Tuesday, July 31, 2012

My poker notes (literally)

I've had poker thoughts brewing in my head for a while; things I could almost see, like images.  Was actually giving me a headache as they swirled around, but couldn't quite gel.

Pieces finally started snapping into place today, and with each piece the others came faster, until they all were totally lucid, and I quickly scratched them down.  Too tired to put them into electronic form today other than scanning them.  Plus I thought it would be cool to have these available to look back on...

...they're my most comprehensive, important thoughts to date.  I'm sure of this because I've both used them in part at high-stakes HU LHE play with great success, and the rest of them directly relate to those that I've used.

So, without any more talk, if you can decipher them, then congrats - feel free to use them in your HU LHE game!  Otherwise you may have to purchase my future book :)

Lastly, a few extra things:

These are specifically for HU LHE; some things I'm fairly sure don't apply to other forms.
Bluffs don't make profit; they recoup loss they've already incurred.  They always lose value effectively.
VBets and VCalls are the only places to make value.

Things still needing to be addressed:

What happens with pre-flop contractions/expansions of ranges IP and OOP?
Also, I'd like to develop a way to play both sides of a match against myself, so I can always practice with somebody who wants to 'train,' i.e. myself.

Sunday, July 29, 2012

flop donk > CR ? I think it's likely.

I've got 5 reasons why I think a flop donk is better than a CR when OOP in HU LHE. Here they are, in order from my rough estimate of importance, from lesser to greatest: 5) by acting first, you force your opponent to choose the correct response 100% of the time, and since no player is perfect, you force them to make the first mistake. 4) I actually WANT the times that I check/check-call the flop/turn to be exploitable, where I'll often call the flop but then fold the turn. This will encourage my opponent to CBet-bluff the turn, which allows me to then shift otherwise VBet hands into my check-calling range to gain value (while bluffing more in my betting range), and then to shift those hands back if they catch on, allowing my weak hands to get free cards on the turn (and then VBetting more strongly when they're then calling more often vs. my donks). If I CR the flop, I don't have any options to incentivize my opponent to make these mistakes. 3) your opponent never has an option to simply check back the flop or turn to realize more equity with hands you want to VBet/protection bet against (like AH/KH hands, or QH with a GS, etc). (note: even if they raise the flop for a free turn card, then they'll run into my #1 point below). This becomes an even greater issue if they tighten up pre-flop and have a higher % of HC hands on average when there's a lower flop/turn. 2) you're getting 20% price on the flop for pure-bluffs instead of 33%, which I feel is more important than getting extra value from value CR on the flop since in the end, VBets/raises are EV neutral. 1) you force your opponent to put in extra bets for a bluff raise, giving yourself either 2:1 profit on your own Value:Bluffs if they raise the flop, or 2.5:1.5 if they raise the turn. This means you can get paid off on your value hands from their bluffs at a higher price to them, which allows you to bluff twice as often as you VBet (in the 2:1 structure). Ex: if you VBet (donk) 20% of the time on the flop, you can donk-bluff as much as 40% of the time, while if they try to bluff-raise the flop, in small bets your EV = 2(0.2) - 1(0.4) = 0. For a flop CR, you're getting 3:2 on both either the flop or turn, which if you V-CR 20% leaves you with EV = 3(.2) - 2(.3) = 0 for EV neutral, showing you can only bluff max 30% of the time. This feeds back into point #2 above, where not only can you bluff more often, but those bluffs are then more profitable.

