horrible downswing or horrible play?

    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      Lately, I have been playing the hyper-turbo sngs on PokerStars. Since the entire tournament is basically jam-fold, the variance in BI is expected to be high. Here, I will share some stats that I have, suggest a few conclusions, and ask you for input regarding the conclusions that I have drawn.

      The structure of these tournaments is as follow: 6-max, 500 starting chips, 3-minute blinds (25/50/10,50/100/20,100/200/40,...), top 2 get ~2.5BI back, 3rd gets ~1BI back, 2% rake.

      number of tournies: 254
      number of hands: 2883
      ROI/$won: -3.9%/-$369 ($37 avg buy-in)
      ITM%: 46.1%
      chips won: 9331 (500 to start)
      EV chips won (HEM): 24622.19
      $EV: unavailable - I cannot find any software that can correctly calculate this for satellites that end when two players remain.

      Conclusions: chip EV and $EV do not necessarily go hand-in-hand. Since I won a significant number of chips but lost a significant number of dollars, I am either making +chipEV -$EV calls, or opponents are making these when I push. Is there another explanation for this?

      SNG variance is very high. Hopefully someone else can comment on this since I have little to compare it to, but the difference between my chip EV and $EV is very high. Even when adjusted for the size of the BB, I have never heard of an EV diff that high (I previously played cash games). Because of this, I am not sure if I am a losing player or a winning player with some very bad luck. Is there any way to estimate $EV from the data that I have here?

      If anyone can help to explain the disparities between $ won, chips won, and EV and/or comment/speculate on my long-term profitability in these, I would greatly appreciate it.
  • 24 replies
    • miskokvo
      miskokvo
      Bronze
      Joined: 28.03.2008 Posts: 1,502
      lol 254 tournies is no sample size for such turbo sng ... go up for 3000+
    • AugustusCaesar
      AugustusCaesar
      Bronze
      Joined: 24.05.2008 Posts: 377
      Hello, my questiong may be a little offtopic, but please, can someone answer this ?

      1) Does "EV chips won (HEM): 24622.19" mean the expected value of chips won when pushing preflop?

      2) Where can I get my stats of this when I play SnGs ?

      3) Does it mean that if the value of "EV chips won (HEM): 24622.19" is much bigger than "chips won: 9331 (500 to start)" you where running bad, so you won less chips than you would in the long run (suckouts and so) ?

      Thanks a lot
    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      lol 254 tournies is no sample size for such turbo sng ... go up for 3000+


      Note that I am not asking "Is this what my ROI should be?" I am asking if my chip EV difference could {impossibly, unlikely, possibly, likely, definitely} mean that my EV ROI is positive. If you cannot draw any definite conclusions from the sample size, at least offer any input that you can in the form of uncertain conclusions. Being 30 BI of chips below chip EV after 254 tourneys seems significant to me, but I was hoping that others could provide insight. I do not
      want to continue to play if, given average luck, I might expect only to achieve a negative ROI closer to 0.

      Originally posted by AugustusCaesar
      Hello, my questiong may be a little offtopic, but please, can someone answer this ?

      1) Does "EV chips won (HEM): 24622.19" mean the expected value of chips won when pushing preflop?

      2) Where can I get my stats of this when I play SnGs ?

      3) Does it mean that if the value of "EV chips won (HEM): 24622.19" is much bigger than "chips won: 9331 (500 to start)" you where running bad, so you won less chips than you would in the long run (suckouts and so) ?

      Thanks a lot
      1. It means the expected value of chips won when all players are all-in and there are still cards to be dealt. For example, if you push AA and get called by KK on a with a flop of {A,7,2}, but see 2 K's come up on the remaining streets, your EV chips won will go up almost as many chips as you lost, but you chip won will go down by exactly the number of chips that you lost.

      2. I assume that you have HEM. If so, Tourney -> Reports -> click the plus under the stat-selection list -> add $ (EV adjusted). This will give you the chip ev stat (it treats it like a cash game, so $EV is actually chip ev for tourneys on HEM).

