It's not hard to explain, obviously.

In the example I gave, where we're basically ranking (3-2-) against 3A, the simulator would seem to do this:

A2 -- valid hand
A3 -- NOT valid (second card not a deuce or less)
2A -- valid hand
23 -- NOT valid (second card not a deuce or less)
3A -- valid hand
32 -- valid hand

Since A2, 2A, or 32 can each happen 12 ways, and 3A can happen 9 ways, we end up with 45 possible hands -- 24 winners, 12 losers, and 9 ties. That matches up exactly with the output of the simulator.

The problem here is that the simulator is taking the (3-2-) literally, and throwing out all hands where the second hole card isn't a deuce or an Ace. The user most likely didn't intend for the two hole cards to be treated differently.

My suggested fix would be that the hole cards should be checked in BOTH orders against the rule. In other words, don't throw out A3, because 3A actually does match.

No. The second simulation makes a distinction between (3A)4 and (A3)4, and won't allow the second hand. Using that syntax forces a range heavier in 3A and 2A holdings by eliminating the possibility that the second hole card can be any higher than a Deuce.

In other words, the syntax (3-3-) means either card can be a 3,2 or A, as long as there are no pairs, while the syntax (3-2-) means that the first dealt card can be a 3,2 or A, but the second-dealt can only be a Deuce or Ace.

Edit: zing, you're quick. I don't consider this distinction a fail. I use this feature all the time to shade an opponent's range stronger when appropriate.

But the weighting is artificial. People don't play Razz differently based on whether they get dealt A3 in the hole or 3A.

For the simulation that I gave, you're basically saying that your opponent is likely to play all of his A2 hands this way, but only half of his A3 or 23 hands this way. And the 2:1 ratio is an artificial artifact of the algorithm; you didn't ask for a 2:1 ratio...

Based on posted simulations, I'd bet that 90% or more of the users of the simulator would be surprised by this "feature"...

Well okay, but a tool isn't busted just because people use it incorrectly. It took me a while to figure it out, but I like it now.

Thank you for the heads up, folks, now I see what you are talking about.

I think I agree that this 'feature' is not really a feature. I will look at fixing this up tomorrow morning.

- bachfan

Seriously, this is the most productive thread hijack I have ever seen.

Now I'm on Steve's ****-list for maybe screwing up his feature...

Don't sweat it. There's always the Razzmaster 2000.

The "feature" is mostly useful with wider ranges anyway. Let's say you have a moderately-loose villain and you want to assign him a range: Any 3-card Seven or better and sometimes 3-card Eights. You can use a range of (8-7-6). This means that only half of his Eights are counted, which shades the range appropriately toward better hands, though not perfectly.

A better way to do it would be to make the range (8-5-6) + (5-8-6), which means he's only playing 865 or better. The other syntax allows him to play half his 876s.

Not a list you wanna be in the top 40 of, lol. Btw, I didn't mean that the tool was a fail because of this issue, but that users were often getting failed results without realizing. I'd be just as happy with a warning on the web page as changing the tool. Steve's proposed fix is the best option of all ldo.

Well, I went over my razz code this morning. That was fun. Feeling the usual mix of pride/horror at looking at old code. I have a tentative fix for the problem, but I want to get some code cleanup/optimizations done while I'm in there.

I'm on twitter.com/propokertools if you want to know when this is complete.

Thanks for your patience,
bachfan

In the process of working on this, I've come across some interesting issues around generating razz hands with the 'correct' probabilities. I'll make a separate post about it or post to my blog when my thinking on this is clearer.

My brain hurts so I'm going to take a breather - my poor family must think I'm crazy.

Thanks,
bachfan

update plz after brain swelling subsides. kthx

I am at the point where I know what I need to do with the razz simulator. Now I just need to find time to do it. In order to prevent complete brain explosion, I am likely to complete the bulk of the work on my beta -> main site migration project first, then go on to this razz issue, and then finally continue work on the downloadable simulator.

Cheers,
bachfan

Working on this right now. Should have something up tomorrow or by the end of the week at the latest.

My fix for the razz 'ordering' problem is now online. I posted about it here:

http://pokercoder.blogspot.com/2010/04/fixing-razz.html

Thanks all, for bringing this to my attention.

- bachfan, ProPokerTools dude

P.S. Sorry electrical for breaking your 'feature'.