**Keizer May 19, 2010 at 3:22 am**

http://blog.pokerjunkie.com/online-poker/poker-stars-testing-the-rngI once read an eBook about an alleged Inducer algorithm and Equitable Distribution being implemented after the shuffle to determine the outcome of the game. Although it does not make the actual hands predictable, it does make who will win predictable and gives you an added level of thinking beyond just plain old reads for good hands vs bluffing, and I have charted this.There are several basics that I’ve observed:- Net Winning Percentage- Showdown Winning Percentage- Number of Hands Played- Number of SeatsNet Winning Percentage (NWP) is the “number of hands won *without* showdown” divided by the “number of hands played”. This number is factored into determine who is dominating the table or potentially bullying (i.e. forcing too many people to fold all to often) since you never go to showdown. If this number is too high, it defeats the fairness aspect that keeps PStars players (even the worst of them) coming back, which means less of a rake if those players disappear. So, claims have been made that they include an equitable distribution algorithm to combat this.Equitable distribution functions as a mode of fairness to prevent any single person from dominating the table. So at a 9 person table, they’ve figured that the net winning percentage for everyone should be very close to 11.111111% or (100/9) over the long term. If this were a 6 man table, the NWP for everyone should be 100/6 or 16.6666667%. At a HU table, it should be 50/50. This is also why proponents for online poker have long touted only focusing on maximizing profit on profit-making plays. Why? Well, if everyone had an unlimited supply of chips and everyone limped in every single game for 1 million games, it would all boil down to 11.111111% winnings for everyone at a 9 man table and everyone would eventually end up with the same number of chips as what they started with. This is equitable distribution – but it only works this way if and only if you have an unlimited supply of chips and every single person pushed all in for every single game for 1 million games. In reality, this never happens, which explains why some people are actually making profit… mostly from high buyins and large blinds. At a HU game, the logical NWP should be 50% over the long term, but eventually, the man who wins the most money will usually have the lowest NWP.So, what does the Net Winning Percentage (i.e. wins without showdown divided by total number of hands) have to do with this? Well, it’s quite simple. If your Net Winning Percentage exceeds your 11.1111111% or will exceed the 11.11111111% after a win, then the theory is that the game will counteract against you to drop cards flop, turn, river that will favor the person with a lower NWP so as to bring theirs more in balance. Sure enough, after observing and charting 300 hands manually using the player notes section on Pstars, I was able to calculate each person’s NWP and predict the winner or the person’s next move, even if I didn’t know what they were holding. But it’s not only about the NWP, it’s also about your showdown winning percentage.Your “Showdown winning percentage” (SWP) is calculated for you on PStars. It shows how many times you have gone to showdown and automatically calculates how many times you’ve won. This is factored into determining whether or not you are playing good vs. bad poker, tilting, tossing your money away – which means (for Pstars) the likelihood of a rebuy if you don’t quit the program and walk away. Most poker books say this is your stopping point (i.e. knowing when to get up and leave the table without blowing your bankroll entirely). The simple way to know this if your SWP drops below 89% (or 100% minus 11.1111111%). So, you should never go to showdown unless you are absolutely certain you have the nuts. What this does is it encourages and rewards slow play so you do not earn money very very quickly and it discourages all-in preflops, high raises etc. Because you will automatically go to showdown if someone calls. So to counteract this, the algorithm is allegedly designed to favor persons with SWPs that are closest to 100%. Theoretically, a SWP of 100% means you are playing very very very very slowly and likely not playing enough hands to make a large enough profit fast enough. This is what PStars wants to keep money in the system as long as possible.So what happens when your SWP drops way below 75%? It’s very easy for example for your SWP to drop to 50% after having sent 2 hands to showdown and only winning 1 of them. Well, at this point, you have to start playing “clever” poker. This means, you have to win 8 more showdowns in order for it to go up above 89%. Well, Pstars knows you won’t normally play bad hands (donkey hands), because your end goal is to “make money”. In order to do this, you have to