Sklansky hand groups v.s. half million party poker hands

[theRealMacoy]

Hello one and all,

I was just checking out the list of hands from highest to lowest by EV at Party Poker:

EV list here

It seems that the list actually dovetails fairly nicely for the first couple of Sklansky's groups. As one continues down the list the order on the EV list starts to diverge from Sklansky's hand groups. JTs for example appears much lower on the EV heirarchy list.

I realize that these stats essentially show an average party poker player (i.e., all players hands are combined); however they are generated from actual live games across many limits (thus playing them differently than "average" would affect their EV). Data of this sort was not previously available, even in the fairly recent past. In particular this strikes me as notable, due to the live-game source and reasonably large sample size.

I am certainly not saying this is the be all or end all of hand strength lists; however it makes me sit up and want to do some thinking.

Discussion anyone?

The Real Macoy

ps. I appologize if this has been previously debated.

[MaxPower]

These are from PokerRoom not Party Poker and they have been discussed many times on this forum.

[Aceshigh7]

I like these +ev hand charts and I prefer this to the Sklansky grouping method.

[shummie]

I'm not sure these are the PokerRoom hands that I've seen before.

I didn't properly research the site, but I had an idea anyway. If these stats were recorded based on known hands in a DB like Poker Tracker (or some other game observing software), they may be skewed because the hands known in Poker Tracker are just the hands that make it to show down (besides your own hands). So crappier hands will look better because they don't make the list all those times they miss the flop and get folded.

If this is a list of hands from some other source without this restraint, then nevermind.

- Jason

[Jim T]

[ QUOTE ]
I'm not sure these are the PokerRoom hands that I've seen before.

I didn't properly research the site, but I had an idea anyway. If these stats were recorded based on known hands in a DB like Poker Tracker (or some other game observing software), they may be skewed because the hands known in Poker Tracker are just the hands that make it to show down (besides your own hands). So crappier hands will look better because they don't make the list all those times they miss the flop and get folded.

If this is a list of hands from some other source without this restraint, then nevermind.

- Jason

[/ QUOTE ]

No, when you have roughly the same number of each pocket pair, then you are clearly not working with just hands that went to showdown.

As for the value of the list in comparison to others out there, it is obviously heavily skewed by poor play. When 72o doesn't even make the 20 worst hands, being behind 72s, A2, etc., what does that mean for other hands?

I'd suspect that poor play can have a devistating effect on the overall profitability of speculative hands (like the previously mentioned JTs for example).

[bilbo-san]

Another consideration is position and number of players. JTs is *much* more profitable on the button vs. 4 limpers than UTG heads-up against the button after he raised and drove out the blinds.

I suspect even if you ignored number of players and just did a breakdown like this for every position, the information is a lot more useful. KJo, for example is +.08, but I bet if we only looked at KJo's EV when it's played from the button, we'd get a much different number.

These kinds of heuristics are pretty important -- it's similar to what allows world-class chess players to outhink computers (at least the non-mainframe computers [img]/images/graemlins/smile.gif[/img]). Thay can dismiss so many moves out of hand, whereas the computer has to consider every move.

Most of the players on these forums dismiss KJ UTG out of hand -- most players on Party do not. That, as much as post-flop play, is what makes our lists of top EV hands so different than this list.

[theRealMacoy]

[ QUOTE ]
Another consideration is position and number of players. JTs is *much* more profitable on the button vs. 4 limpers than UTG heads-up against the button after he raised and drove out the blinds.

I suspect even if you ignored number of players and just did a breakdown like this for every position, the information is a lot more useful. KJo, for example is +.08, but I bet if we only looked at KJo's EV when it's played from the button, we'd get a much different number.

These kinds of heuristics are pretty important -- it's similar to what allows world-class chess players to outhink computers (at least the non-mainframe computers [img]/images/graemlins/smile.gif[/img]). Thay can dismiss so many moves out of hand, whereas the computer has to consider every move.

Most of the players on these forums dismiss KJ UTG out of hand -- most players on Party do not. That, as much as post-flop play, is what makes our lists of top EV hands so different than this list.

[/ QUOTE ]


I agree with your analysis regarding position.

Check out the EV breakdown at Poker Room (thank you) based on position.

EV list based on position

There are also breakdowns based on the number of players at the table, which makes for interesting comparisons. You can also run analyses by hand and/or limit and finally by the number of players at the table.

The main EV page - analysis by hand/number of players/limits

Just as in any theoretical field at the outset theories are based on little evidence. As the field progresses more and more evidence becomes available and the theories are refined. Finally, particularly when many advances are made through experimentation or simulation, theories are a reflection of the data available. This would be generally true for my own field of Psychology. Original theories were completely devoid of empirical data, whereas modern ones are rooted in it. Will there be some refinement to the Sklansky hand groups as the amount of empirical evidence increases? Whether now or in the future as people's personal hand histories of live play surpass millions of hands, giving us data on good players?

Additionally, I have often read how B&M card rooms are so much softer than playing online. If this is true then the quality of play represented by these stats is at least a notch above B&M.


Cheers,
The Real Macoy

[theRealMacoy]

bump

anyone give me the links to previous discussions?
searches thus far haven't given me much relevant info.

thanks,
the Real Macoy

ps. I am not trying to diss the Sklansky hands with this discussion; simply looking for some debate regarding their utility in light of various new data. also, much of the data actually supports the groupings (particularly the initial groups). further, looking at the data based on the number of players also supports such ideas as not playing low pocket-pairs based on number of players.

cheers,
the real macoy

[pzhon]

PokerRoom statistics are averaged over all of the times people get the hands. You may see XYs do worse than XYo. Does that mean XYo is stronger? No, but some fish play any two suited, a mistake, and they drag the average for XYs down. People know 72o is garbage so the vast majority of players muck it UTG. However, many players don't know to muck JTo UTG, so you see JTo losing more than 72o UTG, 0.14 BB/100 versus 0.09 BB/100 at $1-$2. Don't base your hand rankings on that.