      3). You are correct. I started each of the 254 sngs with 500 chips. I won 9331 chips, but this was actually losing over 15000 more chips than I should have.
    • AugustusCaesar
      AugustusCaesar
      Bronze
      Joined: 24.05.2008 Posts: 377
      Firstly, thanks cyzo for the answers :-)


      I am asking if my chip EV difference could {impossibly, unlikely, possibly, likely, definitely} mean that my EV ROI is positive. If you cannot draw any definite conclusions from the sample size, at least offer any input that you can in the form of uncertain conclusions. Being 30 BI of chips below chip EV after 254 tourneys seems significant to me, but I was hoping that others could provide insight. I do not
      want to continue to play if, given average luck, I might expect only to achieve a negative ROI closer to 0.
      Let me repay you in a form of a bit theoretical math aspects and some conclusions I made:
      - If you could provide the table of all the tourneys with the ROI and Chip EV than we could use some basic regression models to see what is the relation between these two parameters and make some conclusions, however, as MiskoKvo pointed out, the sample size is not so great (you should have at least about 1000 observations if the errors are Normally distributed for a 95% confidence level)
      - Simplier approach would be in my opinion to find out how many chips in average are needed to win the tournament (lets say before the push-fold starts), finish 2nd or 3rd .. You have an average of 96,93 expected value won chips, so we could compare it to that

      I hope I was at least a bit helpful :-)
    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      Originally posted by AugustusCaesar
      Firstly, thanks cyzo for the answers :-)

      Let me repay you in a form of a bit theoretical math aspects and some conclusions I made:
      - If you could provide the table of all the tourneys with the ROI and Chip EV than we could use some basic regression models to see what is the relation between these two parameters and make some conclusions, however, as MiskoKvo pointed out, the sample size is not so great (you should have at least about 1000 observations if the errors are Normally distributed for a 95% confidence level)
      - Simplier approach would be in my opinion to find out how many chips in average are needed to win the tournament (lets say before the push-fold starts), finish 2nd or 3rd .. You have an average of 96,93 expected value won chips, so we could compare it to that

      I hope I was at least a bit helpful :-)
      The entire tourney's are push-fold. Third place always gets paid their buyin back but always ends with exactly 0 chips. First and second get paid out the same amount but would average out to 1500 chips per person.

      On a theoretical level, I do not believe ((chips won) - (chip EV)) to be normally distributed. chip EV is an average, and someone who is most often ahead will therefore have less room to be above EV. As a result, that player would be above EV more than half the time, but the times that they were below EV, they would be further below than when they were above EV (data skewed left).

      Since these tournaments are satellites that end when the final two players are reached, I have found HEM's $EV analysis to be wrong (Their ICM analysis assumes first is paid out 33% of the prizepool and the rest of the money disappears), and juk's SNG Luck Analyzer returns errors since the tourneys are never played to conclusion.

      Unfortunately, I do not have chip and chip EV stats for each individual tournament.

      Edit: I do have $ and $EV values counting only tournaments I lost or got 3rd in:
      $-2244.1 $EV -459.155

      Obviously, it is to be expected that these would underperform EV and that the tourneys that I won would outperform EV. The $ for the tourneys that I won would be ~$1900. Therefore, the $EV would need to be ~$500. I am not sure what the actual value is, though.
    • Anssi
      Anssi
      Black
      Joined: 03.07.2008 Posts: 2,173
      Originally posted by cyzo

      Conclusions: chip EV and $EV do not necessarily go hand-in-hand. Since I won a significant number of chips but lost a significant number of dollars, I am either making +chipEV -$EV calls, or opponents are making these when I push. Is there another explanation for this?
      Well, that's pretty much what ICM is for.