Tuesday, July 24, 2012

Poker Outline & KISS approach

While poker has been categorized previously into 2 categories (Assumptions, Best Line), I see it as 2 main categories that are slightly different, and which are broken further into 2 categories each, as follows: 1) Assessment a) Assumptions b) What's possible 2) Action a) What's profitable (parts I and II) b) What's MOST profitable In #1, a) assumptions are tied with knowing how to b) read boards / count combinations. For example, you can't assume an opponent is bluffing 70% of the time when they only hold 50% of garbage hands based on the board. More on #1 in a bit (it's where I'll discuss using the KISS approach). In #2 it gets complex. To decide if an action is profitable, you can't only look at the hand you hold in a single scenario, ex: bluffing and having your opponent fold. If your opponent also folds when you hold a made hand, you're not making money with your made hands. #2 a) part I: it's the EV of i) each of your hand ranges, ii) via an action you take, and iii) responded to by your opponent with each of their hand ranges. I call this 'vertical EV' because it is a composite of your best through worst hand ranges across scenarios. *note: this is where you consider balance. #2 a) part II: You must also then consider what I call horizontal EV, which is your vertical EV taken through the actions/streets of a hand, as far as you figure you can make accurate assumptions. *note: this is where the KISS method shows itself in the EV. Doing these #2 steps allows you to evaluate a plan of actions that you can take vs. a "model opponent", an opponent for whose actions you define/know. #2 b) is comparing identified profitable scenarios for the 'model opponent' and choosing the most profitable. This is tough enough on it's own, but then if you do a 'what if' for your opponent's actions, it can easily turn into a multi-variable problem, which basically means you can't solve it. Better might be to take a different given opponent, and repeat steps #2 a) and then b). ------------------- KISS approach --------------------- In #2 everything is away-from-the-table; it's harder. It's also less useful. Knowing how to do #1 really well can pay off much more than doing it poorly and also trying to do #2. Keep in mind that you'll change your strategy frequently as you adapt to your opponent, making any research you did in #2 obsolete if you don't know what to do next. In contrast, you will make decisions based on the board texture/assumptions every decision point of every hand. Just keeping an eye on your balance needed in #2 part a) will ensure you don't have any 'holes.' Also keep in mind that you don't actually WANT to play balanced. You want to play the opposite of what your opponent thinks you're doing. For example, bluff way too much to get them to call, then stop bluffing and VBet more so they'll pay you off. Or, in reverse, only VBet, then when they start folding a lot, stop VBetting and start bluffing. Another way to think about it is that you also aren't trying to profit with all of your actions all the time. You expect to fold your garbage and only VBet sometimes (losing value when you fold your garbage, but making even more with your VBets). This is really important to understand! You CAN'T profit with EVERY action EVERY time; it's always a trade-off. Making these adaptations as early in the hand as is profitable (i.e. if your opponent won't immediately catch on if you do it pre-flop for example) is usually the most profitable because your opponents actions in response, if incorrect, will compound their loss (your profit) throughout every hand. -------------- Start #1 work: Two tasks -------------------------- Task #1: To start focusing on #1 then, we should do a case study to get an idea of how much it's worth. For example, change your pre-flop range vs. the 'model opponent' and see how much your EV changes. Referencing my previous statement that profit is always a trade-off, you may fold more pre-flop (losing your blind) to gain many big bets from your opponent when they think you have weak hands but don't, and they call down. Task #2: Learn your pre-flop ranges by 'squares' on the charts, in 10% range increments for 'paired' hands, then also for HC 'high card' hands; then for composite ranges (ex: top 60% - top 20%) for each of those. Don't try to do combos directly! Two quick reasons for NOT doing combos: i) too hard to count; not enough benefit, and 2) hands change based on square (like when an 'out' comes to make a straight), so you still need to keep track of the 'squares' anyway. It's the KISS approach again; do the 'squares' 100% and you'll get more benefit than doing it poorly and trying to do combos.

Saturday, July 21, 2012

Fluent Poker (& wine)

