Quiz Question: What would you do?

Here’s another one that I found interesting. What would you do?

You are playing $2.5/$5 cash game with 4 other players. You have ~$600 and you are 2nd to act with 86 of clubs after the under the gun player folds. You open limp for $5 (not a good play), and immediately get raised to $20 by the button, who has $1200 behind. Everyone else folds, and you call.

The flop comes 2 of spades, 7 of clubs, ten of clubs, (2s7cTc) giving you a flush draw and a gutshot straight draw. The pot is now $47.50, and you lead out for $35.50. The button raises you to $115.75, and you call the extra $80. The turn comes the ace of spades (2s7cTcAs). You check and the button bets $184.25. What do you do? Your choices are fold, call, or raise all-in ($475 total, ~300 more to your opponent, about the size of the pot). I said I would fold, and Snowie said it would go all-in as a bluff. I will give my reasoning below.

I said fold because you have no showdown value and I don’t see you ever getting enough folds. The site says that your opponent has a very strong hand or a bluff because they would not bet this way with hands like JJ-KK or KT. But I just don’t see how your opponent would continue bluffing here. They can represent the ace effectively because it seems like they can have so many aces in their range: they could have been bluffing with any Ax of clubs on the flop and then improved on the turn. They could also have 22, 77, TT, or AA, which are the hands they are representing, while hero doesn’t have TT or AA in their range because of the open limp preflop. The opponent’s most probable bluff hands are clubs, which you block, or possibly 89, so it just doesn’t seem like you will be able to get enough folds for this high variance play. Of course, you have some equity against your opponent’s value range (unless of course they hold AK, AQ, or AJ, etc. of clubs), but if they do have a set, at least 2 of your flush outs are no good (2 and A of clubs). I guess the shove can get folds from bigger flush draws, but is that really the goal, since your opponent never has a medium strength hand?

As unranked always says, we should be considering our hand within our entire range, and I suppose 86 of clubs is a good hand to bluff with because it has some equity without any showdown value, but I think we would just have too many bluffs in our range if we did this with all the suited club KJ, J9, Q9, 56, etc that we also have in our range (although I have no idea what range we open limp). Would we be folding Q9 of clubs here? How about 45 of clubs? We would want to bet for value with our sets of 2s and 7s, but there just isn’t a lot of value in our range on this board because we probably have a lot of suited connectors that need to improve to straights and flushes to have value, and there are none on this board. So, I don’t really see the need to turn this into a bluff.

Just an interesting one to think about.

1 Like


Uhhh no, giving you a 10-high flush/straight flush draw, and a 10 high straight draw.

Remember Joe, they didn’t limp, you did… he raised you to $20…

So its… 8c 6c vs Xx Xx …with … 2s 7c 10c As…

First of all you’ve been raised twice for whatever reason, and now … not only did a A pop, but theres another flush out there possible and you got no help for the 115 you just spent. It matters to me how the past few hands played out before I can say any other play is worthwhile… In a vaccum, I would’ve already play’d this differently and gotten a better price to see the turn hopefully or I would already be gone… sure the alure of a str8flush is there, and its unbeatable if it hits, … but you have to assume you are behind at this point, all you really have is A high… yes you will be 1st to act after you see the river, but all he needs is 1 pr higher than a 8 to beat you @ showdown if you don’t hit your flush. He has you 2:1 in stacks so odds are you will get called by a raise/shove…

I could say alot more Joe, but I think I’ll leave it there… its simple, either gamble on the draw all the way ( almost shove but assume you’ll call any raise all in ) or fold.

I am folding for a very simple reason: if I cant come up with a line that makes sense to me here, I have no reason to think a good player will make a good one up for me and fold.

I open limped in a 5 person game and didn’t re-raise. In my head, I have zero A’s or pairs in my range so no one would give me credit for hitting my A or having a set by the turn as a result. So, what on earth could I have here other than a drawing hand? In my mind, I am maxed out at representing an OESF draw and even that is fishy because I’m raising 8/9s not limping it. Only other possible hand is 2 pair 10/7. Guess I could limp 10/7s? Am I ever leading out with a meh-10 on the flop on that board? Probably not so the 1 pair thing isn’t on the table really.

I don’t think villain has the nuts or anything near it or even a made hand yet. In fact, I’d say a good portion of the time he has whiffed everything here by a good stretch. So, its not that I think he likely has a hand he can’t fold, its that I’m not sure what I’m selling is believable. What percentage of his range has whiffed and can only improve to 1-pair on the river? That’s all I’m folding out with the shove I think.

I could be 100% wrong on this one but the 0EV play is the fold and I’m going there without a better idea why I should be doing something else. I can make a case for villain being on air because of the betting pattern and I think he bluffs the A turn always because hero cant have one. Nope, still cant find a shove here no matter how hard I try.

1 Like

can’t add much more reasons then already is mentioned, but i agree on the fold.

the only thing i can add is to fold pre, not only with the limp thing but also after the 4BB raise.
on the flop, i agree with the semi bluff since it’s a flop that don’t hit very well, but instead of 75% i would bet about 60-65% of the pot, to prevent blowing it up like this with close to the same result. as played i agree on the call because of the odds and implied odds you get.
on the turn i do agree with the check because you can’t represent much except for the draw you already have.

Plain and simple, FOLD .

Yeah, I was surpised that snowie said shove, but I wonder if it has to do with the machine learning being based on previous situations in which players have open limped. You would think, depending on the data source, that analyzing the results of players who have an open limp range here would not be useful because they would be bad players (i.e., if you put trash into your model, you will get trash out). But maybe I just haven’t read enough about how snowie works, and it could be doing range calculations based on the pure math rather than results data. But then how is it coming up with these scenarios that involve weird min raises and open limping. Either way, this seems like an unnecessary bluff. But I could be wrong…


I could be totally off base here but if the goal is to develop an unexploitable strategy against all other strategies, doesn’t that mean you have to include sub-optimal technical plays into the data? Now, I do see a good amount of limped pots live - not a huge percentage of them but not an inconsequential amount either. It happens and I’d think you’d need to incorporate this into an overall strategy.