      Btw these are only tournaments I have made -$ :( . Might be variance or people having too loose calling ranges, not really intrested to spend more time on finding out.
    • Meiffert
      Meiffert
      Bronze
      Joined: 13.10.2008 Posts: 151
      Originally posted by cyzo I am asking if my chip EV difference could {impossibly, unlikely, possibly, likely, definitely} mean that my EV ROI is positive.
      Hi, with the payout structure 40/40/20 (or something similar) the chipEV and $EV can be very different (only in winner takes all tournaments and cash game it's the same). You cannot make any conclusions from this statistics as the ICM is different in every situation.
      You have to analyze individual hands using ICM (you can you software as SNG wizzard) to find out the answer whether you are running under $EV.
    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      None of that software works for these tournaments since they end with 2 players left, and nothing has that structure programmed in as of yet. I contacted holdem manager support, and they intend to fix this in the next beta (should be just a few days). I think I can definitely conclude that I am behind EV. The chip EV difference gap has grown to ~60 BI worth of chips, and it has only been 300some games. Whenever I get that analysis, I will probably look for a backer to stake me. I understand that such an arrangement would be -EV for me, but I need quite a bit of money for university this upcoming semester, so I would be willing to lose profit in exchange for a reduction in risk.
    • Anssi
      Anssi
      Black
      Joined: 03.07.2008 Posts: 2,173
      SnG wizard works, you can modify prize pools, blinds, antes and stacks.
    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      Originally posted by Anssi
      Originally posted by cyzo

      Conclusions: chip EV and $EV do not necessarily go hand-in-hand. Since I won a significant number of chips but lost a significant number of dollars, I am either making +chipEV -$EV calls, or opponents are making these when I push. Is there another explanation for this?
      Well, that's pretty much what ICM is for.


      Btw these are only tournaments I have made -$ :( . Might be variance or people having too loose calling ranges, not really intrested to spend more time on finding out.
      I saw you playing a bunch of these today. How are they going?

      I contacted the support for Holdem Manager, and their software now performs correct $EV calculations for these, giving me the following graph of $won and EV$won (>1k SNGS):


      (http://img176.imageshack.us/img176/7241/aug.jpg)

      That's over $2k below EV. Is it even reasonable to run this badly over 1k sngs? The most recent downswing there is starting to diminish my hope. Perhaps I will spend some weeks reevaluating my play in an attempt to improve it. Any comments?

      edit: played over 100 today - lost over $1k - ROI was 20% below EV ROI
    • Anssi
      Anssi
      Black
      Joined: 03.07.2008 Posts: 2,173
      Started with nice upswing, +1k. Then came horribly down, probably -2k. I decided not to play these until I have at least 250 Buy-ins. The swings are awful. It's annoying when you play according to ICM and opponents are even tighter than they should except for few idiots yet you still lose to "luck". Especially enjoyed the 1 minute where my aces got cracked 3 times, 2 twice my Ax. Notes are very useful here, you know who you can push into and who will call with K3o.

      Well, at least I run good/ok in normal sit&goes :)
    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      I play very close to the nash equilibrium. I try to make it exploitative when I can. The blinds do not call enough as the ante would merit when you are not on the bubble, and I am sure that I gained most of that $EV from others playing too tightly, but there are still those players who will call you on the the bubble with, going with your example, K3o. Your 55% equity will be enough to ensure that both of your lose money. I really should take more notes on that kind of stuff, but it is hard when you are playing with stacked tables. Maybe its time to break my monitor out again. I actually had three hands at the same table in a row where I lost KK vs A2o (after previously doubling up), tied with KK vs QJo, then lost AA vs KQo.

      I am sure that 250BI would be enough, but playing 1000 games and being 50+BI below EV would still hurt, especially if you are going to play those high buyin ones ($87.72).

      The hourly winrate is just sickening ($EV winrate - not actual) for these if you can put any volume in. Let me know if you want to discuss these or anything, and good luck with your normal sngs until you decide to try these again. I am going to play some today (Sunday is the day when the games are softest, and some 0% rake sngs pop up every now and then), but I will probably take a break for a little while. Yesterday was over 1.5k below EV.
    • shortfuse
      shortfuse
      Bronze
      Joined: 02.07.2009 Posts: 450
      Originally posted by cyzo
      I play very close to the nash equilibrium. I try to make it exploitative when I can. The blinds do not call enough as the ante would merit when you are not on the bubble, and I am sure that I gained most of that $EV from others playing too tightly, but there are still those players who will call you on the the bubble with, going with your example, K3o. Your 55% equity will be enough to ensure that both of your lose money. I really should take more notes on that kind of stuff, but it is hard when you are playing with stacked tables. Maybe its time to break my monitor out again. I actually had three hands at the same table in a row where I lost KK vs A2o (after previously doubling up), tied with KK vs QJo, then lost AA vs KQo.

