Predicting BB/100 based on other stats

[jerome baker]

i dont know if this means anything but,

1. i ave 148 hands/ hr even though i play 4 tables.
this probably means i leave quickly when conditions change.

2. most of my hrs are from 12am (west coast) to whenever im "burnt out"

3. the games might be "harder" than they were a yr ago.

4. i know the 5/10 games suck (probably under 20% of the games are "playable").
ave pot is 67 (under session notes), not loose but not that bad.

5. my att to steal is not crazy, but probably 3% higher than "normal"

6. luck

ps. my 5/10 6max stats looks pretty messed up also.
(i only have 28585 hands)
my wtsd is 42.26
my wsd is 48.92

[MicroBob]

[ QUOTE ]
the formula provides some statistical validation to our sense of what makes for a winning hold 'em player - figuring out how to get down and dirty to win pots, while avoiding paying off with second best hands.

[/ QUOTE ]



That's pretty much what I took from all this.

Great stuff Nate.

[sin808]

Didn't even bother to think of that at the time. Thanks, that makes perfect sense.

[beachbum]

[ QUOTE ]
BB/100 = (.753 * VPIP) - (.0102 * (VPIP^2)) + (.435 * W$SD) + (.658 * W$WSF) - 56.12

[/ QUOTE ]

Being an engineer by trade, this analysis intrigued me a little. Nate, in your equation you have a dependent variable (BB/100) and 3 independent variables (VPIP, W$SD, and W$WSF). However, the 3 independent variables are not truly independent as they're dependent on each other.

Excel has a neat little feature called Solver (Tools -> Solver). It's an Excel add-on, so if it's not installed just pop in the Microsoft Office cd again and go to Tools -> Add-Ins and select Solver Add-in.

So if you could use regression to find equations for VPIP, W$SD, and W$WSF in terms of each other, you might be able to use this Solver feature to find optimal values for VPIP, W$SD, and W$WSF to maximize your winrate. You basically give Solver constraints, tell it what your targets are, give variables ranges, etc. Then it just runs and iterates and tries to converge on a solution (a maximum winrate).

It can be a little touchy and you have to keep tweaking it if it doesn't work with the initial inputs and ranges. But, I thought it'd be interesting to see if it gave any useful results.

[whitelime]

Right. I figured the estimate would be much closer to something reasonable like 3.5BB/100 and 24%VPIP in which case the numbers wouldn't change that much.

[stinkypete]

did you consider the effect of river aggression?

varied flop and turn aggression will certainly be reflected in adjusted W$@SD and W$WSF values (ie. higher aggression = lower W$@SD and higher W$WSF), but what about river aggression?

i would think that given two players who otherwise play the same, but one having significantly higher river aggression than the other, the player with higher river aggression would have a statistically significantly lower W$@SD, but without a much higher W$WSF. that is, someone who bets/raises on the river a lot will have opponents folding a lot of losing hands (W$@SD is lowered) but not folding a significant amount of winning hands (W$WSF doesn't increase significantly).

[bicyclekick]

Why is there a second stheif?

I'll just AIM it to the real one in case you're some lame imposter. :P

[bobbyi]

Isn't it depressing to see that you are stuck $170 and have paid $2650 in rake? Man, I hate rake.

[johnnycakes]

So, to use this formula, in PT, I need to filter hands with between 7 and 10 players at table and then use the numbers in the "Known Starting Hands" section on the General Info tab, correct? Is 7-10 players acceptable, or should it be 8-10?

Thanks.

[MaxPower]

Did you try running any models with pre-flop raise percentage? This should have some effect on win rate, but at 15/30 there probably isn't as much variance in pre-flop raise percentages than there is at the lower limits.

Another question Nate. What is the R squared for this model?