PDA

View Full Version : quantify luck for me please

shooter
10-10-2004, 11:32 PM
from a non-math guy;
I am told there is nothing that cannot be measured, so if AA (or any holdem starting hand)can be measured to win x times and lose y times, is there a way to follow one player through a whole tournament and calculate the number of times that player won by drawing out on a superior hand, then perhaps subtracting the number of bad beats sustained.
Or perhaps one sample hand played many time to obtain a sample, which could then be translated statisticly?
I have seen many(self included) tournament winners who could not have won without this factor applied many times. However I have never heard a number attatched to it. My suspicion is that the percentage of lucky wins, hand to hand or long term is quite high.
How would you go about calculating the luck factor?

Brineboitano
10-11-2004, 12:13 AM
There is exactly one way: guess. Probabilities of the outcomes of hand matches are based on the number of combinations of cards that will make one hand win over another. The problem is they don't account for the fact that players fold their hands durring the course of play. About all you can do is guess how often a player will fold when he shouldn't, how often they will stay in when they should fold, etc. About the only way to quantify how much better one player is than another player is to have them play eachother a lot and see who wins the most often, but you're seldom going to get that much information.

In order to do some kind of statistical analysis and figure out who's got the best/worst odds of winning, you would need to do a LOT of trial runs between the players involved. Without that, about all you can really do is take a guess.

shooter
10-11-2004, 02:08 PM
Thanks for the response. You are right about it needing to be numbers based, so I am setting up the following test on Wilson software:
one player will be frozen with JsTs. four players will play premium hands (not including J,T), four will play lesser unpaired hands. Personalities and play styles will be representative on both the premium and subordinate sets. Our test subject will be tight/aggressive.
This trial will be contrasted to the above by our subject also holding AA for one trial and two random cards for another. I would suppose one million hands should wash any anomolies.
Now the question is have I forgotten anything?

kleraudio
10-11-2004, 03:13 PM
hello,
please reply back w/ the results i am very interested in this test.

shooter
10-11-2004, 09:34 PM
OK. After hours of simulations I settled on using;
seats 1-4 = loose, aggressive, tight and average w/ premium hands over the base line Js,Ts without using a J or T,
seats 5-8 = same player profiles w/ hands inferior and no pairs,
seat 9 = tight aggressive holding exactly JsTs,
minimum of two opponents on the river, 100,000 hands- Then test 2; same parameters except seat 9 has AA and where possible seats 1-4 kept their hand (except for the two aces); Test 3, same as before except seat 9 has any two random cards- deal

Test 1 J/10 win 33.1% lose 65.2 (remainder are split)
Test 2 A/A win 56.4 lose 41.8
Test 3 random win 42.9 lose 53.2

My surprise was seat 9 had out won seats 1-4 by a huge margin with all three hands. ??? I gotta go back and study that profile hard.
Next I found a profile who was a calling station, no betting or raising, only calling to the river...I know you never met anyone like that.

With that player in seat 9 I did it all over again:

Test 1 J/10 win 23.5 lose 75.3
Test 2 A/A win 60.1 lose 38.5
Test 3 random win 25.5 lose 72.7

Again surprise, seat 9 outplayed 1-4 by a lot, not as much as tight/aggressive. A lone standout was one of the premium players did much better against the calling station.

I will have to chew on this for a while before something comes of it (after someone turns off that buzzing sound).
Right now it looks as though with four players holding high cards, they simply ate eachother....
Still don't know how to assign any of this to luck..
sorry