Poll

6 votes (26.08%)
10 votes (43.47%)
10 votes (43.47%)
1 vote (4.34%)
2 votes (8.69%)
4 votes (17.39%)
6 votes (26.08%)
3 votes (13.04%)
2 votes (8.69%)
3 votes (13.04%)

23 members have voted

sammydv
sammydv
Joined: Mar 25, 2016
  • Threads: 9
  • Posts: 607
January 26th, 2017 at 10:24:43 AM permalink
Wizard. I watched your vid playing the UPBS and I noticed that you were keeping non winning pairs quite a bit.
I thought with video poker, if a losing first draw including low pairs were laid, that the odds favored throwing all in for a 5 draw. If I'm correct, could it be because it's a multi deal that that theory/technique doesn't hold up?
Thanks.
That game didn't do too well in that run either unfortunately from the looks of it.
Wizard
Administrator
Wizard
Joined: Oct 14, 2009
  • Threads: 1388
  • Posts: 23315
Thanks for this post from:
mamat
February 9th, 2017 at 4:24:20 PM permalink
Quote: sammydv

Wizard. I watched your vid playing the UPBS and I noticed that you were keeping non winning pairs quite a bit.
I thought with video poker, if a losing first draw including low pairs were laid, that the odds favored throwing all in for a 5 draw. If I'm correct, could it be because it's a multi deal that that theory/technique doesn't hold up?
Thanks.



Sorry for the tardy reply. I had no idea what I was doing at that sitting so pretty much followed conventional video poker strategy. I'm sure I made similar errors in my example page for the game.

I'm happy to say my page on Ultimate X Bonus Streak is pretty much done. The mathematician with VideoPoker.com was nice enough to send me the list of games, pay tables, bonus streaks, and returns. I probably would have never figured out this game by myself. Perhaps the most complicated casino game to analyze I've ever seen.
It's not whether you win or lose; it's whether or not you had a good bet.
APIndy
APIndy
Joined: Jan 23, 2017
  • Threads: 0
  • Posts: 7
February 10th, 2017 at 4:36:55 AM permalink
Quote: I'm happy to say my page on Ultimate X Bonus Streak[/link

is pretty much done. The mathematician with was nice enough to send me the list of games, pay tables, bonus streaks, and returns. I probably would have never figured out this game by myself. Perhaps the most complicated casino game to analyze I've ever seen.



If you have done any work on UX, then it is pretty easy to set a lower bound for the return on UXBS and create a decent strategy, especially for 10-play. In ordinary UX, the average multiplier for the next hand is 1x very often, and the average multiplier is fairly high on a lot of hands. In UXBS 10-play, the average multiplier for the next hand is 1x less than 8% of the time, and the average multiplier only fairly high after you flop a big hand. Only 13% of hands have an average multiplier of more than 3x.

I.e., the variance of the average multiplier for the next hand is much lower. More hands are played with a middle-of-the-road multipliers, so a simpler strategy works better for UXBS than for simple UX. It also works better for games where the flush pays more, since you are going for more flushes in UX, generally.

Are the expected returns from the mathematician Monte Carlo or closed form values?
Wizard
Administrator
Wizard
Joined: Oct 14, 2009
  • Threads: 1388
  • Posts: 23315
February 10th, 2017 at 6:57:19 AM permalink
I started work on a single-strategy return but my analysis is unfinished. The returns are closed form.
It's not whether you win or lose; it's whether or not you had a good bet.
LuckyPhow
LuckyPhow
Joined: May 19, 2016
  • Threads: 53
  • Posts: 683
February 10th, 2017 at 11:36:29 AM permalink
Wiz,

Rule 4 has "the player" duplicated.

Quote: Rules, Item 4

... then the player the player will win up to five multipliers...



Lucky, Director
Bureau of Redundancy Bureau
stephen.hall
stephen.hall
Joined: Aug 3, 2015
  • Threads: 1
  • Posts: 31
February 10th, 2017 at 3:54:18 PM permalink
The DD coffee is my favorite.
APIndy
APIndy
Joined: Jan 23, 2017
  • Threads: 0
  • Posts: 7
February 10th, 2017 at 7:02:45 PM permalink
Quote: Wizard

I'm happy to say my page on Ultimate X Bonus Streak is pretty much done. The mathematician with was nice enough to send me the list of games, pay tables, bonus streaks, and returns. I probably would have never figured out this game by myself. Perhaps the most complicated casino game to analyze I've ever seen.



