stats (theory) question

[AlwaysWrong]

So I've been doing pretty well the last month, coming in a decent clip above 3/100. A friend of mine goes off on me about being weak-tight when we discuss a hand (unimportant). I mention that my wtsd and showdowns won % are in accepted 2+2 parameters and he doesn't watch me play, so it's bs to call me weak-tight. A while later I go to review my stats and it turns out my went to showdown has gone down a couple points recently, but my won $ at showdown has gone up a couple points. And I thought I was playing good! Hmm..

One guy posts his stats here and they are:

went to showdown %: 35%
won $ at sd: 55%

Seem about right?

another guy posts:

went to showdown %: 33%
won $ at showdown: 57%

(If this still seems ok to you, then 32/58? 31/59?)

We say he isn't getting to enough showdowns.

But does this make any sense at all?

If so, why?

Edit: was pointed out that you can't trade % 1 for 1, these numbers aren't linked as simply as I made out above, but they are linked.

[danzasmack]

100% of hands exist

35% of time went to showdown. 55% of time won $ at showdown.
19.25% won money at a showdown.

33% went to SD. 57% won $
18.81% won money at a showdown.

Also

you are seeing showdown 5.71% less often. Winning 3.5% more often.

EDIT: didn't really think about that 2nd part.

[AlwaysWrong]

Ok, sort of makes sense.

Let's see if this works.

See flop 20 times.

(1) sd: 0.35*20 = 7 w$asd: 7*0.55 = 3.85

(2) sd: 0.33*20 = 6.6 w$asd: 6.6*0.57 = 3.76

[what would w$asd have to be to get 3.85? 3.85/6.6 = 58.3]

Ok, so there isn't a 1:1 correspondence, makes sense. But there is a correspondence. Say someone posts who has a 60% w$asd. What would their went to showdown % have to be for you not to question their play, if it's possible?

[einbert]

Before analysing these numbers, it would be very helpful to know the sample sizes and limits involved.

[AlwaysWrong]

they don't exist.. neither of these are my stats.. I made them up.

10/20 6-max
50,000 hands

(I'm 34/58)

[einbert]

My inital estimate is that both of you are folding a good deal too much after the flop.

A long term W$@SD of 56% is way too high, IMO.

[AlwaysWrong]

ok, fair enough.

Propose a good w$asd number and a good wtsd number and tell me why a lower wtsd and a higher w$asd isn't better.

[einbert]

[ QUOTE ]
ok, fair enough.

Propose a good w$asd number and a good wtsd number and tell me why a lower wtsd and a higher w$asd isn't better.

[/ QUOTE ]

The reason why a lower wtsd and a higher w$sd isn't better is because you are folding the best hand, or folding when it is profitable to draw too much. For example, I could have a W$SD of 100% if i only played the stone nuts and folded everything else, but I would be losing quite a bit of money very quickly.

[AlwaysWrong]

Yeah, it's pretty clear that in practical terms you can't get too much above 50% without folding the best hand a ton. You just can't have that good of a read on people. But if you could somehow see your opponent's cards you would have a w$asd of around 90% say (you'd bet as a bluff sometimes and get called) and be killing the game like it's never been killed.

But in practical terms there has to be a tradeoff here. Going to showdown 45% of the time is too much, going to showdown 30% is too little. The less you go, the more you should be winning.

Why are whatever numbers you think are right, actually right? How do you convince a skeptic that these are the best numbers?

[AlwaysWrong]

Einbert: give me your numbers, I'll give you some other numbers back, and you tell me why those other numbers wouldn't be better (or tell me that they aren't practically possible).

[waffle]

I made a post a while ago that does a little preliminary investigation into the relationship between W$WSF, WtSD, and W$SD. I don't know how to use this info to better someone's game, but, you can read a little bit more about how they're linked here (http://archiveserver.twoplustwo.com/showflat.php?Cat=&Board=headsup&Number=2301485&For um=All_Forums&Words=&Searchpage=0&Limit=55&Main=22 99641&Search=true&where=bodysub&Name=14123&dateran ge=1&newerval=1&newertype=y&olderval=&oldertype=&b odyprev=#Post2301485) .

[einbert]

[ QUOTE ]
Why are whatever numbers you think are right, actually right? How do you convince a skeptic that these are the best numbers?

[/ QUOTE ]
These are good questions. I'm going to try to come up with some good answers and I'll post again when/if I do.

[AlwaysWrong]

ooh, this is nice. Thanks.

There's clearly a lot going on here. If you have two people with the same vpip you can make a meaningful comparision between their CSD/CWSD, but does it make sense between two different vpips? Also, if two people with the same vpip have the same CSD, the person with the lower WtSD is doing better (??)

[Jeff W]

[ QUOTE ]
If you have two people with the same vpip you can make a meaningful comparision between their CSD/CWSD, but does it make sense between two different vpips?

[/ QUOTE ]

Funny you should mention this. Yesterday, there was a discussion (http://forumserver.twoplustwo.com/showflat.php?Cat=&Board=headsup&Number=3231634&For um=&Words=&Searchpage=0&Limit=25&Main=3231257&Sear ch=true&where=bodysub&Name=11131&daterange=1&newer val=1&newertype=w&olderval=&oldertype=&bodyprev=#P ost3231634) on this topic.