Has anyone done much thinking about starting-5 hand range classifications similar to the 169-starting hand charts from NLHE?

(full disclosure) I am wanting to eventually compile a chart/list on http://www.openfaceodds.com/index.html , and I'm also interested in discussing and writing about strategy based on strength of hand (first 5).

Obviously there's going to be literally thousands of combinations of first five, but I'm looking to reduce and generalize into some sort of manageable continuum of strength.

At first I thought this question would help define starting hand ranges but I have this feeling it's more of an exponential can of worms, e.g.:

Is there a consensus as to what a strong or weak starting hand is?

Can we even prove that one hand is superior to another?

Is AAAAK definitely better than 109234 rainbow?

Is it just impossible to make conjectures/proofs about these types of things without a computer running thousands of scenarios?


From infinite space..............

Yeah but the question clearly didn't mention pineapple

OFC_OMG,

I love what you're doing with openfaceodds. However, I don't think a starting-5 hand range classifications is helpful simply because you can't fold and your betting unit is fixed.

I guess it's helpful in OFC variations where you're dealt 7 cards and can discard 2 or you're dealt 5 cards and can raise the max.

What variation has raising? Is anyone playing a limit-style betting system throughout the hand? Also, is this 7-card deal a newer variation, or just one of the ways people are dealing FL in Pineapple?

I see what you're saying about the first five in standard OFC - in Holdem you can rank the hands in a chart and fold what's out of your range. And, of course, there is no folding in OFC.

Maybe it's a question of EV ranges for xxxxx depending on the contour of the hand, in which case I'm trying to define 'contour ranges', and tactics based around maxing EV (as well as having a realistic expectation for what kind of masterpiece you can paint with the particular 5 cards in front of you).

Off the top of my head: three 'contour ranges' where you could have higher or lower EV:

a. 78910ccccAd: Fantasyland-bound plus likely scoop and royalties

b. 9976Q: Possibly Fantasyland-bound but likely not scooping and fewer royalties

c. J9224 rainbow: Forget it buddy/defensive play only/get lucky


Also 'contour range' possibly including known cards if not first to act

I was using hypotheticals.

You're making it too complicated. OFC is like a more complex version of Blackjack. Your play depends on what your opponent has and there's a mathematical estimate for that situation.

If you still want to pursue the idea of "contour ranges," I recommend taking up machine learning courses. It's unlikely any equation (or series of equations) will help you figure out the EV for ranges. But if you can do it, more power to you.

Thx for the input! If anyone else has thoughts on this topic I'd love to hear them.

I think focusing on the bolded part would be much more helpful. Obviously Hand A will have a higher EV than Hand C. The thing about hands in Group A is that optimal setting requires little thinking. However, if you asked this forum how to set your Group C hand of J9224, you'll probably get lots of different answers. And the best setting for this hand will depend alot on cards we've seen.

So I think it would be a good idea to have a strategy for setting common groups of hands like your Group C hand and discussing tactics to increase EV.

For example, we could establish a base UTG setting that has the highest EV.

Maybe that's x/94/22J (I'm not sure that this is true, but let's suppose that it is for the example). So we can now discuss what is a realistic expectation for this hand. Perhaps for this hand it's High Card/Pair/Two Pair or Trips. We can also discuss how to make the most of this below average hand. Maybe that's putting a 6 through 8 up top if we get one in one of our first few pulls so we can try to get top row royalties.

Then we can discuss how to change the setting of the hand if we aren't first to act. What if your opponent has a Jack or a pair of Jacks? Then maybe we change the setting to x/J4/229. What if he has one Jack and one 4 are gone? Etc.


I think this is better than knowing whether AA987 or QA998 has more EV and which is the better starting hand when they are obviously both good. Like you said, it's probably opening up a can of worms, but if you want to go there, then I think discussing optimal strategy for groups of hands is a better way to go, starting with common or difficult ones which many of us have problems with.

Yeah, that's pretty much what I had in mind -- groupings of hands with similar 'contours'.....maybe the wording's a little flowery.....

Would we agree that AA987 and QA998 belong in the same category?

For J9224 I personally would go 9/J/224 and then set next live card in back......but I can also see a case for x/224/J9 and even 224/J9/x and I suppose there are other sets with merit. So an EV for each scenario I guess?

Yeah, something like "Strong Fantasyland Potential Starting Hands."