single out playng against someone whose SWP and NWP you’ve calculated to both be lower than yours, OR someone whose NWP already exceeds the allotted 11.1111111%. This requires observing the table and charting (using the notes) the percentage of wins at showdown, wins without showdown and comparing them to your own. What you will notice is that playing a donkey hand against a person whose NWP is extremely high will give you an opportunity to suck them out when they’re on a good hand. The problem with this (and programmers know this) is that if a TAG is on good hole cards, he will likely play them with very high raises. I tested this down in the 0.01/0.02 NL games, and sure enough by the time the river dropped, I pulled out with a better hand even with a donkey hand.So, what happens if both of you have NWPs that are within range (i.e. lower than 11.111111% on a 9-man table) and really good cards? Well, the person with the SWP closest to 100 will more than likely pull out the win if one of you does not force the other to fold before showdown. This is why there is the “Inducer”.The inducer (if it really exists) is what I see as the problem behind PStars, not the randomness of the initial shuffle for your hole cards. The inducer is designed to drop cards flop, turn, and river that will encourage action. This algorithm should be renamed “The Trapper”. It usually drops cards to make you think there is a possible hand you can make. I would like to believe that this does not exist, but in the 300 hands I charted and observed versus my own hole cards, I could not help but see how the cards dropped induced action between players. This is the reason so many people lose money, because the inducer is meant to keep you in the game rather than encourage you to fold. Sure enough, when I observed this, it was always the case that the person with the lowest NWP not above 11.11111111% won.But, what if you both have NWPs lower than 11.11111111% and similar if not exactly the same SWPs and both relatively good hands? Well, two things could happen: (1) the cards that drop flop, turn, river, will generate a split pot OR (2) the hand with the better starting percentage preflop will win the hand. The only time you will usually observe this happening is at the very start of a tourney where it is always the case in the 1st hand that the better hand with the starting win percentage will win. This is the only time the RNG is truly fair to anyone with a good poker hand, because the hole cards would have been distributed and flopped, turned, and rivered with the high entropy shuffle and no post shuffle equitable distribution. So how does PStars get around this in their fairness policy? Well, if you entered 1 million tournaments (all 9 person tables), and everyone dumped all in on the first hand at every tournament, then you would win your starting hand to continue on in the tournament around 12% of the time.Think about it (we’re going back to equitable distribution theory). In one freeroll, you have 9000 players. At 9-man tables, that is 1000 tables in the freeroll. If everyone dumps all in during the first hand, then best starting hand at each table will win. That knocks out 8 players per table all at once. That is 8000 players gone in one sweep in the 1st hand, with 1000 players left over, each having the exact same number of chips, each has the exact same showdown winning percentage (100%) and each has the same net winning percentage. BUT… here is the flaw. If you have 1000 players left and 9 man tables, that leaves you with 111.111111 tables – that cant be. There need to be at least 112 tables. Let’s ignore this numerical error for now and assume we can indeed have a partial table. The next time 1000 players go all in, that means there will be 1 winner for 111.11111111 tables (or 888.888889 people have been eliminated). That means 111.11111111 players are left in the tourney by the third hand, all of which have 100% showdown win percentages and the exact same Net Winning Percentage 0% in this case, so… the best hole cards would have won. Now you are left with 111.1111111 players and 12.345679 of these 9-man tables. Everyone goes all in again, and what do you have? 8 players going home, so 8 * 12.345679 = 98.765432 players going home. There are 12.345679 players left in the tourney, each has a SWP of 100% still and each has a NWP of 0%. Now we have 2 tables left in the tournament. We have to assume in this case that one table has 7 players and the other has 6. If both are all in, then 6 people go home at one table and 5 people at the other. This creates heads up play. Now each of you has a NWP of 0% and a SWP of 100%. Who will win this one? Well, if both are all in, then of course whoever started with the best hole cards will win. But… what if you all actually decide to play poker? Well, this is why there is the inducer and why the NWP and SWP are