I'd like to point out math errors in the result's listed by Davinci Hand Analyzer. You will need to enter these hands to follow along. DaVinci AC AS. Michelanglo AH 6C. Flop KH 9H 4H. It lists Michelanglo has having 361 of 990 winning hands which isn't correct, the correct number would be 363 of 990. It lists the number of flushes at 360, which would be correct. Michelanglo can also win with 3 6's. There are 3 remaining 6's in the deck, there are 3 combinations of 2 6's that can come on the turn and river. That would be a total of 363 and not the 361 listed. I purposely choose a hand where the math would be very simple. I'm sure there must be other errors on hands where the math would be much more difficult to do. I welcome comments, and challenges to my math.

Sears Poncho

POnch, go get www.pokerstove.com and run some comparisons for us. Nice Catch.

__________________

No. There is only one way for him to get trip 6s and win. Specifically, 6d6s (order doesn't matter). The 6h makes a flush and can't ever result in trips. You are double counting those hands.

In other words, DaVinci's Hand Analyzer got it right.

-Jon

Just for the record, I agree with all of this.

The HandAnalyzer was just a free giveaway to test some underlying stuff while working on the main app(s). However, I do plan to update it with a lot more features soon.

-Jon

OK, I too have found a discrepancy here. I'm not sure who is right, and these numbers are almost insignificant, but there is a difference.

POKERSTOVE:

1,712,304 games 0.203 secs 8,434,995 games/sec

Board:
Dead:

equity (%) win (%) / tie (%)

Hand 1: 93.8582 % [ 00.93 00.00 ] { AsAc }
Hand 2: 06.1418 % [ 00.06 00.00 ] { Ah6c }


DAVINCI:

DaVinci: AcAs 93.37%
Michel: Ah6c 5.65%
Tie: 0.98%

According to Pokerstove, there is no ties. Now, somebodies got a logic or math error there. Maybe its one of them is not looking at all five cards. I don't know.

__________________

Well stop for a second and think about it...This one is easy to prove against PokerStove [ACTUALLY I just realized what PokerStove is doing and it isn't an real error, just a rounding/cutoff error in the printing - see my next post].

It is real easy to come up with examples that tie.

If the board is KdQdJd10d9d (for example), they both have an Ace high straight and it is a tie. It is almost impossible for there to be no chance of a tie before the flop has even happened...

Not sure what PokerStove is calculating wrong but I will stand by DaVinci's Hand Analyzer against all this proof...

I wouldn't bet my life on it but a heck of a lot of people have used it without finding a real error...But keep throwing 'em up and I will attempt to knock them down.

-Jon

Ok. A little more research and I think what is happening is that the labels on PokerStove are deceiving you (I think their labels are poor but this stuff is confusing which is why I worked on an easy UI) ...but their math is identical to mine.

Their first number is the EV or expected value of the hand. That is basically the win% + tie%/#splits or in this case 93.37 + 0.98/2 = 93.86. My currently released version doesn't show the EV but I have a version that does. This is obviously more significant if there are more than 2 hands...

I think their printing of the other values is just suffering from a rounding/cutoff error since they are not printing a percentage (in other words multiplying by 100). The first one should be 00.938 and the last one should be 00.0098 but they are just cutting the values off in the printing to 2 digits rather than truly rounding (e.g. 00.94 and 00.01).

In either case, the two programs agree to the last decimal place if you understand their values are cutoff rather than rounded and their first number is an EV...

-Jon