I have already designed a runner-runner chart specifically for the 3-flush, HU. At some point I will add players and create tables for every situation

http://www.openfaceodds.com/charts_3_flush.html

+1 on this

That chart seems to be correct (I didn't check every figure, but enough to be convinced).

It's virtually impossible to publish all relevant charts. I have an excel with all the data that can be calculated with the formulas I have researched, and it contains tens of thousands of cells of probabilities.

Should be JJ2 bottom with 10-6 middle

Not sure if this is posted in the right forum mods please move if not. Just got into an argument and a bet has come about. Scenario: I have 8824 on the bottom QQA in the middle Q9 up top. I pull an A and place it in the middle with 2 8's left to make trips on the bottom. We're playing criss cross so there's 4 hands being played. With an equal amount of cards left do I have a true 25% chance of hitting an 8. He's saying with a compounding variable I have less of a chance which I can't help but call him slow

I'm not sure what you mean by "With an equal amount of cards left".

Anyhow, with Q9/AAQQ/8842 there are 12-15 cards in the deck, with three more to come. The chances for hitting at least one of two outs are 37.1% / 39.6% / 42.3% or 45.5% depending on your position. Where did you get the 25% from?

I somehow managed to use 6 points for flush (instead of 4) in the above calculation. The correct figures are:
- all clubs alive: .316 * 26 + .494 * 5 - .19 * 6 = 9.862 points
- 4 clubs alive: .316 * 26 + .226 * 5 - .458 * 6 = 6.914
(with some rounding errors because the probabilities are not exact but rounded before calculation).

The conclusion is still the same

^ one more addition: I used the royalties that "normal" in Finland, which is 25 for straight flush, if it's less the result changes still.

This seems like a mistake and it's because we lower the # of games per simulation for quicker results. We'll increase the # of games per simulation on our next update.

Also, Warren isn't perfect at the moment. He has simulations for most spots but because there are so many combinations for starting hands, if he hasn't see a situation before, he'll make an estimate. If the estimate seems off, our customer can run the simulation multiple times to verify the results.

We're also training Warren at the moment to recognize these spots better. In fact, according to our tests, the most recent Warren beat the previous version by a significant statistical amount.

It may be just me, but to me Warren seems a bit expensive for an application/service with that kind of issues. Especially with a subscription model - if you buy an application and it doesn't work properly, you can get a fixed version for free. If you subscribe to a service and it doesn't work properly, you pay the normal fee even for the period it isn't working properly, or?

1 specific hand of his 2 hands getting 1 of 2 cards left in the deck in a 4 hand game. His odds are 25%.

We've been very up front with our customers from day 1. Warren isn't perfect yet, and he might never be. He's going to make mistakes but he will continue to get better. In fact, the most recent version of Warren is crushing small- and some mid-stakes players.

We will release Warren in Supercomputer Mode this week that enables Warren to use supercomputer servers to compute simulations to make decisions instead of using estimates to make decisions. I doubt many players can beat him consistently then, especially in Pineapple.

Fortunately for us, our customers understand that Warren isn't perfect. But as long as our customers are crushing their opponents, they can probably live with that. We will continue to work hard to improve Warren and if there's a day when we don't provide value to our customers, they'll leave and we'll close shop. I love our chances though. Deep Blue wasn't perfect when it played against Kasparov either.

Care to explain that again?

No, there's clearly a fundamental flaw. No number of iterations should lead to this result, as long it's not 1.

I don't think you understand the uncertainty associated with simulation. As slow habit alluded to, a lot of these simulations are based on very few runs, since each run takes a long time. When this is the case, you get suboptimal results pretty easily. I can attest to this bc I wrote a ofc simulator myself.

first to act: A A 3 Q Q

are you putting the 2 Queens up top and going for the gold? if so do the aces go in the back or middle?

How many players? One thing people might not consider is that all things being equal, taking a shot at FL is a better gamble with more players, for a couple of reasons:

1. You'll see more cards, giving you more information throughout hand, increasing your chance of unfouling.
2. When you do make FL heads up, your single opponent knows he's usually getting scooped, so he can more or less freeroll and gamble for a big hand without giving up much. But if there are 3 other players they have to keep each other honest so no one can freeroll you.

the last one was HU. also, i just played a hand, started with trip 8s and J 10 both hearts.

i put the trips in back and the J10 in the middle. strange thing was, i made a flush in the middle and fouled with only trips in the back. was this just unlucky or is there an argument to be made for putting the trips in the middle and 2 suited, connected cards in back?

nope.