important, because those percentages are the predictors for anyone’s style of play and determining what is desirable play for a “fair game” (i.e. slow play, non agressive betting, and playing good poker)This also begins to explain why so many people win more at heads up SnGs over cash games. At HU SnGs if you fold often enough against your opponent, that increases your opponent’s NWP since he’s winning hands without a showdown. If you do this often enough so that his SWP exceeds 50% at the time you are in heads up play, then most likely any hand you play after that point will kick in the inducer, which will create an action flop and force a win to you to help balance out the NWP. The problem here, is it will take at least 2 hands played to win. Why? Because even though his NWP is higher, you will likely have the shorter stack from having folded your blinds. So if the NWPs are now uneven and you both go all in on an Induction play, yes, the program will force you a win to try to equalize the distribution of wins (in which case you double up); but he will still have chips in his stack, albeit very few. In essence, he will now have a SWP of 80% (9000 players was showdown win 1, 1000 players was showdown win 2, 112 players was showdown win 3, 13 players was showdown win 4, 2 players was the 1st showdown LOSS). So now he’s in trouble if he tries to play slow, tight, non-agressive poker. The guy with the better SWP and lower NWP will always take the win if you have a better starting hand. This is because, despite having the better hand, the program is designed to drop cards flop, turn, river that favor the highest SWP and the lowest NWP of any hand where action is present. So, the only way out for the guy with the lower SWP is to force enough wins back to you to get his NWP back in equilibrium. This means the ONLY way he can build his stack is if he forces you to fold every time before the river, but… he won’t be winning much by only winning the blinds… so this will take FOREVER at a 9 man table, which is what’s happening at most cash games. At the same time, the program is already working against you, because with each forced fold, you are increasing your NWP (i.e. wins without showdown) ten-fold each time you do it. This means at a 9 person table, for every 1 hand you win on a forced fold, you would have to forfeit 10 more hands to get your NWP back in check. So, the game is designed to work against you no matter how you put it once your SWP drops below 100%, and it is the same reason why your AA’s, AKs and other phenomenal starting hands are still getting cracked on inferior donkey hands. In all likelihood, it has nothing to do with the randomness of the shuffle, which is why people who have looked into this are not satisfied. All the observations are showing the trend that the wins will consistently be distributed to the player whose SWP is higher than yours and NWP is lower than yours if and only if the numbers are out of balance (i.e. above 11.111111% for a 9-man table 16.6666667% for a 6-man table, and 50% for a 2-man table). The equitable distribution piece of the puzzle will work its magic to dish out cards flop, turn, and river to give a win to help balance out that person’s Net Winning Percentage to equilibrium if theirs is too low relative to whoever you are playing against.I tested this as well once my SWP dropped below 100% down to 80% in a 0.01/0.02 cent table after playing into someone whose NWP I knew was already much higher than alotted, and sure enough, their AA got sucked out by my T3o when a T hit on the flop and a 3 hit on the river. At this point, my 4/5 showdown wins when to 5/6, which is now 83.3333%. That’s still not enough to get back in the clear (remember, we need 100 – 11.11111%, or 89% or better for SWPs). But, the only time I’ll get to this is once I exceed 9/10 showdowns. The question remains, how often will a skilled poker player be willing to donk it at high stakes that many more times AND long enough in order to win on a donk hand and move up their SWP? That’s right. They’re not, which is why your SWP only goes downhill from there if you continue to play good starting hands (which is what PStars anticipates). This also explains why donks get rewarded so much at PStars when someone had been dominating a table for so long. It also explains why your good hands get cracked when your SWP drops below 89%. The domineering players’ NWP is simply too high, and your SWP is an indicator that your are playing worse poker – so PStars knows it can get money out of you for a rebuy if you don’t STOP and WALK AWAY soon.So, yes… it may be the case where you have a good run of winning cards, but, this is the randomness at work… If you had 4 straight wins out of 10 hands, you’re likely winning them from forced folds without a showdown. But had it gone to showdown and the other was not convinced to play to showdown, then he likely would have pulled something on the river – how many times have you seen it happen on PStars? You’re ahead preflop, flop, and turn, and on the river you get sucked out. This is equitable distribution at work. So if you indeed won those 4 hands, you shouldn’t be playing another great hand, regardless of what it is, until you have folded 30 more hands to get your NWP back in the safer 10-11% zone. Folding 30 straight hands can take forever – and PStars knows this – so you either have to be patient and suck up folding good starting hands.Is there a way to speed up the number of hands seen so you don’t have to wait so long? Well, of course. If you multi-table at, say 10 hands, and fold a whole bunch of hands, this will do the trick. I tried joining 10 tables so I would at least get 1 good hand I could play at any single table. Each fold at each table made my “number of hands” go up by 10s. I started noticing very quickly that the number of good hole cards being dropped to me by the inducer started appearing much much much more frequently. The problem here, though, is when you are multi-tabling, it’s much much much harder to keep track of who is winning what and the NWPs of players at each table in particular. Looking at the previous win history will not give you any idea, because the stats tracks the number of hands and wins at ALL tables for the session… not just that one table. So, it would make it very difficult for you to keep track of someone else. You WILL notice, however, that about every 10 or more hands, if you have not won a hand, you will start to see some amazing hole cards being given to you, all around the same time. As I said, this is the inducer at work. It is seeing that your NWP is getting very low and is trying to correct the equilibrium and encourage you to play the hand. Does this mean you will win the hand at every table? No. It simply means you likely will win 1 of those hands at 1 of those tables. If you do not believe me, try it. Join 10 tables, take the min buy-in, pay the BB, and once you have a win on the board with 100% SWP and a NWP that is 10% or less, you will notice a series of good hands coming to you preflop – great face cards, high suited connectors, and plenty of high pairs. BUT, this could be fatal if you are playing against someone who has a lower NWP than you, so the correct move in these instance is not to raise the buy-in. If someone else raises, you fold. If they call and you see the flop, you will see which inducer is trying to encourage action and which Equitable Distribution algorithm is going to feed you the hand. So here is where you have to pay attention and play slow enough until you get to the river and are guaranteed the nuts.The only thing to be careful about is chip dumping, which is a built-in mechanism to combat people intentionally folding to give chips to other players. This is merely PStars’ excuse to cover up the fact that if your NWP remains at 0% while your SWP remains at 100%, then it is likely that you are folding even the BB just to throw off someone elses NWP. So, it’s plausible that they’ve disguisted this part of their algorithm as a mechanism to discourage chip dumping – which is the only reason your % of the flop seen at BB, SB is important in the Equitable Distribution algorithm. What this does is 3 things.First, it encourages you to play and act when the inducer is dishing out bait so that your “hands seen in while in XXX position” percentages stay within normal ranges (I don’t know what these are), which means you still lose some money.Second, it discourages bluffing and/or forcing someone to fold too often, because this increases your NWP, putting you over your allotted 11.111111% if not closer to it.Third, it encourages you to play all the way to showdown, even if you both checked all the way through — because even if there is a dead hand determined by a high card, with nice action cards to keep you in play for possible hands, it will kill your SWP and force it to drop below 100%. If you’ve won enough showdowns on legitimate play, say 10/10 showdowns and you lose the 11th showdown, then your SWP is 10/11 which is 90.909090%, and you will likely blow it off as variance and a bad beat. So, my suggestion if you do not have the nuts and there is a better hand on the board that you can imagine, you likely will want to fold this 11th hand to keep it at 100%, especially if there is not a lot of money in the pot and you were both scared to be sucked into the inducer.If you let your curiosity get the better of you and drop in SWP to this 90.909090%, the only way I see to get around this to bulk up your SWP is to stop value betting on the river if you are certain you have the nuts. This accomplishes 2 things. Your opponent will most likely check if you check on the river. This will force a showdown, and it will give you 11/12 showdown wins. This is 