Pokerroom.com starting hand EV stats

[StoneAge]

www.pokerroom.com (http://www.pokerroom.com) has an interesting page you might want to check out. I tried puting a link to the page in but it didn't work, so here is how to get there. Go to pokerroom.com, hit the 'statistics' link, then hit the 'total stats by EV' link in the middle of the page.

I would be interested in your interpretations of these stats. According to the page the last time I was there these stats compiled using 46,831,525 pairs of pocket cards at real money games. About half of the action at this site is 35%-60% btf 1-2 games. The EV figures given are average over all games 1-2 to 25-50, 2 to 10 handed, regardless of postition. You can get more specific info, such as what is the EV of KTs on the button in a 5 handed game, if you click on the statistics link.

I have been thinking about a few things since first checking this out:

Is a list in order of highest EV to lowest a better hand ranking system to make adjustments from other hand ranking system (Slansky for example)?

According to S&M, in loose games high unsuited cards go down in value. The majority of the data has been taken from loose 1-2 and 2-4 games. AK is 8th in EV, 10th hand in Slansky hand groups(SHG); AQ is 14th in EV, 16th in SHG; AJ is 19th in EV, 23rd in SHG; KJ is 27th in EV, 29th in SHG. KQ however is 21st in EV and 18th in SHG. The EV stats also seem to indicate that JTs, T9s, and 98s are very overrated in the SHG, even though the data was taken mostly in loose games where these hands should gain value. Axs does show higher EV than SHG, which would be consistent with S&M advice

This EV data represents the average EV of the average player on pokerroom.com. How much -EV rating could a good or expert player be able overcome by playing better than average? In other words using position, maniacs, isolation, number of limpers, post flop play etc. better than the average player. For example QT and JT both have an EV of -0.03BB. How about 56s (-0.07BB). As a comparison 23 is -0.14 and 72s is -0.15. I realise that the answer is the better you are the more you can overcome, but what is realistic?

Or do these stats have almost no meaning? There is some wierdness to these stats when you start looking at specific hands. A8s has an EV of .61BB utg and -.43BB on the button in a ten handed 10-20 game.

[tewall]

I didn't see any statistics link at the page so I couldn't see the statistics. I'm guessing the statistics don't show the results by position, so that would limit their usefullness by quite a bit. These statistics are showing how hands do in loose games when played by poor players, so the results are interesting by of limited usefulness.

For example, regarding medium suited connectors. To make a profit with these one needs to play when many people are in, it doesn't cost too much to get to the flop, and one needs to get away from a poor flop when the odds don't warrant chasing. It may be that the EV is low on these hands not because they are overvalued by S&M but because they are not well played.

[eMarkM]

FWIW, here's a direct link to EV for 10 players (https://www.pokerroom.com/evstats/totalStatsPositions.php?players=10)

If find these stats pretty interesting. I wouldn't say they should replace the hand rankings, but they do closely approximate them. I wouldn't expect live data to fit perfectly into the SHG model. There's plenty of anomolies in the data (e.g., why does 64s have a better EV UTG than AQo) to make you realize that you have to take some of it with a grain of salt. But the averages of all positions seems pretty close to the SHG (I don't have HPFAP in front of me to compare).

What stood out for me is the great negative EV with Axo hands. Six of the worst 10 EV are A2o-A8o. All those "any A" players are getting dominated from every position. In fact any Axo lower than AT is negative EV in all positions. Hence you don't see those in SHG at all.

It shows winning just the blinds with AA/KK/QQ/JJ/AKs sucks as these are worth more than the blinds. Everything else, you win the blinds, you're ahead. Again, nothing earth-shattering, but it's nice to see this data from real game action and not just sims.

Obviously, this doesn't show how many people were in when the player limped with JTs, etc, but overall I find this data interesting. It be nice if they added a feature that also included how many players were in the pot when coming in at a given position.

[tewall]

I was able to see the statistics link from this address:
https://www.pokerroom.com/info-handhistory/
I was wrong guessing the statistics didn't show position. They do. 98 suited and T9 suited did poorer than I expected even in best position.

I think the statistics can be quite useful to a beginning play in the following way. Look at the hands by position and select the hands that are clear winners and only play those to start off. Then as skill/confidence increases start adding more hands.

Very interesting link. Thanks for posting it.

[Bozeman]

When we discussed these #'s a while ago on this board (there were only 10 milion hands then) I pointed out that the greatest value of these #'s is that they shown where your average opponents are making their biggest mistakes. No expert would claim A2 is worse than 72, but the average player is losing more with it. However, this is also the reason that the comparison to S&M rankings is meaningless.

Craig

[AmericanAirlines]

What do the numbers represent? EV in SB's BB's something else?

Sincerely,
AA

[Clarkmeister]

I happen to think the numbers on this site are significant as long as you know what you are looking at, and what you can and cannot learn from them. As long as you are aware of that great big "caveat" then these stats are useful. I only wish Paradise published the same info.

For example, the idea that AJo and KQo are marginal, at best, when UTG is pretty resoundingly refuted here.

