Try these PT filters

[Aaron W.]

This came out of a recent discussion I had with JJack. I have am suspicious of a few things, and would like to get a dataset that included the playing styles of people other than me. I would hope that you have at least 10k hands in your database, but I'll take whatever is out there from whoever is willing to do it.

The baseline filter:

* Hands with exactly 3 players who see the flop
* No preflop raise
* Money in or saw flop for free

Record the following information for each hand:
* # of hands
* BB/hand
* VPIP

Filter #1: Big blind
Filter #2: Big blind, steal attempted against your blind and called/raised

Here are my values:

Filter #1: 749 hands @ -0.11 BB/hand, 23.90 VPIP
Filter #2: 66 hands @ -0.49 BB/hand, 100 VPIP

Then I want you to compute the actual wins/losses in terms of BB and number of hands VPIP by straight multiplication:

Computation #1: -82.39 BB with 179 VPIP hands
Computation #2: -32.34 BB with 66 VPIP hands

From each pair of filters, it is now possible to get primarily hands that are limped pots -- this method does not get rid of raises from early or middle position.

Computation #1 - Computation #2: -50.05 BB in 113 VPIP hands

We also need to reduce the total number of hands to account for removing the blind defense hands:

Big blind: 749 - 66 = 683

We are also going to add back the money lost due to the blinds because I like positive numbers:

Big blind: -50.05 BB + (.5 BB/hand)*(683 hands) = 291.45 BB

Finally, reduce the number to BB/hand and compute adjusted VPIP:

Big blind: 291.45 BB/683 hands = 0.427 BB/hand @ 113/683 = 16.54 VPIP

So... how do your stats look?

[OziBattler]

I hope I got the calculations right cos damn, I run good. Aaron, FWIW I probably play in juicer games than you and I table/seat select good. Note that Filter2 has a very small sample which is probably standard for passive and high VPIP Tables.

Database size was 47,200 10 seat table hands from 0.1/0.2 to 1/2. Mostly 25c/50c and 50c/1

Base Filter
1327 hands filtered
0.06BB/hand

Filter1 :924 hands @ 0.05BB/100, 24.13VPIP
Filter2: 37 hands @ -0.21BB/100, 100VPIP LDO

Computation1: +46.2BB with 222 VPIP hands
Computation2: -7.77BB with 37 VPIP hands

Computation1 - Computation2: 46.2 -(-7.77) = 53.9BB @ 185 hands

Big Blind = 924 - 37 = 887

Big Blind: 53.9BB + (0.5BB/hand)*887= 497.4BB

Big Blind: 497.4BB/887 = 0.56BB/hand @ 185/887 = 20.8VPIP

[Aaron W.]

[ QUOTE ]
I hope I got the calculations right cos damn, I run good. Aaron, FWIW I probably play in juicer games than you and I table/seat select good. Note that Filter2 has a very small sample which is probably standard for passive and high VPIP Tables.

Database size was 47,200 10 seat table hands from 0.1/0.2 to 1/2. Mostly 25c/50c and 50c/1

[/ QUOTE ]

That looks really good. Turning a true profit in those situations... I wish I was doing that.

I guess I should have noted that my stats are from about 20,000 6-max hands, mostly at 1/2, with some .50/1, 2/4, and 3/6 mixed in.

[kerowo]

I can't figure out what you are doing to get the final number, but here is my stuff up then. This is mostly at .25/.50, around 35K hands.

.25/.50 9k hands
bb/100 (0.06)
vpip 11.66

Filter 1 in BB
1,125 hands
bb/100 (0.27)
vpip 23.29

Filter 2 steal defense
45 hands (???)
bb/100 (0.05)
vpip (100)

Comp 1 = 1125 * 23.29 = 262 vpip hands * (.27) = -70.74BB
Comp 2 = 45 * (.05) = -2.25BB

comp 1 - comp 2 = 217 hands at -68.49 BB

BB = 1125 - 45 = 1080

$ = -70.74 + (.5 bb/hand)*(1080 hands) = 469.26BB

rate = 469.26/1080 = .4345 BB/hand @ 217/1080 =

[Smurph64]

ok here goes 122k hands



Here are my values:

Filter #1: 3886 hands @ -0.2 BB/hand, 26.76 VPIP
Filter #2: 199 hands @ -0.15 BB/hand, 100 VPIP

Computation #1: -208.2 BB with 1041 VPIP hands
Computation #2: -29.85 BB with 199 VPIP hands

Computation #1 - Computation #2: -178.35 BB in 842 VPIP hands


Big blind: 3886- 199 = 3687


Big blind: -178.35 BB + (.5 BB/hand)*(3687 hands) = 1665.15 BB

Big blind: 1665.15 BB/3687 hands = 0.45 BB/hand @ 842/3687= 22.83 vpip

I think I have this right.

[Aaron W.]

Okay, so even with just a few observations:

#1 - My stats are from 6-max, and everyone else is full ring. I sort of figured this would happen, and it's fine. The major influence that this has is that I have fewer opportunities to call when an early position player raises, and also fewer chances for me to be OOP against two players by having an early position raise, a coldcaller, and having SB fold.

#2 - Relatively speaking, I call a whole lot more than you guys when in the BB and there's a raise. I have 66/749 hands (about 9% of the sample), whereas you other guys who have posted so far are much lower (Aussie, Kerowo = 4%, Smurph = 5%). I'm also doing much worse in those situations with the standard sample size concerns. I'll have to think a while longer to really get a sense of what 4% means and what 8% means, if anything. Part of me thinks you're too tight, passing up on these 5:1 odds when defending against a steal raise and not 3-betting, but again it's clearly not a definitive thing.

#3 - The main result I'm interested in is the rate of winning in limped 3 way pots when you're in the big blind. The confounding factor in this are the VPIP hands. I think it's pretty obvious that one stands to do better in a VPIP hand because the pot size is larger (6 SB instead of 3 SB) and you have a hand that's good enough to call a raise. I suspect this is what is pulling your numbers up higher than mine (not only are your hands better on average, but those hands have a 33% stronger influence on your hands than on mine).

I welcome continued stats, and if you have 6-max stats or feel like prodding 6-max players into posting their stats, I'd be curious to see how things continue to stack up.

[Smurph64]

40k 6 max hands (which were included in first calculation)

Filter 1: 1931 @ -0.19 bb/hand 20.20 VPIP
Filter 2: 127 @ 0.01 bb/hand 100 VPIP

Computation 1: -74.1 bb with 390 VPIP hands
Computation 2: 1.27 bb with 127 VPIP hands

computation 1 - 2 = -75.37 bb in 263 VPIP hands

BB = 1931 - 127 = 1804

Big blind = -75.37 bb + (902 bb) = 826.63 BB

826.63/1804 = 0.458 bb/hand @ 7% VPIP

[OziBattler]

bump

I'd actually like to see some numbers from others, including those who aint vets or think they dont run good. Who knows, if you post your numbers you might actually get some comments that are useful to ya [img]/images/graemlins/smile.gif[/img]

Yes, it might take some effort to work out what Aaron is on about and do the calculations but these sort of threads really can do with more love and the more people who post hte better IMHO. Just do it bros. mkay.

[Bona]

OK here is mine: If I followed the computations correctly

Prefiltered hands are 63,798 with VPIP of 16.9 and +2.09BB/100
~35000 hands are 6 max the rest are FR I could separate and recompute if helpful. I don't really get what you are after here.
__________________________________________________

Values filter #1: 2366 hands @-.08 VPIP 17.54
#2: 102 Hands @+.03 VPIP 100.00%

Comp#1= 415 hands -33.20BB
Comp#2= 102 hands + 3.06BB

Post reduction: 313 hands -30.14BB

Addback= BB + 1101.86 BB

Final: 1101.86BB/2264 hands = +.49BB per hand @.14VPIP

I did it but I still don't get it. [img]/images/graemlins/confused.gif[/img]

[bennyhana]

23K hands spread over a few limits.

I just plugged in numbers, now maybe one of you math wizards could put it in blue collar terms.

Filter1 :773 hands @ -0.19BB/100, 31.82VPIP
Filter2: 64 hands @ -0.16BB/100, 100VPIP

Computation1: -146BB with 249 VPIP hands
Computation2: -10.24BB with 64 VPIP hands

Computation1 - Computation2: -146 -(-10.24) = -136.76BB @ 181 hands

Big Blind = 773 - 64 = 709

Big Blind: -135.76BB + (0.5BB/hand)*709= 218.74bb

Big Blind: 218.74BB/709 = 0.3085BB/hand @ 181/709 = 25.52VPIP