What Would It Take???

[Frank Kneeland]

Am I to assume that everyone would like to track Four of a Kind frequency?If so, it would be best to track each rank independently as that gives us more data points.Here's the problem, 4K frequency can be effected by strategy. I had wanted to make the method strategy independent. Does anyone have an idea how to achieve both goals???1. Tracking 4K Frequency2. Removing the effects of strategy~FKP.S. And where the heck is new2VP when we need him? Is he OK?

[moneyla]

Okay Frank, let me make a suggestion.  As you point out, the "final hand" will always be affected by the strategy of the holds, and of course that will vary by game. So if you want to test "randomness" all you can do is record and chart dealt hands to see if all 52/53 cards come up in a random distribution on the initial deal. Over X number of hands, all 52/53 cards should appear equally.  After the initial deal, any measure of randomness gets thrown out the window by the way the hand is played and the strategy used on any particular game.

[ko king]

They are Class lll, I wasn't paying attention to how many times I pushed the "l" button.

[ko king]

ko king,
 
If you are suspicious then maintain detailed records. It appears you have taken a couple of "shots" at doing this. However, it is easy to be fooled once you think you have seen a pattern. Your mind will start looking for it. The only way to avoid it is to keep exact records.
 
Once you keep records on this for something like 10-20 thousand hands you should be able to determine if your suspicions were correct. Prove it to yourself.
 
Here's one way to accomplish this. Bring a 3x5 index card with you. Keep track of 4Fl and Flushes hit for each hour. Mark them down on the index card. Compare the number to the total number of hands played. Do this for 10 trips.
 
I could keep the numbers in my head but there are counter devices you could purchase to make it easier. I know someone that used to place nickels/dimes/pennies in his shirt pocket each time an event he was tracking occurred. He'd remove them each hour, log the results and start again.





I've thought about that "getting fooled" part and won't deny it's possible so I agree the best way to keep up with it may be the index card.

[ko king]

I agree.
 
I also applaud the effort, and concur that the poster who couldnt read your original post probably has little communications skills when he writes anything anywhere.

Were you refering to me?

[BillyJoe]

I'm going to save people a ton of time by pre-packaging the tools for testing. I'm not responsible for what they test or their conclusions.

~FK
 
Frank, I think it is great that you are willing to help develop a tool for us to use for this. In my case, as I have said, I would like the tool to deal with the infrequent SFs, RFs and Quads. Although I have an MBA (not self-exalting, just a fact), I am not a statistics wiz, so I do not know how to best parse a sample for the significance to reach 99% certainty, which is what you are looking for out of the tool. I hope that someone can answer that question, because I would really love to use the finished product.  

[BillyJoe]

Am I to assume that everyone would like to track Four of a Kind frequency?

If so, it would be best to track each rank independently as that gives us more data points.

Here's the problem, 4K frequency can be effected by strategy. I had wanted to make the method strategy independent. Does anyone have an idea how to achieve both goals???

1. Tracking 4K Frequency
2. Removing the effects of strategy

~FK

P.S. And where the heck is new2VP when we need him? Is he OK?

I would offer that the only way to do this effectively is to 'fix' the strategy for drawing quads, such as "only hold pairs when they are dealt". This, of course, would fly in the face of two pair FH draws, and 4 to the SF/RF with a pair SF/RF draws. I really don't know..

[Frank Kneeland]

Okay Frank, let me make a suggestion.  As you point out, the "final hand" will always be affected by the strategy of the holds, and of course that will vary by game. So if you want to test "randomness" all you can do is record and chart dealt hands to see if all 52/53 cards come up in a random distribution on the initial deal. Over X number of hands, all 52/53 cards should appear equally.  After the initial deal, any measure of randomness gets thrown out the window by the way the hand is played and the strategy used on any particular game.Eliminating the effects of people's choice and errors is a big concern. Depending on how people play and how they draw it would skew any system that looked purely at the frequency of final made hands. I agree that looking only at dealt card frequency would be one way to eliminate the bias, but I don't think people would be happy with the information it provided them, which would be limited to dealt cards.I'm envisioning a blending of a test for dealt card frequencies and draws.For the draws, we'd have to disconnect it from base rate hand frequencies and instead go off only the times a person actually holds certain cards and draws to them.Put another way, it would be fine to track how often one got a RF drawing to 3RF, as long as how often they drew to 3RF was disconnected from how often they got dealt 3RF.The dealt hand frequencies, and the draw and make a hand counters, would be separate.I believe this is doable. Good suggestions!~FK

[Frank Kneeland]

I've thought about that "getting fooled" part and won't deny it's possible so I agree the best way to keep up with it may be the index card.I just had dinner with two couples from Canada that have been trying AP VP for the first time. They are newly retired and in their first year a of serious play. They had a beautiful index card system printed out of MS Excel to track a whole bunch of things and I believe I can use that as a template for what we are trying to do.They were only tracking made hands and RF draws, so it'll need a little tweaking, but I now have a good start on the book keeping side of the project.Their index card is only for NSUD Deuces, so again it'll need some modification.I'll post it when I have it finished.

[Frank Kneeland]

[QUOTE=Frank Kneeland] I'm going to save people a ton of time by pre-packaging the tools for testing. I'm not responsible for what they test or their conclusions.~FK 
Frank, I think it is great that you are willing to help develop a tool for us to use for this. In my case, as I have said, I would like the tool to deal with the infrequent SFs, RFs and Quads. Although I have an MBA (not self-exalting, just a fact), I am not a statistics wiz, so I do not know how to best parse a sample for the significance to reach 99% certainty, which is what you are looking for out of the tool. I hope that someone can answer that question, because I would really love to use the finished product.  [/QUOTE]In a book I read on probability math I remember it stating that by averaging Ten 100 trail samples you get a higher level of certainty than if you used the average of One 1000 trail sample. I also remember it saying that depending on the probability of the thing being tested there is a sweet spot for maximum accuracy in minimum trials. For instance One Thousand 1 trial samples would not be better:)I have no idea what the sweet spot would be for VP or how to calculate it, or even what math to use to calculate it.Where is new2vp?