[AmericanAirlines]

I agree,
Comparing to 2+2 rankings may not tell us a lot. But could we infer that every hand that has an EV less than 72o is one to examine and be careful how you play it, if at all.

The idea being that if people are losing more with those hands than they should, there's something not understood about how to play them.

Granted, it could be that the sample represents that people just like to gamble with those hands for irrational reasons.

These rankings also might tell me what hands I should consider playing if I don't think I'm an expert, perhaps?

I.E. Play the ones that "everyone" has a +EV with? Then maybe focus on how to play them optimally? Also, if you read a "fish" with one of the average +EV hands... consider that they may be harder to beat when they hold it... as in maybe those are hard hands to get someone to miss-play?

Thoughts?

Sincerely,
AA

[KSU78]

I guess that I am not surprised by the performance of AXos. At the lower limits, many players do not know how to play these hands correctly. They fail to realize that they're looking for trips or two pair to make a viable hand, the nut flush draw or a open-ended straight draw.

[Glenn]

Hi tewall-

You wrote:

"98 suited and T9 suited did poorer than I expected even in best position."

I noticed this too, but I believe this is becuase people cold call raises with these hands way too much. If you looked at only hands where they paid one bet to see the flop, I think T9s and 98s would perform much better.

[StoneAge]

I thought this might have been discussed before but I missed it. Thanks for posting again Bozeman.

[StoneAge]

[StoneAge]

These stats do seem to show that the value of these unsiuited high cards is higher that I thought from my interpretations of S&M and that they do hold their own in loose games- AJ seems just as profitable in a 1-2 10 person table as a 10-20 10 person table. I do need to be a little cautious with this though because I 'want' this to be true. But then it seems that if an 'average' player can play AJ profitably utg 10 handed that an 'above average' player should be able to.

[Clarkmeister]

Glenn,

While your point certainly has merit, I think the important thing to note about the medium suited connectors is just how position sensitive they are. Holdem is still a big card game and conditions need to be ideal to play non-pair little cards for profit. Clearly, as you stated, you don't want to be paying a premium to see the flop with these hands, nor can you stand to be more than 2 off the button.

OTOH, note how even marginal hands like K9s pretty much make money anywhere.

Holdem is a big card game.

[rigoletto]

Does anybody know if rake is taken into consideration in these tables? If not a lot of these marginal winners will turn into loosers!

[Clarkmeister]

It is my understanding that rake is taken into consideration. I did some analysis of these numbers about 8 months ago and I seem to remember that rake was clearly taken out.

[StoneAge]

First off thanks for participating in this thread as your input has been great.

A couple questions for you:

Did you give enough validity to this data to let it change the mix of hands you played and the positions you played them in?

Can a player making 1.5-2.25BB/hr (50 hands/hr)to turn a hand an average player has an EV of -0.03BB btf into a pofitable hand? -0.08? Or is this asking the wrong question?

[Clarkmeister]

I think its asking the wrong question. I think the key is understanding how these hands play, and where they make their money. Also understand that you need to be in a very good game. As a very basic example, A4s makes money in every position, but that doesn't mean you can cold call an UTG raise with it. A4s makes money in these looser online games UTG, but that doesn't mean it will make money in a reasonalby tight, tough middle limit table. If you want an estimate, I would suspect that a player as you describe would be able to add between .03-.05 per hand.

Just be careful what conclusions you draw.

As for me, the data more or less confirmed things that I had already modified in my game vs. the standard Hand Rankings. I already had played any suited ace and any pair in any position for a single bet. Did I add a few hands? Yes, I added K9s, KTs into my mix of hands UTG when in an appropriately loose game. I also started being much more picky about where I would pay hands like 78s.

[AmericanAirlines]

Hi Everyone,
I was looking over the cumulative EV's on the pokerroom.com site at this link.

https://www.pokerroom.com/evstats/totalStats.php?order=value

When I noticed that the count for AA was 212,549... and the count for 72 was 424,833...

What's wrong with this picture... Are people really playing 72 twice as often as AA?

Sincerely,
AA

[Clarkmeister]

There are 6 ways you can get dealt AA:

AcAs, AcAh, AcAd, AsAh, AsAd, AhAd

But 12 ways you can get dealt 72o

7c2s, 7c2h, 7c2d, 7s2c, 7s2h, 7s2d, 7h2c, 7h2d, 7hds, 7d2c, 7d2s, 7d2h.

[Homer]

Those numbers are how many times someone has been dealt that hand. Since there are twice as many ways to get 72 as there are to get AA, the 72 hand count of 424,833 is almost exactly double the hand count of 212,549 for AA.

[tewall]

I think the last part of the post would be good advice for people starting out. That is, concentrate on the hands which are positive E.V. for everybody. The other hands even with proper play will not cost much in opportunity loss even to a good player and could easily cost an inexperienced player money.

[AmericanAirlines]

Hi Homer,
Ok, guess the assumption that these were hands that were at least limped in on was false... Hands dealt makes more sense...

Doh!

But is that the correct way, then to calculate the EV? Hands dealt vrs. Hands actually played, so to speak? Perhaps so, since effectively, a percentatge of your blinds is your "ante" for each hand in an orbit?

Sincerely,
AA