I can't understand how these pay tables can possibly be right. You are showing a massive decrease in multipliers for the 10-play versions.
Three-Play Return Table
Double Double Bonus 9-5 2,4 2,4,8 2,4,8,10,12 98.40%
Five-Play Return Table
Double Double Bonus 9-5 2,4 2,4,8 2,4,8,10,12 98.31%
Ten-Play Return Table
Double Double Bonus 9-5 2,3 2,3,4 2,3,4,8,12 98.36%

There must be a difference in the hands where these multiplier apply, otherwise the EV on the 10-play game would be around 86%.

Your old web page had: Three of a kind 2,4 2,4 2,3,4 for the STREAK1

I think the entry should look like this:
Ten-Play Return Table
Double Double Bonus 9-5 2,3,4 2,3,4 2,3,4,8,12 98.36%
Wizard
Administrator
Wizard
Joined: Oct 14, 2009
  • Threads: 1388
  • Posts: 23315
February 10th, 2017 at 7:38:04 PM permalink
Quote: APIndy

I can't understand how these pay tables can possibly be right.



You're absolutely right. I made a typographical error in the 10-play table. Each game has two different streaks only. Please have another look. Thank you for the correction.
It's not whether you win or lose; it's whether or not you had a good bet.
drrock
drrock
Joined: Mar 6, 2012
  • Threads: 3
  • Posts: 51
February 13th, 2017 at 6:02:02 PM permalink
I was pleased to see that you had another paper from Dr. Koehler on your game page for Bonus Streak. I know Dr. Koehler posts here from time to time. If possible, I would be interested in seeing a listing of what he terms the state biases and steady-state probabilities for the 3-Line version for one of the games that he has already analyzed. I would suggest one of the games with only 2 streak types, so that there are 12 1-line states, resulting in "only" 364 3-line states (rather than a 14-state version coming from a 3-streak type game that would need 560 3-line states). I was thinking that the 364 states could be compactly named by 12-digit numbers, with each digit being a placeholder for the 1-line state and the value in each digit being from 0 to 3 to indicate the possible number of each state; i.e., 000000000003, indicates all 3 hands being of 1-line state 1, 020010000000 indicating 2 hands with 1-line streak type 11 and the other one being of type 8, etc.

I offer a couple possible corrections to Dr. Koehler's paper, with one being a very minor typo. The 9-6 TDB game has the straight value of 5 listed instead of 4. I was able to replicate the 1-line EV for TDB 9-6-4, so I'm fairly certain that this should change. One I am not so certain about is the end value for the 9-5 DDB game. He is reporting an EV of 0.991220 for the 1-line version and 0.987923 for the 3-line version. I was able to replicate all of the EVs for the 5 or 6 1-line versions that I tried except this one. I could have made a mistake but can offer 0.987372531 as a possible answer. I do not presently have a facility for easily convolving video poker probabilities to multiple line versions, so I cannot offer any help on the 3-line version, but the Wizard's new page has 98.40%. I would note that the difference between my 1-line number and IGT's 3-line (0.34%) is fairly close to the difference in some other 1-line vs. 3-line comparisons, including the one that I am questioning. My calculations are essentially "by hand" in Excel, so there are lots of places where I could have erred.

Thanks in advance if you are able to provide any 364-state listing of biases and frequencies.
GaryJKoehler
GaryJKoehler
Joined: Oct 22, 2015
  • Threads: 5
  • Posts: 157
Thanks for this post from:
APIndytringlomaneonenickelmiracle
February 14th, 2017 at 3:49:04 AM permalink
Hi Drrork,
Well, between the Wiz, me and you, weíve now accounted for everyone who ever looked at this paper. LOL!

First, thank you. You indeed spotted two actual errors. The TDB table in the paper should have had a 4 for the Straight value (the code did so the results given were ok). The DDB game listed streaks of 2,4,8,10,12 but my code had 2,4,8,11,12. After changing the values in my code, the 1-Line EV is 0.987373, so your answer is in agreement.

Iím sending you the 364 lines of steady-state values for the Deuces Wild game so as not to clog the blog. If there is a simple way to attach a text or excel file, I can make it publically available.

As for encoding the state, that isnít the challenge. Rather it is the blow-up in size Ė the old curse of dimensionality. However, I am working on a new idea to shrink the state space to more manageable sizes (as was possible with Ultimate X).

At some point Iíll bother the Wiz to repost the paper with the changes and any new results that might come if I reduce the state space.

Again, thanks!

  • Jump to: