 40% done.

That's what I'm getting but contrary to what I've said it turns out that lower spreads make these percentages larger.

Sincerely,
Cac

[Cacarulo]

1) Red Seven (counting R7's as +1 & B7's as 0) in RC mode

 play-all | rounds played = 100.00% 

 spread   ev/h   sd/h   ror%      n0    di  score      ekb   avb    unit  kelly 

 ------------------------------------------------------------------------------ 

 1 -  4  0.736 23.818  13.53  104688  3.09   9.55   770.66 1.664  12.976  1.000 

 1 -  8  1.895 37.559  13.53   39301  5.04  25.45   744.59 2.193  13.430  1.000 

 1 - 12  2.888 49.611  13.53   29507  5.82  33.89   852.19 2.626  11.734  1.000 

 1 - 15  3.604 58.542  13.53   26387  6.16  37.90   950.94 2.943  10.516  1.000 

 1 - 16  3.833 61.403  13.53   25663  6.24  38.97   983.66 3.044  10.166  1.000 

 1 - 20  4.744 72.916  13.53   23621  6.51  42.34  1120.64 3.450   8.923  1.000

2) Red Seven (counting all 7's as 0.5) in RC mode

 play-all | rounds played = 100.00% 

 spread   ev/h   sd/h   ror%      n0    di  score      ekb   avb    unit  kelly 

 ------------------------------------------------------------------------------ 

 1 -  4  0.759 23.817  13.53   98370  3.19  10.17   747.00 1.664  13.387  1.000 

 1 -  8  1.932 37.423  13.53   37520  5.16  26.65   724.88 2.189  13.795  1.000 

 1 - 12  2.948 49.556  13.53   28267  5.95  35.38   833.16 2.628  12.002  1.000 

 1 - 15  3.667 58.328  13.53   25301  6.29  39.52   927.77 2.941  10.778  1.000 

 1 - 16  3.904 61.244  13.53   24615  6.37  40.63   960.85 3.044  10.407  1.000 

 1 - 20  4.817 72.528  13.53   22671  6.64  44.11  1092.02 3.445   9.157  1.000

3) Red Seven (counting R7's as +1 & B7's as 0) in TC mode

 play-all | rounds played = 100.00% 

 spread   ev/h   sd/h   ror%      n0    di  score      ekb   avb    unit  kelly 

 ------------------------------------------------------------------------------ 

 1 -  4  0.754 23.705  13.53   98786  3.18  10.12   745.04 1.669  13.422  1.000 

 1 -  8  1.843 35.630  13.53   37370  5.17  26.76   688.78 2.110  14.518  1.000 

 1 - 12  2.766 46.095  13.53   27775  6.00  36.00   768.20 2.472  13.017  1.000 

 1 - 15  3.409 53.570  13.53   24688  6.36  40.51   841.70 2.727  11.881  1.000 

 1 - 16  3.618 56.010  13.53   23969  6.46  41.72   867.14 2.811  11.532  1.000 

 1 - 20  4.415 65.370  13.53   21928  6.75  45.61   967.98 3.129  10.331  1.000

4) Red Seven (counting all 7's as 0.5) in TC mode

 play-all | rounds played = 100.00% 

 spread   ev/h   sd/h   ror%      n0    di  score      ekb   avb    unit  kelly 

 ------------------------------------------------------------------------------ 

 1 -  4  0.776 23.677  13.53   93189  3.28  10.73   722.79 1.668  13.835  1.000 

 1 -  8  1.883 35.628  13.53   35787  5.29  27.94   673.98 2.112  14.837  1.000 

 1 - 12  2.817 46.019  13.53   26695  6.12  37.46   751.87 2.472  13.300  1.000 

 1 - 15  3.469 53.467  13.53   23752  6.49  42.10   824.01 2.728  12.136  1.000 

 1 - 16  3.679 55.870  13.53   23067  6.58  43.35   848.53 2.810  11.785  1.000 

 1 - 20  4.488 65.222  13.53   21120  6.88  47.35   947.85 3.128  10.550  1.000

Take a look at the differences in percent. The ranges go from around 4% to 6%!

Sincerely,
Cac

[Don Schlesinger]

> Take a look at the differences in percent. The ranges
> go from around 4% to 6%!

I am not the least bit surprised by these results. To me, it was axiomatic that counting all the sevens is clearly superior to counting half the sevens as twice the amount. The latter leads to greater fluctuation or results, which is always bad.

If you increase s.d., you lower SCORE.

Don

[Cacarulo]

> I am not the least bit surprised by these results. To
> me, it was axiomatic that counting all the sevens is
> clearly superior to counting half the sevens as twice
> the amount. The latter leads to greater fluctuation or
> results, which is always bad.

> If you increase s.d., you lower SCORE.

K-O and even Hi-Lo are better than Red Seven!

Conditions: 6D,S17,DOA,DAS,SPA1,SPL3,NS,5/6,heads-up,5000 million rounds

1) KO + C22 (exact EM-indices)

 play-all | rounds played = 100.00%  

 spread   ev/h   sd/h   ror%      n0    di  score      ekb   avb    unit  kelly  

------------------------------------------------------------------------------  

 1 -  4  0.651 22.192  13.53  116043  2.94   8.62   755.97 1.548  13.228  1.000  

 1 -  8  1.728 34.199  13.53   39179  5.05  25.52   676.93 1.966  14.773  1.000  

 1 - 12  2.664 45.066  13.53   28621  5.91  34.94   762.41 2.316  13.116  1.000  

 1 - 16  3.531 55.353  13.53   24575  6.38  40.69   867.73 2.642  11.524  1.000  

 1 - 20  4.352 65.203  13.53   22446  6.67  44.55   976.87 2.950  10.237  1.000

Sincerely,
Cac

[Norm Wattenberger]

> K-O and even Hi-Lo are better than Red Seven!

An interesting result. One can debate whether counting only Red 7's is difficult or not. But it is certainly more difficult than counting all 7's

[Parker]

>
> K-O and even Hi-Lo are better than Red Seven!
> Conditions:
> 6D,S17,DOA,DAS,SPA1,SPL3,NS,5/6,heads-up,5000 million
> rounds

5/6 is pretty good pen. The "conventional wisdom" (whatever that means) says that, while KO may be better in deeply penetrated games, Red 7, with its lower pivot point, should do better in the games with mediocre or worse pen that are so common today.

Care to run some sims to confirm/refute this?

I know you've got nothing better to do. :-)

[Cacarulo]

> 5/6 is pretty good pen. The "conventional
> wisdom" (whatever that means) says that, while KO
> may be better in deeply penetrated games, Red 7, with
> its lower pivot point, should do better in the games
> with mediocre or worse pen that are so common today.

Yes, that's what I heard too but we should agree that the "conventional wisdom" is based on a false premise (R7 = DR7).

> Care to run some sims to confirm/refute this?

Please, specify "Mediocre" or "Worse" penetration that you consider to be common today. 3.5/6? 4/6? The problem is that in order to be precise I need to calculate indices for an specific pen.

> I know you've got nothing better to do. :-)

No problem, I'll try to do it at 3:00 AM during my usual insomnia :-)

Sincerely,
Cac

[Norm Wattenberger]

> Yes, that's what I heard too but we should agree that
> the "conventional wisdom" is based on a
> false premise (R7 = DR7).

John Kenneth Galbraith had much to say about conventional wisdom and the problems caused by such.