Sunday, August 31, 2008

VPIP and PFR debate in Poker Calculators

Phaedrus, JP and Mattywein debate some new features that extrapolate deeper VPIP and aggression measurements in poker calculators.

I note your enthusiasm for this feature Matty. If I understand it correctly, you would like a 20 percent, or 30 percent hand range broken down into probable holdings and your hole cards matched up against say the average of that range.

Its a nice idea but I really think it will create too much techo reliance and lead to some pretty bad reads. For example the program would need to take into account position and action in front, Mzone probably ICM the looseness
, tightness and stack sizes of the players behind just to name a few variables. By the time you had reliable data with all these variables your opponent will most likely have died of old age.

Mostly I think of VPIP this way.

A sub 20 percent range is incredibly tight. The are playing big pairs and big cards and maybe most pocket pairs. The may play some suited connectors but not many. These guys are easy to read post flop and its great to be in pots with them because you won't lose much.

A 25 to 35 percent range is much more dangerous. These guys are playing quality cards with just enough suited connectors in the mix that any flop can help them.

Over 35 percent is still dangerous to play against. The individual player is playing too many hands and will most likely be a losing player over time. But they can have pretty much anything in their hand and so its very hard to read them correctly, particularly if they are also aggressive. Best approach is mostly to trap the before the flop or on the flop.

But notwithstanding all of that, you need to be mindful, especially in sit and go tournaments about what blind pressure is doing to players ranges. This is why this particular program feature would not be helpful to my game. Tight players know to loosen up when the pressure is on and their game will change dramatically

If you tracked me over a SnG you would find I am a rock or TAG at the start and a maniac at the end....virtually every tournament.

In an multi-table poker tournament online you will quite often track me as a maniac or calling station at the start and then Tag in the middle, maniac at the end. Or depending on stack sizes, maybe a rock at the end. Its all situation dependent.

A much more reliable read of peoples ranges is their stack size at the time. Small stack almost always = big range. Big stack can mean different things though.

One other observation is that I use Poker Office for my cash games at Pacific. The HU display put numbers all over the screen (quite distracting really) I can select which numbers I want to see from a massive list of indicators. But at the end of the day, VPIP, pre flop raise percentage, post flop aggression, went to showdown and win at showdown together tell a very accurate and complete story. I really don't care about how often someone steals blinds or how well they defend their blinds. Blinds are minuscule compared to stacks. I'd rather have the info that allows me to make the best decision for a 100 big blind bet than a poultry 1.5BBs

In tournaments blind steal and defense stats are also pretty meaningless. It all depends on the size of the blinds and the size of the stacks. most all players are thieves when under pressure and you soon learn how well the guys on your left defend without needing a calculator.

-----

I disagree with this. I think those stats are valuable and when reviewed in combination with other stats and the current situation can reveal a lot about the player and what they may be doing in that situation.

For just one example "defense", not sure what number you look at for that but I look at PFRR (pre flop re raise) AND combine it with pfrr-W$SD which means IF they pfrr then they W$SD If they don't win at showdown often when they pfrr then it tells me they pfrr too much and are more likely to defend with a weak hand so their re raise to me has a high probability of being a defensive re steal. Granted, EVERYTHING else about that hand, that player, the actions leading to that moment, stack sizes, the hands leading up to this etc... will ALL play a role in what happens and the decision making process. But stats, when used correctly, are a tremendous aid to a player in order to "play the player" when you can't simply play your cards. (how often they fold to steal bets in bb and sb are main defense stats but they don't play a part in this particular example unless they have high fold but this time are re raising and they have low re raise or hi win if pfrr. Then I fold)

-----

I hear what you are saying JP and agree with it at a theoretical level. The problem for me in practice is that there is a pretty small window in a tournament when blind steals and defense are either not a waste of time (ie. blinds are too small) or not pot committing (i.e blinds are so large any raise pot commits you and your opponents reaction is irrelevant).

I have had quite a few players tell me I am a blind thief at the table. When I tell them I have just had a run of good cards, several have then told me about my Poker Edge or Crusher stats. Apparently these must show I raise a lot from steal positions (which I do)

The problem is that this is not necessarily stealing, it is just correct position play. I play a wider range in late position and I typically open raise. But I almost always am playing solid hands for my position. So I really don't know whether that qualifies as stealing...to me is just seems more like pressurizing aggressive play

-----

Phaedrus I totally agree with you on this point, BUT

This statement sort of misses the point I was making. You are still relying on the 2 figures to make your judgment and you have to compensate due to other players inflection points and M values. The figure I'm after would give BEGINNERS a better understanding of where they would be in the hand and what the odds would be against the players range. You're still relying on a figure either way and making a judgment on that and all the other factors mentioned.

I do think this would be difficult to implement though, but only because a pro's 30% range would be a lot different to a beginners, Plus the math would probably be mind blowing. Confused Confused...

Saturday, August 30, 2008

Using a new Poker Calculator for Omaha

As you can see, here, I limped in with what Omaha indicator calls a premium hand, with 27 Hutchison points. The reason why this is not a top 10 hand is because of that eight being a suit of its own, and really offering no strength at all for the low or a possible straight. So in this hand, I leave you really playing for the nut low, because I have two of the deuces, seriously reducing the likeliness of another player at the table also holding Ace deuce. So in that case, again, there's absolutely no reason to raise as you really want as many people in the pot with a hand like this as you can muster.

The flop is somewhat reasonable for me, although there are no diamonds I only need one card for the low end and as you can see in Omaha odds calculator indicator I have 20 low outs. One player bets 90, I call, one player folds, and another calls, making the pot 435 with three players left in the pot.

The turn brings to three of clubs, which gives me the nut low and a reasonable expectation that I will not be sharing it. When you are in the spotlight this against two opponents holding the nut low, you do not want to be pushing either of your opponents out of the pot, because they are likely splitting the high or at the least fighting for it. So to have them both continue contributing to the pot is your main objective, which is the only reason I simply called phantoms that of 360 chips. Unfortunately cubsfan did fold, leaving a pot of 1155 chips, for me and Phantom to fight over.

The river brings a jack of hearts filling out a flush and a possible straight. Phantom checks to me and I checked down as well. There are a lot of players here who would bet in my position, simply because they have the nut low and feel they have nothing to lose by adding more of the pot. That is a huge mistake in this situation because as you can see, I have no chance of winning the high side of this pot, and Phantom could possibly have a share in the low side of this pot, and me betting out here might just result in a reraise, so there is absolutely no point in adding to the pot at this stage.

As it turns out he was afraid of the flush as well, but his set of nines easily took the high side, and he wisely checked it to me after the river which critically devalued his high hand.

Thursday, August 21, 2008

Poker Tracker 3 and Calculatem Pro

You may also find this of interest if you do like the program and would like to purchase it.

Poker Tracker

Basically people who have bought pokertracker2 and PA HUD will be getting a discount applied to their PT3 purchase. Although as far as I know they are planning on having the commercial release of PT3 somewhere around the end of this month. Keep in mind the release date has been changed a number of times now so that date may not be firm.

I purchased an odds calculating program a while back and from my own experience it was a waste of money. Calculatem Pro I think it cost around $100 and I used it for about a week before I ditched it. It does suggest what actions you might want to take, but as BNS has mentioned it is kind of bot like if you plan on doing exactly what it says. I also don't think it's suggestions are all that profitable...mind you it might be good in limit, I tried it out in NL which was probably a mistake in itself.

Wednesday, August 20, 2008

Poker Calculator for Omaha?

Sure it is a small market compared NL holdem, but with all the big pros like Gus Hansen and Phil Ivey playing omaha and omaha HL for hundreds of thousands of dollars most nights on Full Tilt, you would think that such a poker calculator would be successful.

It isn't like there haven't been attempts, but sadly they don't even rate good enough to well, rate. Mostly becuase you have to plug in your own numbers while the game is in play.... Yukkk.

But wait - word is that a new product from the makers of tournament indicator are on top of things again. A self reading, poker calculator designed specifically for Omaha and Omaha HL.... stay tuned - it should be good.

Saturday, August 16, 2008

The Math of the Deal.

Several years ago, a computer engineering team (can't remember the exact school they were from) decided to beat the slot machines. They managed to shell out the cash to buy an old model that was still widely used in their state's casinos. They reverse engineered the source code, and found that the slot machine's programmers had poorly implemented a common random number generator: they intended to make it simpler to work with the slot machine's limited computational capabilities, but without that key piece of code, the RNG's output was almost completely predictable based on its last few outputs. The team made over a million dollars playing these machines before they were caught and tried.

It's a fascinating story, but I can't find it at the moment.

