Day 13 - Swongs Are One Thing
Even until yesterday afternoon, Brian was almost in despair. The day before he started to tilt from "running bad." The mind is weird and it is easy to tilt from small losses. I told him even if you lose every sng you play that day, you still aren't truly running bad. We had checked HEM and saw the results: negative 1.6 million chips in all in EV. I laughed when I saw how insignificant his sharkscope was: -$800. Brian was running super bad, but because he was really good at the game, it was impossible to lose as heavy as one should when they run that bad. We were still printing money, even when we weren't.
I told him about standard deviation and the # of games. See, in regular in MTTs, or in cash games, or in HU hyperturbos, or whatever, sometimes you actually just can't play out of a bad swing. It would take you a month or two months or more to gain back all those sklansky bucks that you lost. So you can sometimes feel hopeless playing say, 5/10 NL when you are stuck 40k. But in 180s you could. You could get out of anything, and thats why they were so great. I told him, 200 games a day, $2 a game conservatively. So you are only truly 2 days behind. Peanuts. And like Knish said in the opening scene of Rounders to Mike McD: "You'll be back before you know it." I told him I lost $35k in one day in my career more than 3 times. And each time I remembered that phrase.
And I knew he was good at the game. I was so firm with that conclusion. Because after going over hundreds and hundreds of HHs one hand at a time, I could make that deduction. His error rate started out high but slowly got better. And his tabling ability got better, from 24 to 26 to 28 to 30 tables at a time. I find it funny that Brian found it amazing that I was talking to him on the phone and talking about strategy at a final table that I was in while 30 tabling, but didn't recognize that he could 15 table and talk on the phone and now he was only a couple months away from what seems impossible today. And then it happened, that same day. 4 final tables at the same time. He won two $15/180s and a $3r/180 in that same afternoon, and booked a downswing-saving session. And after that big win, his lifetime stats? He was finally even.
And Brian impressed me every day that he played, it made me so happy as a coach. Today we decided to review hands other people posted in the MTTSNG and Midstakes MTT forum. And we saw the following hand:
Poker Stars, $7.34 Buy-in (2,000/4,000 blinds, 400 ante) No Limit Hold'em Tournament, 6 Players
Payouts: 300,200,119,80,65,50
SB: 97,947 (24.5 bb)
BB: 41,462 (10.4 bb)
UTG: 33,671 (8.4 bb)
MP: 7,512 (1.9 bb)
CO: 52,842 (13.2 bb)
Hero (BTN): 36,566 (9.1 bb)
Preflop: (2,400) Hero is BTN with 8
K
3 folds,
Hero ?
And he got it right!!
Fold. And I was so happy because every other regular in that thread got it wrong, and it wasn't close. You could put any reasonable calling ranges you want in a computer and it would tell you to fold. Actually it was really sick, because it was like, people didn't trust a computer. And that is what I learned early on when talking about 180s and trying to teach them. My general impression was basically this: People didn't like the idea of a solved game [like 10bb poker] being all about playing accurate technical ranges. Rather, they liked the idea that they were poker
players, in the hotseat and able to make
optimal decisions because they were so active in
exploiting other players, and who cares what K8o's equity is compared to 87s versus their calling range, I have a King and a medium card which is an above average hand and I am on the button with 9bb god dammit, and ICM and a computer be damned, because I am
entitled to win more often than these fish and it is because of my supernatural ability to apply my
skill in this skill game and put my tournament in my own hands.
(This is also why many tournament players today still, in the early levels of an MTT where cEV is very close to $EV, they push very wide in early position with a few big blinds "to protect their fold equity". It's because they hate to be in the big blind and to be forced to call off with a trashy hand. But Brian and I didn't mind. In our eyes, there was only equity and making the correct decisions, and if folding K8o utg now only to call with 43s later was the best play, idiotic as it may seem, we trusted the facts.)
I thought about this on my own. Why are people so averse and so willing to discount what a computer says? I felt it was analogous to a moderate chess player rated 1550 reviewing a game of chess he played where he played a move that was Ng4: -1.00 instead of a computer suggested move of a5: -0.21 and saying "well, my knight is further up the board and is way more active, while a5 isn't very active at all, so clearly Ng5 is better than a5 and who cares what a computer says." It shows a ton of ego to mistrust a computer over your experience, or maybe just a lack of knowledge about how irrelevant your experience is. And then it dawned on me. It was just a lack of understanding.
This is just an aside, but here we have a computer (such as holdemresources beta) that is willing to take whatever ranges you input, and make a relatively simple calculation for you. And it spits out the answer and what the EV difference of the answer is. When Brian practiced these pushfolds, he trusted a computer when appropriate, as did I. And I feel he is in just a much better position to learn and grow than most regulars. Let me try to break it down in more detail.
Poker at its core is about two different things. The first thing is "what is your opponents strategy?" The second thing is "what is the best response to that strategy?" Sometimes you will plug in something into a computer and it will make little sense. You should try to understand it. For example, we looked at a hand where our opponent checked back a flop and then the turn was an A. We guessed at our opponents range and inputted our range and we found first by counting all our combos and all our opponents combos, that AQ was only in the 72th or so percentile compared to our opponents range, and that by computer analysis that checkcalling was better than leading when we had AQ based on our guess of the opponents range and play of our opponent.
Anyways, when you reject the conclusion of what the computer says, for example in the case of the K8o hand, your rejection has to come from the premises -- the computer is just applying logical deductions to those premises to achieve it's conclusion. In particular, in the K8o hand, by rejecting the conclusion, the only logical rebuttal one can give is to be challenging the interpretation of the ICM formula for equity given payouts and stacksizes. --
-- Basically, if you push, most of the time you get a fold and the stacksizes are one way. If you push and get called, some of the time you win and the stacksizes are another way; then most of the time you lose and the stacksizes are yet another way. Because obviously it knows that for any given ranges, it knows what % of the time you reach each possible end state that one could reach given everyone decides to push or fold. That's just a simple implication of the ranges and the assumptions.
It's nice to think that our equity, proportional to the equity that ICM predicts we have, increases substantially as we go from say 8bb to 16bb. But it doesn't.
The first paper written in the neighborhood of the subject was written in 2008 and I read it when it came out. It analyzed 9max SNGs that were ITM, and basically solved the game completely - without ICM and understanding all future game considerations. That gave a very very strong argument that ICM, aside from corner cases, was very, very accurate.
It's been proven in psychology that people act first and then surround their beliefs to be congruent with their actions. And I am not immune, as these last few paragraphs demonstrate. But it takes a big person to admit misunderstanding. Instead of just looking at the K8o hand and admitting they don't understand, instead people try to invent reasons why they are right. In fact it is not even close either: Even if the big stack gave the short stack 20k chips before the hand started, the computer still wouldn't even push K9o !!!. But, if I had suggested to fold K9o in the thread, the participants might have gone crazy. However, I think there were a lot of silent readers who read my advice, made a note and learned from it, and went on without saying a word -- which is more than half the reason I contribute. And the reason that K8o is a fold is because of the equity versus the calling range. People just have a misunderstanding about hands. They think hands are ranked in a linear order: start with all the pairs. Then all the aces. Then the kings, then the queens, then the jacks, etc. Because after all, a pair is better than ace high is better than king high etc, right? So they have this notion that K8o is pretty high up there.
But it isn't! This is pretty instructive: list all the hands in terms of equity versus a "calling" 14% of = {55+ A7s+ A8o+ KJs+ KQo}. That is, you go all in and he called with this range. Now you list all 169 possible hands in order of preference. The answer is listed here, and if you are shocked by this answer then maybe you would learn something and reconsider your strategy about how looks can be deceiving when it comes to how good K8o is for this situation.
Anyways, the last part of this update got a bit blog-ey. But Brian is starting to do well, and I look forward to his progress as he starts to learn more about MTTs and analyze more hands in these MTTs and try to play deeper. I find it interesting because of the pushfold work he has done, he always gives accurate ranges (not just vague guesses, but right on the ball exact listings of hands like "55+ A7s+ A8o+ KJs+ KQo" for example. But when it comes to ranges in deeper play, he starts to stutter because he doesn't have the practice of thinking about ranges constantly in more complicated spots. I hope this improves.
One last thing: if you want your interesting/tough MTT hands reviewed please post in midstakes MTT forums and email me at
alexwice@gmail.com and I will review them in the thread and have Brian look at them for practice. Though I find that most people focus too much on spots that come up once a month and have no relevance to roi, instead of spots that come up frequently. But if Phil Ivey always folded a royal flush he would still be the best player in the world.
All the best, and I look forward to updating everyone next week, and thanks for following me on twitter (@AWice).
AW