I think I should have used better examples for the point I was trying to prove. Let's try this again...I don't think it's that helpful to know whether AA987 or AJ1072x or QQQJJ has more EV or which is the better starting had. We know how we are setting them. Those are all great starting hands and therefore setting them up is easy.

But I do like your realistic final hands and tactics ideas. Basically, we want a plan for our hand just like we would in NLHE. But obviously, there isn't as much information on OFC yet and so I do think we need some basic strategy plans for various types of starting hands. And I think it would be more useful to start out with hand categories that are tougher to play as that would provide the most help.

Does anyone know how many points of EV Fantasyland adds? Or even an educated guess? I'm writing a OFC simulator program for fun and this would save me a lot of time.

I think shaun deeb said its worth around 12 pts. Have heard people say more. Some say less.

I'm going to work on some starting hand classifications. Off the top of my head I'm thinking, in no particular order of strength:

4-flush
3-flush
Pair
Two-Pair
Trips
3-up-and-down
4-up-and-down
Separated ranges i.e. KQ 637
Hybrids/cannibals i.e. 1098109 (your cards cannibalize each others' draw ranges when set in different rows)

I'm also happy to work on some varying sets/EVs of the same hand if some1 can help lay out the equation or point me to an example.

so, for a medium-low strength hand like J9934, set 4 or 5 ways, what is the EV of each set, maybe what is the FL EV

x
34
J99


34
J
99

99
J
34

J
34
99

J99b-34m

About 9-10 is closer on average.

Yea that sounds more accurate in my limited experience. Any simulators out there where you can set it to play in FL hand after hand? Guess you could always just break out a deck of cards too. Any ideas how many hands you'd need to deal to get a reasonable sample size?

You can actually do this in the OFC app. Just change the number of cards dealt to 13 and remove the fantasyland qualifiers. Basically you are playing standard chinese with OFC scoring.

I've played probably 1500 hands this way against a variety of opponents and the computer. Some hands are monsters. Some are real duds. I don't think this is anywhere close to enough hands to be a valid sample- but that's where the 9-10 pt estimate comes from.

On pocket cowboys app? Are you still not able to play computer on android? If so, I can't figure out how to do it.

Yes but on iPhone/iPad. Can't speak to Android version.

man who do you bitch to about the pocket cowboys app? I downloaded the free one for android, bought fantasyland and 4 color deck, and I don't get fantasyland? what the **** get your **** together

They have a Facebook page. I've sent feedback through messages there and gotten responses.

Towards a theory of first-five opening tactics and EV, I put together this *BETA* first-five hand classification system on openfaceodds.

http://www.openfaceodds.com/charts_starting_hands.html

It DOES NOT give EV for starting hands (yet), nor does it rank hand strengths against each other (yet).

What it DOES is roughly sort classes of hands into the categories below. A 'class of hand' is defined as a 'one-pair', 'two-pair', '3-flush', etc. The category sorter then defines how the first five cards interact with each other, in relation to building a horizontally and vertically complete hand. I have about 70 specific example starting hands for illustration. All examples were dealt by a shuffled 52-card-deck javascript random deal program, the same kind that deals 5-card draw.

Here are the categories I broke them down into, with one example each to show you what I mean:

COOPERATORS: Cards can be grouped in separate draw ranges across two or possibly three rows

example: 456KQ

CANNIBALS: Cards share draw ranges and a split option is presented

example: 456AJ

VERTICALS: Cards can be comfortably split into two separate rows for vertical strength

example: A10K88

DISPARATES: One made poker hand or draw plus (awkwardly) unconnected cards

example: 3779K

HYBRIDS: Grouped cards with multiple simultaneous draws

example: 6789A


The next steps:

1) consider multiple sets of various hands in hopes of
2) defining what makes a strong or weak starting hand, and
3) setting up a full chart/matrix of first-5 ev

-- or something along those lines. Any feedback is appreciated.

Yeah i don't have Facebook. How about putting an app together that does what you pay for? Who charges people for non-existent features?

@Slowhabbit
I thought Neural Networks is about adaptive learning. Meaning the bot would use the expierences or losses to learn.

AI is the future I agree

Anyone looking for a game on pocket cowboys, hit me up. Playing a few other 2p2ers if you need references. Looking to play for around $1 a point.

Playing criss cross with FL QQ+...

Dealt 777KK first to act, would be standard to play X/KK/777 correct?