I Think My Biggest Leak Is…

[...] demonstrates the error of thinking the map is the territory in a recent blog post.  The post is about being able to tell the difference between a making a big error and making a [...]

Gary,
Winning poker actually is taking the best possible action in terms of the Expected Value Equation.
If you claim that the conditions of the game are nonstable and nonlinear then this should also be included in the EV equation.
Best wishes,
Pawel

Pawel,

I think that Gary missed the point of the post a bit. He goes to some great lengths to indicate that often it depends. Absolutely true. And for sure, models don’t always hold. Non-linear regions are tough to model and aren’t stable. Really smart people make mistakes modeling difficult things. All true. True, true, true.

To me, the point of Ed’s post was really about when looking for big leaks, they are not in the “it depends” region. That is the region of expert play and good earn vs great earn. With big leaks, it probably doesn’t depend. You are just playing bad. Look for places where you play bad consistently which come up a lot and you’ve found a big leak. Fix it. Win more.

Todd

I had impression that Gary’s article critisizes the approach to poker based on EV, saying that this is a model and not a real teritory.

I too was a bit dismayed by Gary’s article. For example:

“So, it could very well be true that drawing to too many flushes is a leak in your game even though Ed’s analysis wouldn’t uncover it. ”

I don’t think Ed said it wasn’t a leak, he just said it isn’t likely to be a major one for the low-limit players he was addressing.

And also:

“But all their analysis was done with graph paper and slide rules. They never fooled with the dirt.”

This implies that Ed is some sort of ivory-tower mathematician who doesn’t “get his hands dirty”. This does Ed a disservice. The main reason Sklansky and Malmuth brought Ed in as an author was precisely because he had gotten his hands plenty “dirty” at the lower levels of the game. The preface to SSHE says so.

I think Gary is the one who is being too ivory-tower here, by taking Ed’s article too literally. Ed isn’t trying to build precise mathematical models here. He’s just saying “Look for the errors you make most often, and fix those first”.

I’m an engineer too. How sad for me– doomed to see the nonlinear world through linear eyes. When my functions are nonlinear, I’m lost. Oh wait, I might be able to linearize them! Whew, that was close! For a second there I thought the universe was collapsing.

It’s most efficient to use the simplest possible model. Add complexity only when necessary. It’s one thing to say a model is inaccurate. It’s another to demonstrate that it is inadequate. It’s yet another to develop a better one. Gary doesn’t seem to have gotten past step one.

As far as poker is concerned, most of us have far to go to extract what benefit we can from the simplest concepts. We can worry about extensions to current theory when we’re world-class.

So anyway, I do have a question: what [i]are[/i] some examples of big leaks for mediocre players like me?

[...] A lot of defensiveness among the readership over at Ed’s blog about the post I made. True believers [...]

How typical. Now we are all just a bunch of naive and ignorant true believers.

One of the things I have always liked about this web site is that Ed never feels the need to put other people down in order to lift himself up. Too bad so many other people obviously aren’t the same.

What I’m starting to think is, this guy is just trying to increase the traffic to his site by trolling us. I suggest that we stop playing along.

Hi Gary,

Since you won’t comment here, I won’t comment on your blog either.

You say

“I wasn’t talking about how complex you need to make your models. I was talking about the mistake of treating a model as if it’s the actual reality. If they didn’t teach you that in school then you should have gone to a better school.”

I’m still in school and am happy to continue learning, even if it has to be from you. I’ve read your book, I thought it was worthwhile.

It is a mistake to think the model is identical with reality. (Happily, my education hasn’t failed me so far as that is concerned.) However, it is exactly the point of a model that we can treat it as though it were, so long as we do not violate the underlying assumptions and understand its limitations. There should always be a sanity check to verify that the results are consistent with experience. Have these conditions been violated? Maybe, but you haven’t demonstrated it.

I understand you’re an iconoclast and enjoy controversy. It’s your thing. I’m honored to be added to the group of people you’ve insulted over the years. My mom will probably get a good giggle out of it, thanks a bunch.

You’re insane

I’ve never deleted any post critical of me or anybody else. Mason has a long history of deleting my posts even in violation of specific agreements I’d made with his webmaster.

Ed Miller has always supported Mason and that kind of behavior. There’s no reason at all to trust that Ed Miller won’t delete anything I post here.

Lay down with dogs, etc, etc.

Hi Gary,

I would comment on your blog. Must log in, can’t, sigh. So I stuck here in this thread. Sorry Ed.

You really have a bee in your bonnet about EV. Did you have a traumatic experience with a renegade craps table? Infinite improbability drive let you down? My goodness. I think you make some nice points that are diminished by your mean spirited approach. EV. +EV. I think you are also trying to espouse quantum mechanics to an audience that isn’t quite ready to operate the linear accelerator, if you know what I mean. -EV. Measure. Martingale. The audience for that post are well served by Sir Isaac.

I think Ed’s old post was aimed at these people:

…snip…
If you play mostly limit and it’s not real loose but it’s very passive then looking at things with a prism of EV as the Word of God will help a lot, it just won’t help that much when things change. And, I promise. Things will change.
…snip…

Gaping holes in their game. House is on fire, not time to fix the squeaky garage door. Density function. Numeraire.

You are correct that things will change. When they do, Caro’s advice will serve them (read me) well.

Sorry you had a bad experience with Mr. Malmouth. I doubt your the only one. But, still, more flies with honey, etc, etc.

“You’re insane

I’ve never deleted any post critical of me or anybody else. Mason has a long history of deleting my posts even in violation of specific agreements I’d made with his webmaster.”

Since you posted here, I went over to your site, but apparently you don’t have comments enabled. My intent was not to defend my sanity or question yours, but just to return the ‘favor’. You do have some interesting resources there. Worth checking out.

Not everything has to do with your history with Mason. If Ed deletes something that’s his choice and his right. If you want drama, wouldn’t you be better off going after 2+2 more directly?

“EV is important, no question about it. And if you define Expected Value Equation (I like those caps, gotta make sure we indicate the holy status) as some sort of tautology where it’s conditional on whatever unspecified conditions are required then it would be true. But if you’re going to pray to a conditional EV you need to specify those conditions.”

Actually the conditions are always specified. Conditions of a certain game are given precisely each time EV for any play is considered. They are of course rough estimates and that’s what makes them a model rather then reality.

“Making each decision independently of all other deicsions and doing so to maximize the independent EV of that decision will not give you an overall EV maximum.”

In a poker game you never make any decision independantly, but every decision you make has its independant EV. So you have to take into consideration many circumstances to make a decision whith maximum EV.
The example following your reasoning is exactly what I expected to find there.
“Maybe if I don’t call I’ll get a chance to bust him later”
Maybe if call this time you do bust him and then get another chance again later, after he reloads? But thus sometimes you will have to reload… unwilingness to do so is what is probably called being ‘weak-tight’.

It’s ironic that in his attempt to show up Ed, Mr. Carson has illustrated what, at its heart and beyond the poker table, this article is all about. I was close to expounding on this in detail. Gary is very temptimg, but I, unlike he, get it.

Ed,

Very interesting post on the relative size of errors. I’m very impressed with your site and your analysis of most topics, but I have to take exception with one of your points when you posted on this subject back in 2004.

You stated then in regard to giving too much preflop action that “(Now, making that mistake might get you into a situation where you make more mistakes and lose more… but the ORIGINAL mistake costs at most $2. Remember, you can always fold.)” While it’s true a player can always fold and cut his/her losses, there is a paradox that it’s often incorrect to fold because of pot odds (or implied odds) once that player is in the hand.

Take, for example, one of the spots you discussed where you have 72o in the SB. Calling against 2 loose limpers (assuming an equity of 14%), will result in a small +EV all-in, but because of position and playablity issues is considered a small mistake. Say the flop gives you a small pair and you face one bet. It’s quite possible to figure your hand has enough equity to call on the flop, and then reevaluating on the turn and river to come to the same conclusion. In each decision, you are making a small +EV play, but your NET EV can be seriously negative because a lot of the money that you’re chasing on later streets was YOUR OWN money put in on earlier streets. Of course, once you put it in the pot it’s not YOURS anymore and it inflates your pot odds.

I call this kind of early mistake a “compound mistake” where even CORRECT play on later streets compounds your loss. Make no mistake about it… this is a common and major leak, and is even worse in Stud and Razz where you have 3 big bet streets to play through.

I think Gary and Ed both have a lot to teach the novice player. I’ve found that Ed’s approach works extremely well in loose and passive games, but not so well in wilder games, and that’s what it seems to me Gary is saying. Commenters such as Pawel seem to be ignoring how difficult it is actually to know what the EV of a play is without enough information. For instance, on a paired flop, the EV of drawing to a flush is reasonable if no one has flopped the boat or quads and terrible if they have. In the model Ed uses, you might factor that in to your decision, which will tend to make you play too many losing hands and too few winning hands; in the model Gary uses, you’re looking to fold when you’re beaten and improve your results by *not* applying one decision to the whole set of decisions that look superficially the same.

Doctor Razz, I am having trouble understanding your “compound mistake” explanation.

If I have 2 BB invested before the turn, then I fold on the turn, I am sure to lose 2BB.

Now if I am getting pot odds to call on the turn, then by definition the call is +ev and should reduce my 2BB loss to something less. The fact that it is partially “my money” in the pot shouldn’t matter.

Or to look at it another way, if I fold I definitely lose my money in the pot, but if I call then at least there is some chance to win it back plus more.

So how can a +ev call increase your loss?

What am I missing?

I just re-read your post Dr R. and now I think I get it - you said “NET EV”. Do you mean your net for that particular hand? Yes I see how you could end up losing a lot more than just 2BB for that hand, if you call a lot on the turn and river. If each call is +ev it will decrease your loss in the long run, but certainly not on that particular hand. If you are in a tournament and low on chips then I can see where you might not want to make those calls.

JJS -

Thanks for trying to follow what I’m saying. When I talk about EV, I mean it in the standard sense of long run average value over an infinite number of hands. When I say “Net EV” I’m talking about the EV for the hand as a whole, instead of the EV on any given street. The paradox is that on any given street, the pot you use to calculate your EV has been bloated by your own money (in part) such that you can make correct plays that end up costing you a lot of money in the long run.

To show an oversimplified example, say you have 20% equity in the pot on every street, including the showdown. Preflop there is one limper and you’re in the SB, getting 5:1 to call. Of course, you’re out of position, so a call here is likely a small mistake but you go for it anyway. On the flop, you check, the BB bets, and limper calls. You’re getting 5:1 again… worth a call. On the turn, the action is the same and you get 5:1 odds again. Gotta see the river, right? This time the BB bets, limper folds, and you call getting 7:1 odds. A great price with your 20% equity, or is it?

Your line for the whole hand is not so good. You put in a total of 5.5 small bets into a pot of 16 small bets. That’s 34% of the chips, but you only had exactly 20% equity the entire hand. Your Net EV in this example is -2.25 small bets, a lot more than the initial 0.5 small bets you put in preflop. How could you lose so much if you made one small mistake and played the other streets correctly? Like interest compounding on your credit card bill, that little mistake kept growing because you kept chasing your OWN money once you put it in the pot.

That’s a “compound mistake”

“The paradox is that on any given street, the pot you use to calculate your EV has been bloated by your own money (in part) such that you can make correct plays that end up costing you a lot of money in the long run.”

I think that you bring up a good example, however the mistakes that occurred were due to a misunderstanding of what “20% equity” actually means at each street. The river call is the only one that is clearly not a mistake, and that is because there are no future streets to bet on. The 20% previously would only be true if you were going all-in on that particular street. And that’s why Pokerstove calculations are so often mis-applied to limit situations when really they’re more suited to no-limit all-in plays.

I agree with nfactor13, I think the key to this “paradox” is implied odds.

In Dr. R’s simplified example the pot size is 5 SB on the flop and it’s one SB to you. 5:1 pot odds seems to imply a call. However this is a small pot and its likely to get bigger, so you have to think about future betting and implied odds.

We are only getting 2:1 on our money on all future bets (there are only 2 other players in the hand) so implied odds say that we should have at least 33% equity to continue. If you only have 20% equity then clearly the flop call is -EV and should not be made.

Pot odds is an important poker concept but it’s not whole game, by far.

Dr Zen,
“on a paired flop, the EV of drawing to a flush is reasonable if no one has flopped the boat or quads and terrible if they have”
Because you can never know for 100% if someone has a boat, or if someone is drawing to a bigger flush along with you, you need to make a play that has the biggest EV on average, and because game conditions change, you have to make assumptions. Reassuming, you try to make a biggest EV play on average under given circumstances.
How certain one can be to be against a boat or a quads is on the circumstances part, not on the strategy part.
I think there’s a good example worth checking in ‘The Mathematics Of Poker’, pages 4-5.
Best regards,
Pawel

“I agree with nfactor13, I think the key to this “paradox” is implied odds.”

That is exactly the key. That’s why we need people like Ed to guide us thru post flop play.

Pawel, yes, I understand the theory, but my understanding is that Gary believes Ed underestimates the importance of game conditions and applies the model too strictly (in the terms you used, he ignores the specific circumstances: just like a soil scientist applying a model of soil and not giving any mind to the soil he actually has). Separating “circumstances” from “strategy” is nonsense though. That’s exactly what Gary is berating here.

Gary,

Question. And since tone is hard to detect on-line, I’m not saying this in a smartassy way, but in the vain of self-examanation of your own thought processes.

You made a public statement here that Ed can’t be trusted not to delete because of laying down with dogs etc. etc. Is that making a linear assumption about non-linear human behavior? Other factors involved besides he was part of 2+2, ie. he doesn’t post there anymore, he started his own strategy discussion site, he never personally deleted any of your posts, etc. Would these factors affect the probability of Ed deleting your posts on his site?

His history gives strong evidence that he can’t be trusted. He has behaved very badly in the past for no reason other than to try to please Mason.

Maybe he’s grown up. I don’t know. He says he has. I see no need to assume he has. I don’t see a benefit from taking such a path.

It’s not his association with 2+2 publishing that’s the problem. It’s his behavior and things he’s said. John Feeney, for example, is someone I’d trust. I even trust David, although to a limited extent.

I haven’t posted on 2+2 since Ed was about 12, so of course he didn’t personally delete any of my posts on 2+2.

It’s not about probability. It’s about personal trust. I don’t trust him and have never seen him do anything to suggest I should. That’s just as true today as it was a year ago.

He’s had opportunity to show he has changed his stripes and has not taken the opportunity. He says he has changed, but he’s not done anything to show it.

btw, keeping a focus on the model rather than the thing itself works fine for soils engineers or electrical engineers. Not so well for industrial engineers (who work with man/machine systems).

[...] the comments section of an Ed Miller post I mentioned earlier Pawel quotes Dr Zen Dr Zen, “on a paired flop, the EV of drawing to a flush is reasonable if no [...]

Gary,
“It’s possible that those probabilities are such that calling isn’t higher EV than folding. But raising might well be even if calling isn’t. Check-raising might be better than betting. Calling and back raising might be better.”
I generaly completely agree with you here.
I’m not sure however if we generally understand ourselves right.
I don’t take the EV as an ultimate guide whether to play a hand or not. I only agree (with what I have read about poker so far) with the definition of a strategy as the aim of maximizing ones overall winnings. So I would say that ‘if this then do that’ is a way of executing the goal.
Also, I think that one can estimate EV of those different actions one may take, like for example folding equity etc. Still I mean in the given situation.
From what you wrote in the last article it looks like you suggest using some other model for estimating what action is best.
Of course I don’t know any better yet then the approach I’m trying to describe, yet I don’t mind learning, and I would necessarily claim that one must be better from the other.
However, I quite don’t agree with this statement: “In a wild and loose game you’re more likely to be beat or drawing dead.”
I think it’s more likely that your opponents will rather be chasing gutshots and rising with trash hands and doing some crazy bluffs and doing some absurd slowplaying then betting the nuts. At least that is my experience so far and also it looks reasonable if we are talking about games where people play weak hands maniacally.
Anyway, some good discussion seems to arise.
All best
Pawel

P.S.
“and I would necessarily claim that one must be better from the other”
Sorry - of course I meant I would not necessarily claim that one must be better from the other
Pawel