So I started playing around with propokertools and am finding something odd I'm hoping you guys can explain.

AAQJr is at top 3% hand 10-handed according to ppt.

Now when I match it up 4-ways vs.:

1-15%
5-20%
10-20%

I'm getting 24% equity. So why is it ranked a top 3% 10-handed when it only has 25% equity vs. hands that are weaker ranges than 3%?

What I'm trying to actually figure out is how strong AABB is in general. I was shocked to find it ranked 3%, as the equities were about what I expected against reasonable ranges. Is this the Ace-removal issue? If so, is it really a top 3% hand then? The notes I read about the rankings where that they were basically derived by playing better and better hands against each other iteratively until the best hands were determined.

Ahhhh...statistics

Don't confuse comparative strength (top 3%) with absolute strength (24% equity)

regards

Gar

I don't really have the answer, but some immediate thoughts:

(1) Against 3 hands near the top, there is a very high chance all the aces are out. This is a disadvantage to our hand. It makes it less likely we get a house etc.

(2) Its not that likely to have 4 hands occurring at once from the top percentiles.

(3) Against these particular hands, if a low is possible almost certainly somebody has one.

All these things can apply 10 handed versus random hands as well, but in a less concentrated way (if that makes sense).

Oh, and I get this:


Omaha Hi/Lo Hand Ranking (unweighted) ?
Rankings for Ac Ad Qc Jh

Ranking Description Average Best Worst
10H 10-handed iterative (default) 6.0 6.0 6.0
3H 3-handed iterative 13.0 13.0 13.0
6H 6-handed iterative 8.0 8.0 8.0
VR vs. random hand 16.0 16.0 16.0

Assuming I am interpreting this right.......

AAQJds (best) is ranked 3%
AAQJns (worst) is ranked 9%

it's debateable whether we suffer from others sharing the two remaining aces among them. they are the ones who are dominated for high.

AAK9ns is ranked 15%
AAAA is ranked 32%

PPT is *overranking* aces.

one last thought. PPT O8 rankings aren't made specifically for LO8. if AA tend to play better in big bet O8, and I think they do, there's one explanation for you...

Maybe this will help...

find out how many hands have more equity and how many hands have less equity.

Add the 2 numbers, [sanity check:should be equal to C(52,4) (total number of possible hands)]

divide the 1st number by the total, should be a very small number, somewhere very near .03.

Try it a different way...

Imagine a group of 100 people of varying physical strength. Further, let's say that the absolute strength has a range of 75 to 100 evenly distributed

To be a top 3% means 3% of the group, or 3 people have equal or greater strength. so, say...99.25 or better puts you in the top 3%

now, let's say that in a contest, the chances of winning (EV) is porportional to the difference in strength. If you take the strongest guy (100) (top 1%) and match him up against the weakest guy (75) (bottom 1%), the strongest guy is only going to win 25% of the time (100 - 75)

crude illustration, maybe, but that's the principle. you have to take into account the range of the variable's value along with the ranking of a specific value in the range.

Regards

Gar

Good points.

I took the r to imply 'rainbow'.

It all depends on the arena (flop). In some arenas, the weak guy is at an advantage. In most arenas, the strong guy is at an advantage.

I don't get why the strong guy would only win 25% of the time against the weak guy.

I think you would basically need to write your own propokertools program to do this.

Also, from what I read on the ranking system, it's fairly simple and just sims hands against random hands, then takes the stronger ones and sims them against each other, and it does this over and over until all of the hands are ranked. So basically of a survival of the fittest repetition.

If this is true, then our sims kinda suck, don't they? Because we often put in top 15% or 10% or whatever as a range for villains, but if it overweights AA, then the calculations aren't going to be accurate.

PPT is the best out there, and perhaps a 32% for AAAA is justified. I don't know. simulations are good for what they are. they help us to master poker, not to master simulations

IMO the problem with drawing a conclusion on how accurately the hand rankings rank AAQJ from the sim you ran is the SIM.

when dealt 10 handed and 1 player is dealt AAQJ 3 other players are dealt a top 20% hand only 2.8% of the time (as simulated using a PPT simulation engine)in 600000 deals

i assume its because so many of the top ranked hands utilize the same cards. Aces and Kings and Queens, and the AAQJ removes 2 of the 4 aces and 1 queen.
**there are 54130 hands in the top 20% hands, only 3858 don't have an ace
there are 54130 hands in the range top 20% hands, remove AAQJ each of a distinct suit and the top 20% range is reduced to 23322 hands, and the non-ace hands are reduced from 3858 to 3282

when holding AAQJ, 2 top 20% hands are dealt to among the other 9 25% of the time, and just 1 top 20% 50%.

as the program that ranked the hands dealt the cards out randomly, it had to be very rare that AAQJ found itself in a situation where it was against 3 top 20% hands.

additional thoughts:
(IMO) i don't think "survival of the fittest" represents the concept used in determining the ranks very well, although it did involve elimination.

the link: http://pokercoder.blogspot.com/2006/...-of-hands.html

when using the PPThand ranking percentages in a sim such as top 15%, it should be clear that it probably never represents the 15% that someone would choose to play if they only play 15%. its a gross simplification of an individual player's range. for instance 6655 doublesuited is a top 15% hand A5KQ monosuited is not. I'm thinking more people choose to play the A5KQm before the 6655ds

That's exactly right. If we simulate against 9 random hands, we only manage 3 other top 20% hands about 2.75% of the time.


ProPokerTools Odds Oracle Results (2.16 Professional)
Omaha Hi/Lo, Generic syntax
PLAYER_1 AAQJ
PLAYER_2-10 ****
600000 trials (randomized)


How many players match hand range 20%


As has been pointed out before, we need to be careful when using hand rankings - they are really only a rough approximation and should be treated as such. That said, I have seen many cases where an intricately-crafted pre-flop range matches a rough percent-of-hands guess very closely. YMMV.

Cheers,
- Dan @ PPT

This isn't merely technically difficult -- it's conceptually meaningless. Hands don't "have equity" in a vacuum, as they might in video poker or perhaps Caribbean stud, where quads pays a certain amount and a full house pays a certain amount, etc. Omaha hands only have equity when playing against other hands.

[EDITED because now I realize that link above explains the PPT algorithm.]

I'll have to look at that link more carefully when I'm fully awake. On first glance, in the top 15 I see it ranks (A2)(A2) above (A3)(A4) and (A3)A2, which I'm pretty sure any experienced LO8 player would consider absurd.

it's hard to miss the flop with (A2)(A2) - when the board is low, you either have nut low+good high or a set.
while a set of deuces is by no means the dream hand, it is at least playable in most cases.
I'd rather the second deuce be a 3, 4, 5 or 6, but that's about it...
(A3)(A4) and (A3)A2 are better, but not by much.
as you know, nut low+nut flush possibilities are really powerful.

i'll never forget the hand I lost the most bets in: It was 4/8 at the Orleans, I had (A4)(A4) and after the hand I was 11 big bets short (capped pre, capped post, capped turn and luckily only one river bet).

Edit: here it is!

http://forumserver.twoplustwo.com/44...ans-lv-839115/