All right, I took a look over this. I think it's either a matter of "streets" interpretation, or I ****ed it up, or both. After the deal, the UTG player is 17% to hit a one-outer (5 known cards, 47 unknown, 8 cards to come). Before 6th street, however, 15 more cards have been dealt, leaving 32 unknown cards. After setting the 6th st card, UTG has 7 more cards to come, 21 known, and 31 unknown. In this case the equation here should be:

1-C(30,7)/C(31,7) = 22.58% = 23%

What the chart is supposed to do, however, is give the odds before the upcoming street card is seen, so the UTG player 'on' or 'approaching' 6th st. will have 32 unknown cards and 20 known cards, with 8 cards to come (but after the other players have been dealt 15 more cards, which accounts for the large jump in odds even though UTG hasn't gotten any more cards yet), in which case the equation should be:

1-C(31,8)/C(32,8) = 25% -- or exactly the same odds the dealer has right after the first 20 cards are dealt.

Unfortunately the equation I actually used was:

1-C(31,7)/C(32,7) = 21.88% = 22%

So, I need to go back and adjust those. Fortunately the numbers are only off by miniscule percentages so I don't think any1 using it would have been massively miscalculating their odds

Anyways, thanks for the catch. Exactly why I posted that question -- I had a weird feeling about it.

I am most likely doing aa middle qq3 bottom

Fwiw it's a joke people are paying for warren and that they're even charging.

The purpose of AA in the middle, first and foremost, to allow yourself to go to Fantasyland if you pick up QQ or KK. You're taking the risk of fouling in exchange for the pretty damn live shot at FL.

QQ is EXACTLY what you want on top under such scenario. If you set up x/AA/QQ3, you've cut your FL chances in half, any Queens are going in back, so you're only looking at Kings. You now need 2 out of the 4 kings, as opposed to 2 out of either the 4 kings OR 4 queens. And you still incur the same level of risk of fouling, you need a Queen or a 3 to cover, and you still get nailed if another Ace comes.

Meanwhile, QQ/AA/3, while risky, gets you to Fantasyland if you cover. It's the whole point of the Aces in the middle set-up, no?

I'm not saying play it that way; I'm sure the safer play of x/QQ/AA3 is perfectly fine; it gives you a very nice shot at 9s, Ts, or Jacks up top for a nice royalty, in addition to the boat draw in the back.

Maybe I'm missing something, but to me, the two plays are either QQ/AA/3 or x/QQ/AA3. I don't see how x/AA/QQ3 makes sense as it seems to be taking a legitimate chunk of risk without the proper level of reward (since you can only go to FL with two Kings). I do agree x/AA/JJ3, or any other pair, but not Queens, since the whole point of Aces in the middle is to shoot for FL.

DrawNone: QQ/x/AA3 seems silly to me; you sandwich yourself into needing either KK and KK only, or two pair if you can improve the back. I just argued that the odds of picking up KK specifically are pretty low right before this comment, so I can't really turn around and stand by a set-up that puts itself at a decent risk of fouling unless KK comes.

Yes, I know, there are weird runouts where you pick up the last two Aces or Queens to not foul, it's a relatively small amount of the time though.

To me, the FL set-up is QQ/AA/3. First 3 live cards go in the back. You are most likely looking at having a spot like QQx/AAxxx/pair+2 live cards, with something like 6-8 outs to the land. And if that's too much risk, then I go x/QQ/AA3; you can still end up going KK/Queens Up / Aces up to go to FL anyway.

I dunno; maybe I'm just running good with my Aces in the middle sets. I do seem to nail them fairly frequently, but then again I seem to brick the vast majority of other types of draws, but doesn't everyone claim to run horrible at OFC?

The relevance of 13th street odds is about the decision being made on 12th street. Obviously there is only one place to put your card on 13th street, but the odds of hitting any needed outs on 13th st will influence how you set your 12th street card.

I will be re-engineering the chart soon.

I was looking at it to simplistically. In our argument we both agreed that each point was valid but I was way off because it changes depending where you are drawing from. I was just looking at it like there are 2 8's and 12 cards and 4 players the 8 has to go to one of the four players.