Making a new auto-rate rule file for pokertracker.

[OrianasDaad]

This is a crosspost from books/software.
-----------------

Another thread got me thinking about setting up some of my own autorate rules, since playerview takes care of VPIP, PFR, and AGR. I think I can use the poker tracker data to come to conclusions about more subtle things about my opponents play.

For instance:

W$SD < 40, River Agg > 1.5

This might indicate someone who bluffs on the river a bit too often.

My open questions to the forum are:

What types of data do you use to read opponents beyond Bison's auto-rate rules?

At what point are these reads statistically reliable?

How important is this "statistical read" to your winrate?

Looking for lots of ideas.

[easypete]

[ QUOTE ]
What types of data do you use to read opponents beyond Bison's auto-rate rules?

[/ QUOTE ]

Actually, I hardly ever look at the icon description any more. I think it was good for me for about 25k hands, then I found I played better by looking at VPIP, PFR, AF idividually as I needed it.

For instance... I get c/r on turn. First thing I look at is the AF. If it's 0.3... I fold. If it's > 3, I raise. In between, I may call down. I also look at the c/r%.

Bison's rating system (one of the best things that's happened to help play the game) is a good start. But you need to play the player in the situations, this requires digging into the stats when you need them. Bison's system has 12 descriptions (IIRC). I usually think of players as about 36-40 different types.

[OrianasDaad]

That's exactly it. I'm looking to create a set of rules for players like us that don't need the Bison's auto-rate rules anymore. Something that will order the specific reads you can get from the statistical info PT provides.

[SlantNGo]

For postflop play, I find that Agg. Factor is deceiving. Many postflop LAGs also tend to call down a lot, which keeps their AF low, so the telltale sign of this is a high Went to SD % and a low Won at SD %. Those are the main 2 I look to when I'm heads-up or short-handed after the flop and deciding whether to semi-bluff. Same goes for value bets.

[easypete]

[ QUOTE ]
That's exactly it. I'm looking to create a set of rules for players like us that don't need the Bison's auto-rate rules anymore. Something that will order the specific reads you can get from the statistical info PT provides.

[/ QUOTE ]

I don't think I understand completely. Set of rules for what? A new autorate? We only have 12 icons we can have. I don't know... I don't think this is worth doing... If you're past the need for bison's autorating system, then it becomes player dependant. I wouldn't put too much thought into it, nor do I think I would use it. Classifying players into 12 catagories is too limiting.

[OrianasDaad]

Well, for my particular example, mabye it would be better to make it: River Agg > 1.5 and W$DS < 40 and WtDS > as something high.

As an interesting aside, and useful discovery - I found that there is no limit to the number of criteria you can put on a rule.

Here are the stats that PT auto-rate rules can use:

Aggression: Pre-flop, Flop, Turn, River
Amount Won
Attempts to steal blinds %.
BB/100
Check-raise %
Folded BB to steal %
Folded SB to steal %
Folded to river bet %
Pre-Flop raise %
Put $ in Small Blind %
Saw Flop All Hands %
Saw Flop not a blind %
VPIP %
Went to Showdown %
When Folds: No Fold, Flop, Turn, River %
Win rate/100 hands
W$SD %
Won $ when saw flop %

[OrianasDaad]

[ QUOTE ]
[ QUOTE ]
That's exactly it. I'm looking to create a set of rules for players like us that don't need the Bison's auto-rate rules anymore. Something that will order the specific reads you can get from the statistical info PT provides.

[/ QUOTE ]

I don't think I understand completely. Set of rules for what? A new autorate? We only have 12 icons we can have. I don't know... I don't think this is worth doing... If you're past the need for bison's autorating system, then it becomes player dependant. I wouldn't put too much thought into it, nor do I think I would use it. Classifying players into 12 catagories is too limiting.

[/ QUOTE ]

I'd argue that classifying them into the 14 (there are 14 icons) most useful sub-categories might be very worthwhile.

And yes, it's for a new auto-rate file I'm making, hence the title of this post.