Anyway, while I am still comfortable with the decision to fold, I did enjoy trying to think about whether it was possible to steal this pot or not. Thinking about ways to get the money in the middle is a great exercise for me. I do not believe I am close to balanced on my bluffing ranges yet and maybe thought experiments like this one will help me get there?

Added: Did some more reading on Snowie and it appears that all scenarios are derived from playing random hands. That includes all kinds of strange unconventional situations. So, open limps, min raises and the like have all been run through, apparently some crazy number of times per day to come up with these results. The main weakness that the developers acknowledge is the error rate against wildly non-GTO play, mostly in bet sizing. Supposedly a fix for this is on the works but it makes total sense from where I sit since there were similar problems with options pricing models beyond certain input thresholds (volatility mostly). Had to switch models to get correct results and then try to work out what to do in the grey areas. I suppose that is what Snowie is going to try and do with the next release.

Just cool to think about a program running trillions of hands a day and coming up with scenarios to practice against. Sure, it may make some “mistakes” from the purely optimal theory but its so far ahead of where I am that I am getting a ton of utility from playing against it.


Haven’t gone there, hahaha Snowie… you see this is pure genius…
Just think a distributable, non centralized poker AI, capable of winning 10% per 4 hours.
Now, Watson from IBM should be able to learn poker and learn every hand from every hand play’d online and WSOP from every player. You and I cannot just remember every hand someone has played and be able to extrapolate from that how they are currently playing or may play.

In 1 year aprox, I think I’ve play’d what 55-60k hands here… just think if you could instant recall all of them, then sort like a database, ect ect ect ect … on your opponent in realtime or to aid yourself… Think of the progressions …

Seriously, whats gonna pay more… people paying to play simulations against Snowie… or … Snowie being rolled out from 100 IP addresses, and playing for smallish stakes for real cash ??? Could you spot a AI, or would you just think it was a decent player taking your cash ??? All these ppl playing aganst Snowie, is giving it 1 seriously huge database of possible ways to play hands.

All very cool and interesting ideas. I’m not sure this program is of the level you’d need though. Its a training site. From what I understand, there are far more powerful platforms already out there that the top HU players work on. The very top of the AI food-chain still isn’t capable of beating pros on a 6-max table as far as I know.

Since I don’t know much about these things, I have no idea what a break-even rate would be even if you could run powerful bots like this across multiple sites and tables. I assume the costs of running these things are not insignificant so if you factor in the costs to operate and the rake, what would the break-even point be? Any idea? I’m honestly interested.

ok, I have no clue what IBM charges for time with Watson, but when it played Jeapordy years ago ( its alot better now ) it wasn’t factoring in the opponents. Maybe the top 5 players should pony up a mil each lets say, and challange Watson 6-max… but you see Watson is so beyond mearly playing games, when they’re curing cancer or being “farm’d” out to the PGA… Time with/against Watson is pretty expensive…

As far as cost to operate, how much does it cost to run your computer 24-7 ? see what I mean ( winks )… but Snowie “could be” learning all while it is being used as a computer opponent for practice… plus why put up senarios then ask what/why would you do… thats more into the psychological aspect of playing, thats being collected… not just hand histories.

I’m not saying they are, just things that make ya go hmmm.

I like the free quizzes on the Snowie site; the two hands I posted here were a couple of the more “controversial” questions that were interesting to think about. It makes sense (as unranked said earlier) to be prepared for all kinds of unconventional situations.

One thing I wonder about Snowie and GTO in general (which may be what you were just saying), is that it can make sense to make sub-optimal or even downright weird or incorrect plays in some situations to provide range balancing, etc. And to extend upon that, in real-world situations it can make sense to play in ways that are unconventional or even incorrect because it can pay off in future situations in terms of table image, perceived ranges, perceived competence, etc. Playing the pure math would surely pay off against most players, but at higher levels I think it can make sense to occasionally make a high variance play like the one in the question just so that your opponents know you are capable of doing it. If your shoving range is 2/3 or 3/4 value hands, then you will get paid off in the long run. It seems like Snowie is able to take that range balancing into account, but I wonder if it considers the effects of a particular play on table image/future action.

1 Like

I’m actually laughing out loud because I had a similar thought last night. I wondered if there was such a thing as incorporating totally nonsensical plays into your ranges just so your opponents can’t be sure if you’re competent or not and be just a little afraid of you :slight_smile:

I don’t know if it has a concept of future actions or meta-game. My gut tells me it shouldn’t be able to but again, most of this is way above me. If it has learned things like “donk betting” and range-merging … it has a pretty cool arsenal at its disposal whether it can look forwards or not. It seems it actually has a higher incidence of bluffing than most human players as well so not sure if this hand is part of that behavior.

I wish we could get some more detailed analysis of the hands and why the recommended plays are “correct”. On the whole though I think I am learning more from using the program than I had from almost any other source. Since it costs less than I’d spend on 1 hour with a pretty cheap coach, I think its pretty useful, even with the questions it leaves me wondering about.

1 Like

ummmm, Yes… Yes… and ohh hell yeah …


I wondered if there was such a thing as incorporating totally nonsensical plays into your ranges just so your opponents can’t be sure if you’re competent or not and be just a little afraid of you :slight_smile:

…or at least that’s what I say I’m doing with those totally nonsensical plays…yeah, that’s the ticket. I’m just throwin’ off my opponents, so I can swoop in soon for the kill.


I will be watching out for the swoop lol