91.666667%, which is not as dangerous. If you do this long enough (and it will take forever to do it), then you can check your way to a 20th showdown win, which is 19/20 or 95%. Obviously you will never get back to 100%, but if you checked your way to …say… 99/100 showdown wins when you know you have the nuts, then you are back at a SWP of 99%. This gives you a lot of leverage to losing a showdown every now and then, but… the way the game is designed, it will take you forever to hit and keep these numbers. Most people end up with 50 to 80% before they lose all their money or quit playing for the day.So, I believe PStars is rigged — not in favor of any player in particular, but in the interest preserving equitable distribution of winning hands in fairness to ALL players and keeping you playing as long as possible. But, the problem is in overbetting and blowing all your money in one wad without focusing on bankroll management.There is only question that remains in my eyes, and this is: are these percentages ever reset so that you can start winning again if you just fuck up and hit a bad hand?Naturally… I think they are — every time you reload $$ at the cashier or every time you shut down and restart the program for a completely new session.But what if they weren’t? Pokerstars obviously keeps a record of at least your last 2000 hands on its server. Do you think these long term averages are what’s used to keep you in check? Maybe this is one of the reasons you have to keep rebuying, even when you’re on a good run of cards.The only problem is, if you’re rebuying all the time and never learn to play the slowest, and tightest, least aggressive poker possible – you will keep going deeper in the red. On the other hand, if you are playing the slowest, and tightest, and least aggressive poker as possible, PStars will reward this with wins as long as your SWP stays close to 100% and your NWP stays less than 11.1111111%. At this rate, however, you spend a very very very very very long time to move up in limits – unless you’re willing to risk it in a very high buy-in at a very expensive cash game where you can literally begin making lots just off of winning blinds that are extremely large. This is why some players are in it for the long haul and keep the eye on the prize for the long term, and is also the reason they try to play poker in a profitable way with good bankroll management. It’s the patience that’s being rewarded, not necessarily skilled poker play. And if they lose, on a good starting hand… well, they’re chalking it up to variance, when in actuality the only thing that matters are those stats on the 3rd tab in PStars. This is exactly what PStars wants. It wants you to keep rebuying and putting money in the system. And every time you do, you will probably get some monster hands, win a few, get a good run, and wham… you will plateau and go a very long time until your NWP is within suitable range for it to dish you a win.I was a Comp Sci and Math major, and these types of fairness algorithms make a lot of sense to me as a programmer/gaming modeler/number theorist. The only variable here is the human in the play. You can program the game to dangle a carrot, but you can’t predict how people will react if they’re carrot hungry. So, you have to do it with numbers and stats as indicators of how someone is playing. If programmers did not focus on equitable distribution remaining limited to 11.1111111% per player, they would not be able to make any claim of a “fair and equal game for all”, which they are required to uphold by gaming laws. Your job as a poker player, however, is to squeeze as many profitable wins out of that 11.1111111% as possible.So, sorry for the long winded post. The moral of the story is you can’t win money at online poker without playing well for a very very very very very very very long time. Most people don’t have enough time or patience to do that, and this is the only reason why people who are actually winning some money at poker like to bust on those who aren’t, by claiming that the other is simply playing shitty poker.Try looking at it from the NWP perspective, and charting the NWPs for people at your table, and I guarantee you will start to easily predict who will turn out with the winning hands at any given moment. You might not be able to pick out the cards they have since this is taken care of by the Quantum RNG hashed and mixed with white noise and user input DURING THE SHUFFLE, but you will at least be able to sit back and see how the inducer is generating action and see how the cards drop using the equitable distribution adjuster in order to dish out a win to the one with the lowest NWP and highest SWP between any set of players in a hand.I’ve done it.Works every time in predicting a winner.Now, if only I could stay disciplined long enough not to give in to the inducer.

## 0 comments:

## Post a Comment