However, the points are this: 1) you'd need to know the exact RNG and shuffling algorithms used by the individual poker site, and 2) the algorithms would need to be very poorly implemented. There would be no 'catch-all' solution. Furthermore, I'm willing to bet a large amount of money that the majority of poker sites use hardware RNGs as the source of their random bits, against which this attack would be completely irrelevant.

There's also the fact that a decent, well-implemented RNG would never (again, by "never" I mean "never in the lifespan of the universe") shuffle a deck of cards the same way twice. There are 52! (52 factorial) ways to shuffle a deck of cards. That's roughly 8 followed by 67 zeroes. If you can beat odds of that number to one, you don't need to play poker. Just go ahead and win a few lotteries.

So the database these people are compiling is useless. (Assuming they are even compiling anything, and not just spitting out random answers.)


I asked about this topic on PokerEdge forums under "fuzzy math". I was told that FT runs a reshuffle after hole cards, then flop then each street therafter to defeat any possible reverse rng program from working. The main one advertised states that it will show at least one other players hole cards and 3 cards that will hit the board by the river. I asked if this reshuffling would effect winning odds ie 2-1 of completing 4 flush by river ect. I was told that since it's the same number of remaining cards, that the odds did not change. I asked this after having, as most of you probably have too, my allin flop bet on trips or flopped straight being beat by AXs sucking out a runner runner flush on me.

I've also been card dead (Sklansky group Cool for 30- 50 hands and saw one section of the table getting a high concentration of group 3 and above.
But those suck outs are beyond my control. ie, some donk calling my AA allin push with J4s and flushing me out. When I actually get ITM in a STT, I still place first more than anything else. (20 finishes ITM 11 wins)

Wednesday, August 06, 2008

The most advanced poker calculator yet....



Ok I think it would be quite advantageous to have this kind of skill at the poker table. This could be the most advanced poker calculator on earth!

Tuesday, August 05, 2008

At least I was favored!!

A few random hands where my poker calculator shows I am a huge favorite to win, but things just didn't work out.

TI Profile ebook.

There is a free ebook at SitandGoCertified.com that deals with the traditional profile grid as explained by Dr. Alan Schhonmaker AND the profile icons that Tournament Indicator use when it profiles your opponents at the table. I think it is useful to combine the two to show where those icons are actually situated on the grid, and why.

Even if you don't use Tournament Indicator, a lot of poker calculators use the same or similar profile icons so this ebook - which includes 4 videos btw - should prove useful nonetheless.

Monday, August 04, 2008

Poker Crusher sucks, or no?

Some forum members are debating about Poker Crusher and its accuracy:

You are using a "free lookup" and as such it is just a hodgepodge data collection. It is the SUM TOTAL of every game type, cash or tournament, any buy in level, full table or short table, no limit and fixed limit ALL added together then analyzed.

Raz is more uniform or narrow in game types, therefore his "analysis" is going to be a bit more accurate as there are fewer fringe numbers added in to skew the result.

Since you have a lot of FL games in there and you also have short table games in there, you have a larger portion of looser style play numbers added in to the sum total. You do not play the same starters short in FL as you do NL and your positions you play are also different. Likewise if you are short table and HU you play different. You may have vp of 50+ for straight HU play, 25-30 for 5 or 6 way play, 10 for early stages, etc...

That analysis rating is everything thrown together and not a good barometer. It's there to entice you to look deeper. Now if you strip things out and filter them properly...

I've edited and generalized so as not to put specifics out but:

Sun137 FT poker, Full table tourney play, NLHE, Shark, TAG.
VP-(removed exact number but it's under 20)
PFR (Again removed exact but it's under 12)
W$SD (Over 50% is all I'll say Very Happy )

JP's profile summary: Alpha-T - If he raises, reraises or goes all in preflop, fold unless you have AA-KK-QQ as he is more likely to win the hand if it goes to showdown historically. Laughing

Now this is ONLY for a tournament of no limit hold em and a full table of 7-9 players on full tilt only.

Now, if I add in ALL the data from the cash games FL and NL the numbers shift upwards in all cases but one which is a hair downward by comparison. So you see, it's not an accurate portrait of a player as there are too many unknowns in the the matrix. That website is useless for profiling, but the program is not as you can filter out exactly what you want.

Just like poker tracker, it's usless to know you played 224,899 hands, you have to sort it out with various criteria to make is useful to us. AA had it 180 times means nothing as it's just the sum total of how often you had it. But filter it and sort by position, amounts won, hands played etc... and it starts to take on meaning that we can do something with.

It can also vary from site to site! On stars, I am tighter than Sun, on FT Sun is Tighter than me yet we seem to have similar playing style as in a tight aggressive play for NLHE.

But it is entertaining to look at and if you play MOSTLY one type of game, the profile can be close.

Friday, August 01, 2008

69% win odds so I decide to go for it!

Phaedrus75 did a better job of describing my own video, than I did myself.



I have only just gotten around to watching this, and I thought it warranted forum discussion.

In the video Marty shows 2 hands, the first being an all-in with middle pair and a flush draw (K8s) and the second is folding pocket kings to board with 2 Aces and a flush draw.

The two hands are diametrically opposite in terms of risk tolerance, but IMO were both perfectly played.

The K8s hand, Marty has an M of 24.5. He has a lowish table Q and I would expect that his tournament Q is probably pretty low as well.

He gets to see a free flop from the BB with K8s and flops middle pair and a flush draw.

In the video, Marty describes how he was feeling, it was late at night, he had a mindset of double up or go to bed and chose to ignore GCI and let it ride with this hand (which has to be said is a very strong hand on the flop).

I think the most important discussion point in this decision is not so much about green Mzone IMHO, but about poker tournament equity.

Despite his green M, the field is starting to get away from him at this point. The next blind increase will leave him with a yellow stack. He will have then lost the opportunity to play many potential double up type hands such as suited connectors and small pocket pairs. He must then wait for big pairs and AK / AQ type hands or potentially a few re-shoves over limpers.

When he plays these hands from a yellow M, he would be aiming to playing for his whole stack. He therefore MUST win those hands and must not get sucked out on. He must also leave it to pure chance that he will get a playable hand in yellow M before his stack dwindles further and the rest of the field pulls further ahead.

In terms of tournament equity, this situation is not very encouraging. Sure, he may come back sometimes, but to reach a stack size equivalent of what he would win by shoving the K8s hand (and to have that stack size match what his Q would have been by winning the K8s hand) he needs a mini-miracle. Certainly I would estimate that the chances of the come back happening are way less than 50/50.

So instead of waiting for that to occur, Marty takes a 50/50 poker odds chance with his still viable stack. IMO shoving (or in this case calling) the draw was significantly better in terms of tournament equity, than folding to live on to fight another day.

Marty says it was uncharacteristically LAG play which made him get all his chips in with a draw. Maybe it is a bit loose, but when you consider his standing in the poker tournament at the time the hand occurred, versus if he wins this coin flip (and and lets not forget coin flip was about the worst case) then IMHO it is an insta-push or insta call every time (not just late at night).

I have been toying around with the idea that Mzone needs some adjustment for on-line tournaments to account for the speed of the blinds (even in non-turbo MTTs) as well as loose players.

Mzone has been a concept which has been around for decades. It was first given a name by Paul Magriel of backgammon fame in the 1970s I think and since then, has been applied to all tournament situations.

However when the Mzones were first developed (and I don't know if Harrington was the first to do this, but rather suspect the first to publish it), then it stands to reason that most of the Mzone theory is based on bricks and mortar tournaments (and tournaments which were being played before the Danish HLAG brand was injected into the game).

Whilst I don't think radical transformation is required for on-line, there are a couple of differences. Firstly, fast blinds and relatively shallow starting stacks typically mean that in the early middle stage, you are not that comfortable with an M of say 22. Secondly, lots of LAG on-line poker players at the start of MTTs tend to accumulate monster stacks very quickly. Unless you keep up with them to the extent that you have fold equity over them later, these are the guys who will in the later stages run over the top of you and/or suckout when you finally do get AA.

Accordingly, at about the same stage of the tournament Marty is at with the AKs hand, I have tended to be thinking of a 25-30M stack (in the early/middle stages) as boarderline yellow, meaning I can still play suited connectors and small pocket pairs in the right circumstances, but I am pretty happy to play for my whole stack with way less than the nuts.

The KK hand, was of course an excellent laydown in the face of what seemed very likely to be a bluff by a big stack. However tournament equity was now much higher and the big green stack again needs to be treated with divine reverence.

These are just my thoughts on the video and I'd be interested to hear what others say.