      I am sure that 250BI would be enough, but playing 1000 games and being 50+BI below EV would still hurt, especially if you are going to play those high buyin ones ($87.72).

      The hourly winrate is just sickening ($EV winrate - not actual) for these if you can put any volume in. Let me know if you want to discuss these or anything, and good luck with your normal sngs until you decide to try these again. I am going to play some today (Sunday is the day when the games are softest, and some 0% rake sngs pop up every now and then), but I will probably take a break for a little while. Yesterday was over 1.5k below EV.
      do you even know what this means? elaborate please
    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      Originally posted by shortfuse
      do you even know what this means? elaborate please
      I do know what this means. Your use of the word "even" implies that you do as well, though, if you are questioning whether or not I know what that means, I would certainly come to question your understanding of this. If I am misinterpreting you, and you are genuinely unaware of what this means, I would be happy to elaborate. Please get back to me and clarify what you are asking and why you are asking it.
    • shortfuse
      shortfuse
      Bronze
      Joined: 02.07.2009 Posts: 450
      Originally posted by cyzo
      Originally posted by shortfuse
      do you even know what this means? elaborate please
      I do know what this means. Your use of the word "even" implies that you do as well, though, if you are questioning whether or not I know what that means, I would certainly come to question your understanding of this. If I am misinterpreting you, and you are genuinely unaware of what this means, I would be happy to elaborate. Please get back to me and clarify what you are asking and why you are asking it.
      wtf?

      explain how you 'play very close to nash eq' because i'm not sure what you mean?!
    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      Originally posted by shortfuse
      wtf?

      i have no fucking clue what im talking about
      I fixed your post.
    • shortfuse
      shortfuse
      Bronze
      Joined: 02.07.2009 Posts: 450
      Originally posted by cyzo
      Originally posted by shortfuse
      wtf?

      explain how you 'play very close to nash eq' because i'm not sure what you mean?!
      I mean that my play is based primarily on pushing and calling with the handranges which are optimal vs your opponents ranges, having taken into account the $EV of winning, losing, tying, stealing the blinds/ante, folding your blind/ante, folding and letting other players possibly go all-in and give you more $EV and given that your opponents are aware of everyone else's ranges and play perfectly based upon those. Of course, we can only estimate what these ranges might be; even the programs we can use to practice such situations give estimates (with the exception of HU NLHE, for which the equilibrium has been solved). We also make the false assumption that pushing and folding are the only options, though other plays when the blinds are high make you exploitable. Unexploitable strategies with very low that include moves other than pushing/folding AA only have been found to be better in HU play, but the SB is still very much -EV even for those.

      When I say that I my play is close to what I have mentioned above, I mean that I am confident that I can come up with a reasonable approximation to the equilibrium strategy. Also, mentioning exploitative play in itself necessitates that the equilibrium is not followed, as such play is always either suboptimal or break-even (when it is suboptimal, it is almost always still profitable). For example, when the blinds and antes are so high that I can push 60% from the SB and the BB should call with 40%, I ask myself whether he really will call with some of the worser hands in that range, call with hands below that range, or call with that range exactly. If he is too tight, I will push wider. If he is too lose, tighter.

      I hope this answers your questions. Also, I saw on your blog that you refer to yourself as an "aspiring economist." I am not sure precisely what you mean by that, but that interests me, for much in terms of modern poker theory stems from the intersection of mathematics, economics, and computer science. Perhaps you would like to discuss some of this in more depth.
      Well I'm sorry but your pretentious babble still does not justify 'playing close to nash eq'

      nash eq for what? The nash eq is result of play, it is only present after discovering and dissecting what players would have done. In which case, you make no sense saying you play close to the nash eq as if this was some magically unexpolitable path.

      For a nash eqilibirium, we would need an accurate (I'll let you approximate) payoff matrix, and then we can work out what the dominant strategy would be- what both players would do and hence what the payoff would be.

      Once you provide evidence for this, you can say you chose the dominant strategy- NOT you played close to the nash eq which doesn't mean anything. In fact game theory has proven that in many situations it would be benificial for both palyers to cooperate rather than to defect- in which case the nash eq would not the the most optimal.

      If you could roughly demonstrate what you mean with a payoff matrix- I'll let yoiu assign arbitary units for relative stregths of hands you think you are facing against. It is up to you to provide a payoff matrix.

      If you are still following, you will realise you can only choose YOUR strategy and given the opponent's strategy does not change (which we must assume he/she always takes the dominant strategy) then you can claim you made the least -ev decision. But you cannot claim you somehow 'played close to the nash eq' which means NOTHING- the eq is a resuilt of both players strategy and is not the PERFECT method, it is the RESULT.

      Hence, it still leaves me worried that you are unsure of what you are talking about and just randomly throw in some words in play.

      To reiterate- the point is you just misused and confused nash equilibrium to the point where it makes no sense (on so many levels)

      edit: as for my blog titled 'aspiring economist' it is because this is a subject which is of great interest to me, along with mathematics.
    • cyzo
      cyzo
      Bronze
      Joined: 19.07.2008 Posts: 117
      There is a singular equilibrium strategy.

      We can find what this strategy is. We can come close to this optimal strategy. We can calculate an approximation which is similar.

      Why do you believe that one cannot employ an equilibrium strategy? If one can, it is obvious that one could also modify this strategy slightly. In this case, it would still be a strategy.
    • shortfuse
      shortfuse
      Bronze
      Joined: 02.07.2009 Posts: 450
      Originally posted by cyzo
      You seem to have no idea what you are talking about.

      The nash eq is the result of play? It does not matter how your opponents play; there is only one equilibrium strategy. Push-fold poker is such a game where an unexploitable strategy exists.

      I do not see what you do not understand.

      There is a singular equilibrium strategy.

      We can find what this strategy is. We can come close to this optimal strategy. We can calculate an approximation which is similar.

      We know exactly what the matrix looks like. It is an n-dimensional matrix, where n is the number of players, with size of 169 in each direction. A few of these are blank and can never happen. The rest contain the a very large number of values. We know what these values are. We know how to enumerate how each hand does against other hands. We have ICM.

      The equilibrium strategy is a strategy and hence a method that can be employed.

      I do not care what your interests are; you seem to demonstrate a lack of understanding in both topics.

      I look forward to your response in hopes that it will cost you further respect.

      Why do you believe that one cannot employ an equilibrium strategy? If one can, it is obvious that one could also modify this strategy slightly. In this case, it would still be a strategy.
      OK you are really embarrasing yourself here:

      In these push fold stages you must have an idea of your opponents range and thus go for the dominant strategy which may be shoving.

      Why does it not matter how your opponents play? Surely this is the crux of poker! If he has a VPIP of 1% then you shove almost all the time! Now both strategies have been found- then we can find the equilibrium.

      If you are still unsure a quck google will show you that the nash equilibrium is SOLUTION not a method. This is you major mistake.

      Furthermore, you neglect the fact that in poker often there are many hands played (even in your hyper tournaments) so there will be MIXED strategy. Go figure this one out Johnny.

      1) You brought up the point about my interests referring to my blog, so don't get nitty when I answer you.

      2) Strategy =/= equilibrium - Again, here lies you crucial error.

      3) You posted this thread asking for help and understanding, clearly you are unsure and it is evident from your results and posts that you are indeed confused and not playing well.

      If you fail to understand the basic terminology then I suggest in future that you stop using it randomly and making yourself look lost.

      If you still fail to realise you mistakes and correct them then I'm afraid I cannot offer any more help. My condolences. It feels like flogging a dead horse at times.

      Good luck at the tables.
    • 1
    • 2