ssho88
ssho88
Joined: Oct 16, 2011
  • Threads: 45
  • Posts: 501
March 1st, 2021 at 10:06:10 PM permalink
Hi,


How to find covariance of a game ?

There is an example how to find covariance of Blackjack at this link https://wizardofodds.com/games/blackjack/variance/

Firstly, find total variance of N hands per round, and use the formula,

Total variance of N hands per round= N* Variance of 1 hand per round+ N * (N-1) * covariance.

Fro example, variance of 1 hand per round = 1.31 and Total variance of 2 hands per round(through simulations) = 3.565, N = 2,

So, 3.565 = 2 * 1.31 + 2(1)* covariance

covariance = (3.565 - 2.62) /2 = 0.4725.

My question : Is simulations the only way to find total variance of N hands per round ?




Regards

ssho88
ThatDonGuy
ThatDonGuy
Joined: Jun 22, 2011
  • Threads: 97
  • Posts: 4628
March 2nd, 2021 at 7:11:27 AM permalink
My gut answer is, if you can calculate the variance of 1 hand per round, you can calculate the variance of multiple hands per round, but the number of possible deals for even 2 hands per round may be overwhelming.
Wizard
Administrator
Wizard
Joined: Oct 14, 2009
  • Threads: 1367
  • Posts: 22723
March 2nd, 2021 at 1:01:07 PM permalink
I did my covariance articles, including that on video poker by simulation.

Perhaps in multi-card keno one could do an exact analysis. Perhaps I'll work on that.
It's not whether you win or lose; it's whether or not you had a good bet.
CrystalMath
CrystalMath
Joined: May 10, 2011
  • Threads: 8
  • Posts: 1881
March 2nd, 2021 at 2:30:22 PM permalink
Quote: Wizard

I did my covariance articles, including that on video poker by simulation.

Perhaps in multi-card keno one could do an exact analysis. Perhaps I'll work on that.



But in keno, it really depends on the overlap of the marked spots. I know people who mark 5 or 6 spots the same on 20 cards and then use 1 or 2 other spots to spread around the screen. In this case, the covariance is really high. Conversely, you could do a quick pick which would give much lower covariance.
I heart Crystal Math.
ssho88
ssho88
Joined: Oct 16, 2011
  • Threads: 45
  • Posts: 501
March 2nd, 2021 at 4:41:31 PM permalink
Thanks for your replies.
DRich
DRich
Joined: Jul 6, 2012
  • Threads: 75
  • Posts: 7186
Thanks for this post from:
CrystalMath
March 2nd, 2021 at 6:13:58 PM permalink
Quote: CrystalMath

Quote: Wizard

I did my covariance articles, including that on video poker by simulation.

Perhaps in multi-card keno one could do an exact analysis. Perhaps I'll work on that.



But in keno, it really depends on the overlap of the marked spots. I know people who mark 5 or 6 spots the same on 20 cards and then use 1 or 2 other spots to spread around the screen. In this case, the covariance is really high. Conversely, you could do a quick pick which would give much lower covariance.



I was once out with my boss and we were gambling pretty heavy as partners on video poker. We were down a couple thousand dollars and he asked me for $100 to play video keno. As a numbers guy I was offended that he would play a 9% hold game. He put in his $100 and played four card keno with the same seven numbers on all four cards. He hit for $28,000 and just laughed at me for not giving him $100. He then proceeded to ask me to sign the W2-G and I got the last laugh and it only cost me $14,000.
Living longer does not always infer +EV
USpapergames
USpapergames
Joined: Jun 23, 2020
  • Threads: 18
  • Posts: 807
March 2nd, 2021 at 11:59:42 PM permalink
Quote: ssho88

Hi,


How to find covariance of a game ?

There is an example how to find covariance of Blackjack at this link https://wizardofodds.com/games/blackjack/variance/

Firstly, find total variance of N hands per round, and use the formula,

Total variance of N hands per round= N* Variance of 1 hand per round+ N * (N-1) * covariance.

Fro example, variance of 1 hand per round = 1.31 and Total variance of 2 hands per round(through simulations) = 3.565, N = 2,

So, 3.565 = 2 * 1.31 + 2(1)* covariance

covariance = (3.565 - 2.62) /2 = 0.4725.

My question : Is simulations the only way to find total variance of N hands per round ?




Regards

ssho88



What a great question. The answer is no. I don't use computers to solve any math for me except computation & I still always get the numbers right. What your asking for is extremely difficult to teach and requires years of game analyze to identify the mathematical relationships between game mechanics and rules. I can analyze ANY game as long as I am provided the data for it (video games can hide the math behind them very well). Just some examples of the types of data I can get on anything related to: Variance, Probability, Volatility, Standard Deviation, Game Theory, Expected Revenue, & believe it or not even a Skill, Strategy, Luck ratio analysis which I invited since I made some major discoveries in the science of game design. No joke, I can tell you exactly how much luck there is in any game the only thing is I do all my math work by hand that its not worth the effort for me to over analyze. If the game is simple then yes getting the exact value of luck in the game could be important. But any good game is going to be difficult for any player to play perfectly so the amount of perceived luck in a game is often much higher than the actual amount of luck. Which is why I've learned that the ability to calculate estimates that are within your acceptable tolerance is the better approach in the long run. 1st, focus on the game analysis skills then I'll teach you some tricks for getting accurate estimates that way you will intuitively be able to get a sense of any games variance just from experience because your not spending a lot of time of getting the exact numbers. Game design is like engineering, it only needs to be 97% right to get it into production.
Math is the only true form of knowledge
gordonm888
gordonm888
Joined: Feb 18, 2015
  • Threads: 44
  • Posts: 2809
March 3rd, 2021 at 7:02:54 AM permalink
I am interested in this covariance question as well.

In BJ, the dealer's hand is nominally 1/2 of what determines the outcome so the covariance when playing multiple hands against a dealer must be high. I'm not sure what the metric is for covariance, though.

I also struggle with questions such as: if you randomly partition n objects into 4 piles, what is the variance in the size of the piles, or variance in the average deviation of the piles. Clearly the size of the 4th pile is determined by the size of the three previous piles, so there is covariance in this problem.
So many better men, a few of them friends, are dead. And a thousand thousand slimy things live on, and so do I.
marcel55
marcel55
Joined: Feb 21, 2021
  • Threads: 0
  • Posts: 18
March 3rd, 2021 at 9:16:33 AM permalink
Quote: Wizard

I did my covariance articles, including that on video poker by simulation.

Perhaps in multi-card keno one could do an exact analysis. Perhaps I'll work on that.


Well I check that out but there is something missing I think so.
USpapergames
USpapergames
Joined: Jun 23, 2020
  • Threads: 18
  • Posts: 807
March 3rd, 2021 at 10:50:18 AM permalink
Quote: gordonm888

I am interested in this covariance question as well.

In BJ, the dealer's hand is nominally 1/2 of what determines the outcome so the covariance when playing multiple hands against a dealer must be high. I'm not sure what the metric is for covariance, though.

I also struggle with questions such as: if you randomly partition n objects into 4 piles, what is the variance in the size of the piles, or variance in the average deviation of the piles. Clearly the size of the 4th pile is determined by the size of the three previous piles, so there is covariance in this problem.



It's funny you mention that because combinatorics plays a HUGE role in determining variance and covariance.
Math is the only true form of knowledge

  • Jump to: