gordonm888
gordonm888
Joined: Feb 18, 2015
  • Threads: 20
  • Posts: 518
Thanks for this post from:
Zuga
May 7th, 2017 at 8:04:08 PM permalink
First of all, let me say that I LOVE the Ultimate Texas Holdem (UTH) calculator, written by J.B., on the WOO site HERE . I think I know what J.B. had to do to write this calculator and to do the calculations that are posted on the WOO UTH page -and I am in awe of how much work he did and how useful this material is. I am using this calculator everyday in my "research." Even though I am about to post some problems with the UTH calculator, I am pleading that you keep the current calculator on line until a revised version is available - I really am using it!

Okay, I have been using this calculator for only a few days but I have discovered what I believe are three completely different problems with the calculator.
1. Different results for same hand
2. Problems with combination math in evaluating options
3. House Edge calculation problem


I am posting the first issue (with screen captures) immediately and will then wait to make sure that either J.B or the Wizard is paying attention before I do all the work to post the next two issues.

Problem 1: Different results for the identical case

I have been using the calculator to study the 'Raise 2x or Fold' decision when the player's hand is J-x (off); x= 2-9 and the flop comes AKQ rainbow. It's interesting because its a case where you sometimes should raise 2X with a gut-shot straight draw. And this is what I am getting.

Player has Js5h and the flop comes AcKdQh.



That screenshot says Raise = -2.205525 and Check = -1.795540. I think that is a wrong result. The next screenshot shows the identical hand with Raise = -0.194818 and Check = -0.181275.




Now here are 3 screenshots with this hand: Player: Js4h Flop: AcKdQh. These 3 screenshots have these different Summary results for the same hand:
- Raise = - 1.838365 Check = - 1.403751
- Raise = +1.299305 Check = +0.878537
- Raise = - 0.564032 Check = - 0.543535







I suspect that the different answers are a function of the sequence used to identify the cards in the hand.
Last edited by: gordonm888 on May 7, 2017
JB
Administrator
JB
Joined: Oct 14, 2009
  • Threads: 334
  • Posts: 2053
Thanks for this post from:
Zuga
May 7th, 2017 at 11:07:38 PM permalink
I'm here, and just subscribed so I'll get an email when you reply.

I'll look into it. I suspect it's a caching issue, that is, your browser is caching the results from a previous scenario and displaying it as the results for the current scenario, since it pings a URL behind the scenes which performs the calculations and returns the results. If this is the case, I think I can fix it on our end.
JB
Administrator
JB
Joined: Oct 14, 2009
  • Threads: 334
  • Posts: 2053
Thanks for this post from:
Zuga
May 7th, 2017 at 11:12:17 PM permalink
First attempt at resolving the issue. Can you please clear your browser's cache, and try it again?
gordonm888
gordonm888
Joined: Feb 18, 2015
  • Threads: 20
  • Posts: 518
May 8th, 2017 at 10:16:48 AM permalink
I did that, and my cache was indeed full, but I am still seeing the problem with an empty cache.

I have only seen this problem occur when I am changing a card in the player's hand with a 3-card flop on the board.

This sequence of data entry makes it go haywire:

1. Put Js2h in players hand.
2. Put AcKdQh in the flop.
3. Remove 2h and replace with 3h (in the player's hand).
4. Remove 3h and replace with 4h.
5. Remove 4h and replace with 5h.
6. Remove 5h and replace with 6h.

Usually, by step 5 or 6, the calculator goes haywire. It seems to be triggered by successive replacements of a card in the player's hand while the flop is on board
gordonm888
gordonm888
Joined: Feb 18, 2015
  • Threads: 20
  • Posts: 518
May 8th, 2017 at 4:12:44 PM permalink
J.B.: Here is the 2nd problem:

This occurs on hands where only the player's 2 cards are defined and the calculator is providing information on the Raise 4X vs Check decision.