Fluent Poker (& wine) Had a great wine last night; 2010 pinot noir from Wollersheim winery, a winery in Prarie Du Sac, WI. It runs about $14, and was great! http://www.wollersheim.com/ Working at my new job, I realized how much information a person works at mastering to have a job where you get paid to know technical processes, etc. You are basically learning to be fluent in a language. I think poker is like this too. There's no quick way or trick to learning poker; you just have to force tons of information into your brain for practice, and repeat. Many times I feel that I still haven't really become fluent in poker; I've simply 'hard linked' scenarios to outcomes, but this is like memorizing phrases in a language; you don't really know how to speak it unless that specific scenario presents itself again. To become fluent in poker, I feel that you have to know these things: 1) How many squares are on a poker chart for the top 10% range increments (i.e. top 10%, top 20%, top 30%). For example, the top 30% has 19 squares (where AK is a square, AQ is a square, etc), plus all the pocket-pairs (PP) 55+. 2) How many made squares of made hands are possible based on a board card in each range. Ex: In the top 30%, if an A comes, there are A5+ = 9 squares of made hands (pairs). This accounts for most of what you'll deal with in wider ranges, the rest being PP or HC (high-card) hands. Ex: top 30% has 19 squares, 19-9 = 10 HC still remain, plus 10 PP hands. 3) After this learn to subtract top 10% pre-flop range increments from each other. Ex: top 10% has 4 squares and 7 PPs (88+). Top 30% has 19 squares and 10 PPs (55+). So the range top 30% - top 10% has 19-4 = 15 squares plus 10-7 = 3 PPs. But also beyond this know which hands for each rank, as in #2 above. Ex: top 30% had A5+, and top 10% had AJ+, so top 30%-top 10% has A5 - AT = 6 squares of Ace made hands (pairs). 4) Then do all of this given 2, then also 3 board ranks. 5) Once these are mastered, then learn how many HC are in each range given the ranks on board, then in each subtracted range scenario. This will account for the majority of hands you'll see, especially in HU LHE where the ranges are wider (my particular game of choice). You'll start to get a sense of how 'heavily' flops hit ranges, etc. After this, though, there is still much more to consider: scenarios where I hold a pair, double ranked boards, ST 'straights' (which will take up squares on the chart similar to pairs, but between board ranks), FL 'flush' (which will be roughly a % of HC and PR 'pair' hands available), 2PR 'two pair', trips, etc. Then also OE & GS draws.

Wednesday, July 11, 2012

A bit of everything, esp Poker (Esp HU LHE)

I decided my blog is going to be for me, and of course anyone who wants to read this drivel! So I'm going to blog mostly about what I'm doing/interested in that day, whenever it strikes me (as most terrible blogs do!). Namely: poker (a lot of this for sure! Especially HU LHE), playing guitar, staying fit, drinking wine, learning to speak Thai, or my websites (plug: www.thepeopleswines.com , www.hulhe.com , www.hulhe.com , etc.). Today is poker!


 Value betting IP is different form VBetting OOP (will discuss more distinctions in a later post). One main difference is that your opponent is often already betting and when you're IP you can just call, or you can VRaise.

VRaising is like VBetting in that you want to be a 50% favorite vs. your opponents value range, but VRaising is different than VBetting because when you're opponent is bluffing, you have an option of collecting bets from bluffs by calling, where as those hands will likely fold if you raise. So the decision pits value from inducing bluffs vs. value from their made hands.

You collect 1 extra bet from their worse value hands, but only those which are > 50% of their value range, but still worse than your hand! This is because if you're exactly a 50% favorite, raising is a wash; you'll gain 1 extra bet from worse hands, but lose that extra bet from their better hands; it's a half-life scenario that continues until betting is capped, back and forth with your opponent.

Seen another way, for example, they 3-bet pre with top 30% and fire the flop & turn. If they held 20 possible hands and would call with 10 of them, if you hold the 4th hand from the best in that list, you're a 50% favorite vs. their top 8 hands, and also vs. 2 more lower calling hands.

If you raise, you gain 1 bet for 2 hands = 2 bets gross. But if they were bluffing with 3 hands, by just calling you'd gain 3 bets gross.

Flow of thought: count # of opponent's bluffing hands. To VRaise, count # of opponent's better hands, double this, and add the # of hands they were bluffing with; this is how many hands with which they'd need to call for a VRaise > VCall.

Summary: a value raise vs. call IP must compare # of extra calling hands beyond 50% worse vs. # of hands pure-bluffing. Whichever is greater determines your more profitable action with your IP value hands.