So it’s only fifty five days until the WSOP and I’m getting excited just thinking about it. Last year I went in quite rusty not having really played a hand of poker in eleven months and was hoping to just play good enough to get lucky and bink something. I made money and it was by all means a successful trip but I was pretty disappointed in myself for going 0/11 in the bracelet events. At the end of the series last year I told myself I would be better prepared this time around and ready to play to the best of my abilities for the full six weeks.

For the past couple of months I’ve been playing bitcoin poker almost every day, railing and analyzing the nosebleed games, and watching strategy vids and discussing hands with friends. I haven’t done a great job of working out, but I’m really going to try to make that a priority over the next two months as it plays a big role in the marathon that is the wsop. I plan on playing 25-30 events hopefully and have rented a house with a ton of poker players this time. There’s a eleven of us in a five bedroom house but not everyone’s there the whole time.

In other news I bought a condo! I don’t close for another twenty five days but I see no reason why the sale shouldn’t go through so I’m pretty excited. My wife and I ended up spending a bit more than we were initially planning, but it’s a nice place and I’m sure we’ll be happy there. It’s a 2 BR/1 BA in belltown in a nice building with all the amenities hot tub, gym, rooftop deck etc. The negotiations went well. The place was listed for $475k, we offered $450k, they countered at $465k. At this point our realtor urged us to take the deal. The place had been on the market less than a week and good places sell very quickly in Seattle. A couple other condos we were contemplating making offers on in the past would be under contract a day later before we made up our minds. I decided to ignore the realtors advice however and gamboool and offer $460k which they ended up accepting. It was risky because they had an extra 24 hours to shop that offer around to other potential buys to try and get a higher offer but it ended up paying off so we’re happy.

But that’s enough boring life updates I’m sure you’re just here for the poker strat! I downloaded Pokersnowie today as I’d heard about and thought I’d check it out(no they didn’t pay me to write a review, although if they read this and would like to retroactively I will happily accept donations :)). I played about 200 hands of headsup NL vs it and apparently it didn’t think I played too badly:

Unfortunately I can’t say the same for it. In reality I’m by no means a world class HU NL player and I’m pretty confident the best ones would be able to beat this bot for a healthy winrate. Having said that it’s pretty fun to play and I think it could improve your game quite a lot – certainly well enough to be able to beat midstakes. Below are two hands it says I played badly and I think it played badly and don’t think it’s even close.

The hand went I minraised button, and the bot check/called every street. Evaluating the bot’s play there’s no way that check/calling on the river is a better play than going all in. If you notice it shows that the EV of calling is 31.93 bets and calling is only 11.93 bets. It’s really hard to make backdoor flushes especially out of position. You might say yeah I mean he has the best hand 90% of the time but when he’s called he’s behind 70% of the time, but vs an actual human player there’s no way this is the case. Most humans are going to have a very hard time folding 2 pair+ to the raise and if they are folding 2pair+ to the raise every time well… it’s going to be pretty easy exploiting them by just check raising them every single time we get to the river with a hand that doesn’t beat AT since the in position guy will have a strong Ace thru top set much more often than he will have a flush.

Evaluating my play it says my bet with the AT is a mistake. I disagree. Most human players are going to be check raising their 2pair+ sometime before the river, so I basically lose to A9,J2s,J3s(24 hands), A2,A4,A5,A6,A7,A8,24,25,26,27,28,29,2T,2Q,2K of clubs though some of these will be 3bet preflop sometimes(15 hands) and the occasional random backdoor flush or slowplay. I beat A4,A5,A6,A7,A8 both offsuit and suited non clubs(75 hands). I chop with AT and most people will 3bet AJ-AK. Most people fold J2o, J3o preflop. That’s 39 hands I’m beat by and 75 hands I beat. Even if I add in all A2 and A3 combos that only adds about 30 hand combos making it still a value bet. The only way this is not a value bet is if the oop guy is never check raising this board on the flop or turn. This may be optimal, but in practice most humans I think will have a flop or turn check raising range.

The second hand I disagree with pokersnowie on is a paired board where again it went I minraise button and the bot check/calls three streets with a very strong hand instead of check raising the river.

This is such a strong hand and very easy river checkraise. I think worse hands that bet/call in my spot are any 7 so Q7,T7,97,87,76,74(96 hands). Better hands that bet/call are all boats so K7,J7,75,KK,JJ,55(66 hands). 55 probably checks the flop or turn and A7 chops obviously. Interestingly, Q7 is actually a losing river check raise. It beats 80 hands vs 98 hands it loses to. Of course this is all based on these bet/call ranges which or may not be correct but for humans I think they’re pretty close. You could add in a 73s,72s,73o for button opening ranges and some people may hero call AK or KJ which would probably push Q7 to a check raise, but that’s very villain and read dependent.

Evaluating my play it thinks the K8 river bet is a mistake. I’m going to be barreling most gutshots AQ,AT,QT,T9,89 and plenty of diamond flush draws on the turn and bluffing many of these on the river. I need to be value betting thinly to balance for this. I’m not going to go through all the exact combos for this one, but very similar to the previous hand if the oop player has any sort of flop or turn check raising range which 90% of players do this is a very easy value bet.

My other big complaint is the level of accuracy it gives. It should be giving a range of EV where its 90% confident or something. Not exact EV down the the decimal point for each bet. When reviewing my “errors” it says I made it would bring up all these spots where calling might be -0.03 bets and folding was 0 bets and says I made a mistake by calling. There’s no way it can provably say for example that I am losing money by calling a 3x raise with 96o from the bb like it believes. Also when I made a bet size it didn’t like it would tell me I was making an error by betting 2/3 instead of 1/2 pot in a single raised pot. There’s very little chance that is a provable mistake. The errors should be organized in largest to smallest and ones that fall under statistically insignificant or too small to be proven shouldn’t come up as errors IMO.

I’m sure they will get there eventually but for now I actually feel pretty good about man vs machines chances at heads up NL.

And some funny vines for you

Leave a Reply

You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>