Sometimes, in this situation, the "Full Details for Raising 4X" table will display rows and combinations for scenarios in which the player Raises 2X. Here are two screenshots of a "Full Details for Raising" table in which "Raise 4X" should be the only player bet size in the list of scenarios:





In my experience, this defect in the "Full Details for Raising 4X" table occurs randomly -seemingly on 1 in 4-5 hands. You can input the same hands that I am showing in these screenshots and not get it because the erroneous table entries are not repeatable, but if you try a half dozen hands and scroll down the 'Full Details on Raising' table you should encounter it. This error is mathematically insignificant -amounting to about 1000 combinations out of 20 billion - but its kind of creepy that it happens at all - and the randomness of it makes me worry that something more fundamental could be wrong. Anyway, I thought you should be made aware of it.

I have also seen instances, when evaluating the initial 2-card hand, in which the total number of combinations shown in the Full Details tables was slightly different between the 'Raise 4X' and 'Check' options - and they certainly should be the same. But I Looked at it closely and the difference in total combinations seemed to be due to this same error -usually 990 combinations for Raise 2X scenarios being included in the Raise 4X combinations math.
gordonm888
gordonm888
Joined: Feb 18, 2015
  • Threads: 20
  • Posts: 518
May 12th, 2017 at 2:26:50 PM permalink
I am assuming/hoping that J.B is still working on trying to fix the UTH calculator. Here is another error I noticed -this one may help J.B. to narrow the possibilities.

With all 5 cards on the board and the decision being "Raise 1X or Fold", I saw a hand where the summary was:

Raise = -2.17xxxxx
Fold = 0.0

Clearly, the EV for fold should always be -2.0. So, this was clearly not the results to some previous case that has been pulled out of the cache.

I went into the "Full Details for FOLD" table; this full details table looked pretty normal except the last line also said Total 0.000 - i.e., the full details table was showing the same error.

I have been clicking NEW HAND to clear before all calculations and re-entering in all the cards for every hand I wish to evaluate - and I am still getting errors. I even get errors on the first hand immediately after reloading the calculator page onto my browser. The errors never occur on a 2 card hand -they occur after the 3-card flop and sometimes after all 5 common cards are dealt.

Good luck, J.B.
beachbumbabs
Administrator
beachbumbabs
Joined: May 21, 2013
  • Threads: 76
  • Posts: 9045
May 12th, 2017 at 3:08:14 PM permalink
Gordon,

I am positive JB appreciates the detailed bug reports, based on past troubleshooting. Please keep them coming.
"If the house lost every hand, they wouldn't deal the game."
JB
Administrator
JB
Joined: Oct 14, 2009
  • Threads: 334
  • Posts: 2053
May 12th, 2017 at 4:21:32 PM permalink
Thanks for the report. I haven't had much time to work on this yet but will definitely get to it.
gordonm888
gordonm888
Joined: Feb 18, 2015
  • Threads: 20
  • Posts: 518
May 13th, 2017 at 4:14:55 PM permalink
Here is another interesting glitch. This is a "Full Details for Raise 2X" table after the flop. The number of combinations (41,571 and 3,569) are correct for the the first two rows -the Royal flush rows. But then it all goes wrong. The next rows are for 4OAK and Full House Combinations -but it is impossible for the player to get either a 4OAK or a Full House in this situation.



On the fourth row down, when player has 4OAK and the Dealer wins -this is calculated to be 2 combinations. Its as if the calculator thinks the player has a pocket pair and could make 4OAK while the dealer has 2 possible ways to make a straight flush. Or something. I mean - exactly 2 combinations is a highly unusual result.

This is not a "repeatable" error. Every time I input these five cards -or change Spades to Clubs or whatever - I seem to get a different answer.

I am beginning to think we have the first case of an on-line calculator that is suffering from demonic possession. J.B. may need an Exorcist for this one.
JB
Administrator
JB
Joined: Oct 14, 2009
  • Threads: 334
  • Posts: 2053
May 24th, 2017 at 2:28:49 AM permalink
I'm happy to say that I believe this is fixed.

It all boiled down to a rookie mistake on my part. The problem couldn't occur during development/testing because I was the only one using it at the time. I had made the arrays which hold the results for an analysis be static (global) when they should have been instance (local).

So what happened was you were seeing someone else's results, or a hodgepodge of yours and theirs. Hence the intermittent nature. If nobody else was using the calculator at the time, it worked just fine. But if someone else entered a pre-flop hand at the same time you entered a post-flop hand, you might be seeing the results for their hand, or they might be seeing yours, or some of your results might have been combined with some of theirs, depending on the timing.

  • Jump to: