Two Plus Two Older Archives  

Go Back   Two Plus Two Older Archives > General Poker Discussion > Poker Theory

Reply
 
Thread Tools Display Modes
  #1  
Old 12-19-2005, 12:37 PM
Guest
 
Posts: n/a
Default Theory of Deadweight Information Loss

Theory of Deadweight Information Loss --

Purpose: To distinguish one line of wagering as superior to a parallel line when the end result between the two (total amounts wagered by all parties, the pot size) is the same.

The following must be in effect for two wagering lines to be considered parallel:
I - All action for both lines must take place within one betting round (one street)
II - All action for both lines must result in the exact same total wagers and total pot size at the end of the betting round.

The Theory of Deadweight Information Loss:
Given two parallel wagering lines, the superior line is the one that reveals the most information about the opponents' hand prior to him revealing his cards.

In this fashion, a player gains two benefits:

A - The player gains more information in order to make the proper decision for this hand.

B - The player gains more information on how an opponent plays his cards, often just before those cards are revealed. This information is valuable for other hands in the future.

In many cases benefit 'A' is not enough to dissuade you from making the incorrect decision; but it has provided information none-the-less.

--------------------------------------------------
For instance, consider the following NLHE example:
--------------------------------------------------

Hero -- Holding AA from the Button with 40xBB
Villain -- Holding KJ from the SB with 25xBB

Hero - AA raises 3xBB…
…Villain - KJ calls from SB.

flop K J 4 (7 BB pot)

Villain leads out with a 3BB bet…
…Action to Hero is 3BB to call 10BB

Two scenarios:
------------------
1) Hero raises all-in ; Villain calls.
2) Hero raises to 12xBB ; Villain pushes all in ; Hero calls

Even though in the second scenario the Hero was provided extra information about the strength of the Villains hand, the pot odds and the strength of his cards necessitated a call.

Consider that when a player is making an incorrect guess as to his opponent’s strength, (as in the above case) he has an 'Evidence Threshold' (ET) that is always 1.00. When that level is breeched, he realizes his mistake because of the overwhelming evidence at hand, and changes his judgment. In this case, Scenario 1 represented a ET impact of .02 (only 2% evidence out of 100% needed to persuade Hero that he is beaten) Scenario 2 represented an ET impact of .22 at the second decision point ( An extra 20%, but Hero still needs 100% to lay it down, so he calls)

So we made an extra 20%, but then pissed it away. Well it didn’t just disappear.

Similar to a dead-weight loss of mutual benefits in economics, there are betting lines that result in a dead-weight loss of information – Accumulation against the ET that may be useless to the short-term outcome, but is gleaned for future reference.

Additionally, good players accumulate many small bits of information by obeying the Theory of Deadweight Information Loss in choosing their betting patterns, and in this manner accumulate more knowledge about their opponents which results in a greater likelihood of a 'threshold breech.'

A player who avoids deadweight loss of information in his play will become sensitive more quickly to an opponent, and will better be able to judge his later actions.

All of this is just a restatement of a lot of obvious stuff, but I think there is something to be said for information gathering in every nook and cranny you can find… including making sure your betting is getting the maximum information out of your opponent for use in this and later hands..

I have a few more scenarios to post later on, involving weaker hands and how accumulation of small amounts of information that might otherwise have been missed can lead to laying down hands to a trap that you sniffed coming, and lead to pushing a marginal edge to cause your opponent to lay off a hand. Both are things we do every day, but it is always a keen idea to explore those habits and better understand them.

I appreciate your thoughts;

Keith
Reply With Quote
  #2  
Old 12-19-2005, 03:21 PM
AaronBrown AaronBrown is offline
Senior Member
 
Join Date: May 2005
Location: New York
Posts: 505
Default Re: Theory of Deadweight Information Loss

Most poker theory takes a different approach. The information in B is not necessarily valuable. A good player will keep track of the information she has revealed, and act in the future to frustrate people trying to take advantage of it. Granted most players have some degree of predictability, so it does make sense to assume they are likely to act as they have in the past. However the value of this kind of information in poker is much less than in other games, and the better the players, the less valuable it is.

The usual poker advice is if you are willing to call a raise (including an all-in raise) you should make it yourself. The main exception is when you are slow-playing, you also need to work in exceptions for deception. But in general, it takes a stronger hand to call a raise than to make a raise, and to bet less than the maximum amount you are willing to bet gives options to the other player, and thus is passive.

As you can see, this sacrifices information about the other player's hands in the interest of limiting their options. It is closer to minimizing the information about their strategy than maximizing it. Gathering information is passive, forcing issues is aggressive, and conventional wisdom is that aggressive is better.

Therefore, this is a radical way of looking at the game. That doesn't make it right or wrong, but most people's first reaction will be to argue with you.

I don't like your specific example because calling an all-in raise with only an overpair is suicidal. Villain is representing a set, which has a 90% chance of winning. Even if he is loose enough to call 3 BB preflop with KJ and go all-in with two pair, he has a 73% chance of beating you. This ignores any flush possibilities. You are only likely to beat a pure bluff.
Reply With Quote
  #3  
Old 12-19-2005, 04:43 PM
Guest
 
Posts: n/a
Default Re: Theory of Deadweight Information Loss

[ QUOTE ]
The information in B is not necessarily valuable. A good player will keep track of the information she has revealed, and act in the future to frustrate people trying to take advantage of it.

[/ QUOTE ]

I would content that this is a constant concern in poker, and any method of information gathering merely results in a better chance of anticipating your opponents actions. (You can better read when and why he is changing his plays up.

[ QUOTE ]
Granted most players have some degree of predictability, so it does make sense to assume they are likely to act as they have in the past.

[/ QUOTE ]

Again, there is no implicit assumption that says that along with the absorption of information on your opponent you must assume that the opponent will act in this exact manner again. Instead you can use the information to better gauge what level your opponent is thinking at.

[ QUOTE ]

The usual poker advice is if you are willing to call a raise (including an all-in raise) you should make it yourself. The main exception is when you are slow-playing,


[/ QUOTE ]
or value betting.

[ QUOTE ]

But in general, it takes a stronger hand to call a raise than to make a raise, and to bet less than the maximum amount you are willing to bet gives options to the other player, and thus is passive.


[/ QUOTE ]

The Gap Concept. TOP. but I am confused when you say that betting less than the max you are willing to bet gives them options. Again I must say that value betting should be an important part of your arsenal. The fundamental TOP says that you want to induce your opponent to make mistakes, and putting too much money in for their draws is the best kind of mistake you can induce... half the pot is a sufficient amount. Don't need to jack it always...

Harrington says in HOH that if you find yourself pushing at pots to protect a lead, you need to re-evaluate if this is the most +EV thing to do. You can often get someone to chase.

Thanks for taking the time to comment. I apprecite you taking the time to offer comments and criticisms....

In closing, I would just say that information gathering in NLHE, especially tournaments -- is not passive in ANY WAY... the way you get information is by betting into someone. When you check to them you do not know what their betting represents. This is a fundamental resonance that fills TOP and HOH.. BET BET BET for information... I just cannot be dissuaded from this belief, it is a central tenant to how I play poker.
Reply With Quote
  #4  
Old 12-20-2005, 12:14 PM
elmitchbo elmitchbo is offline
Senior Member
 
Join Date: Jan 2005
Posts: 129
Default Re: Theory of Deadweight Information Loss

quite frankly, i agree with both of you to some extent.

i think it is valuable to gain information at every available oppurtunity. i also think that sacrificing aggression for the simple purpose of gaining information is, more often than not, an incorrect play. you don't want to BET BET BET for information, you want to BET BET BET because it gives you the best chance to win pots. any information that is unveiled in the process is a happy byproduct, but it should never be your primary concern in the way you play a hand.

i would like to see you're other examples before i make more commments. i think you might be on to something, but your example isn't the greatest.(in quoting HoH you say pushing to protect leads isn't always the most EV play, but in your example you're way behind)

suggesting this type of play is more valuable in a tournament is crazy. you may gain lots of information from a hand that gets you busted out by a player you will never see again. information gathering is more valuable in ring games, because you are more likely to get a second chance to put it to use.

let's see some more hands that back up your ideas.

good post... got me thinking!
Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -4. The time now is 10:27 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.