Stat's Interpretation (What the Autorate is Cappable of) (long)

[nomadtla]

[disclaimer] Stats and autorates are not reads, get reads, use stats/autorates sparingly and don't base all decissions on them, they can be helpful but don't make them a crutch. [/disclaimer]

I agree with Wookie's statement in This Thread that we should deffinately not base all our poker decisions on what stats tell us, or even worse base them on broad groupings of those stats. I was still interested in how far the autorate could go, esspescially if it could be used to better judge postflop play.

First let me say "If I saw farther then others it was because I stood on the shoulders of giants", and I have the utmost respect for "BisonBison" and "Harv72b". I also did not do my fair share on this compared to what "Kwaz" did, and the ideas he oppened up for me.

For those interested Bison's clasic autorate can be found HERE
While mostly an autorate it is generally mis-referenced as a thread on note taking. While it does contain some good note taking information if you look for it, most of the replies are on the autorate.

The other known autorate is Harv72b's which can be found HERE
This is allso a good system though the stats it bases on take a while to converge.

The problem I had with both of them was that they were overly focussed on preflop IMO. I am more concerned with whether a player is aggressive postflop (or will pay off) then I am with their aggression preflop. I like most any PT user, can look at a PFR # >15 and know they raise too much preflop or <5 and know they raise premiums only. * see diclaimer

With that all said:
Here are some things I have done. Please if you are interested do your own further research into this matter as I feel this merely scratches the surface. But I feel it opens some doors for people to again "see farther then I have"

The problem is PT only has 14 icons to autorate with, so something from the old autorates has to go. Since VPIP has a bering on a players AF postflop, the only logical choice was to drop PFR as a variable for autorate.

The idea we came up with was to use a general autorate based on postflop AF that only took up 10-11 icons. That way we would be able to use 3 or 4 others to single out special groups.

The 11 category general autorate looks like this:

1 Sub Group 1 (SG1)
2 Sub Group 2 (SG2)
3 Sub Group 3 (SG3)
4 Loose Reasonable (LR)
Vol. Put Money In Pot % is greater than or equal to 35.00
Aggression Factor - Total is greater than or equal to 1
5 Loose Passive (LP)
Vol. Put Money In Pot % is greater than or equal to 35.00
Aggression Factor - Total is less than or equal to .99
6 Semi-Loose /Passive (SLP)
Vol. Put Money In Pot % is greater than or equal to 23.00
Aggression Factor - Total is less than or equal to 1.19
7 Semi-Losse /Aggressive (SLA)
Vol. Put Money In Pot % is greater than or equal to 23.00
Aggression Factor - Total is greater than or equal to 1.2
8 Tight Passive (TP)
Vol. Put Money In Pot % is between 16.00 and 22.99
Aggression Factor - Total is less than or equal to 1.49
9 Tight Neutral (TN)
Vol. Put Money In Pot % is between 16.00 and 22.99
Aggression Factor - Total is between 1.50 and 2.19
10 Tight Aggressive (TA)
Vol. Put Money In Pot % is between 16.00 and 22.99
Aggression Factor - Total is between 2.20 and 3.19
11 Tight Super Aggressive (TSA)
Vol. Put Money In Pot % is between 16.00 and 22.99
Aggression Factor - Total is greater than or equal to 3.20
12 Super Tight Passive (STP)
Vol. Put Money In Pot % is less than or equal to 15.99
Aggression Factor - Total is less than or equal to 2.19
13 Super Tight Aggressive (STA)
Vol. Put Money In Pot % is less than or equal to 15.99
Aggression Factor - Total is between 2.20 and 3.49
14 Super Tight Super Aggressive (STSA)
Vol. Put Money In Pot % is less than or equal to 15.99
Aggression Factor - Total is greater than or equal to 3.50

Obviously you want to leave the first few spaces of your autorates open for your subgroups so that it weeds out the specifc groups that bother you or benefit you. Then let the general autorate catch and divide the straglers to give an idea.

As far as the icons go pic whichever ones work for you to make the conection. Play with it a while and see what works for you.

If you want more room for subgroups you can combine some of the categores like drop Tight Super Aggressive. Or combine the loose groups. Or drop all the super tight groups. Like I say this is your autorate play with it and do what you want with it. I am merely trying to help guide people.

You'll notice that the AF line between passive and aggressive varies as the VPIP goes up. This is a understandable correlation but one that is hard to quantify. But as a player's VPIP increases their AF decreases though they may be still as sound postflop because the more flops you see the more marginal situations you are put in where calling is correct. I developed an equation for an estimate of the line between passive and agressive AF for any VPIP.
[equation] 1 = (AF x VPIP)/30 [/equation]
If you plug a players VPIP and AF into this equation and it equals a # greater then 1 they leand towards aggressive, if it's less then 1 they lean towards the passive side.
As I said this is just an esstimate and not Gospel.

So now you may be asking yourself well what subgroups should I put in. Well, thats where you come in. You could have one for super high VPIP (>60)like wookie does.

Me and Kwaz did find there are interesting relations between street AF's and a players "regular line"

So I have these sub groups:
Chasers:
Vol. Put Money In Pot % is greater than or equal to 23.00
Aggression Factor - Flop is less than 1.00
Aggression Factor - Turn is less than 1.00
Aggression Factor - River is greater than or equal to 2.50
As you can see these players are passive till they see the whole board and then it's either raise or fold.

Calling Stations:
Vol. Put Money In Pot % is greater than or equal to 23.00
Aggression Factor - Flop is less than 0.65
Aggression Factor - Turn is less than 0.85
Aggression Factor - River is less than 1.00
As you can see these players are passive all streets probably not betting without something serious.

The No Folder:
Vol. Put Money In Pot % is less than 22.99
Aggression Factor - Flop is greater than 2.00
Aggression Factor - Turn is less than 1.70
Aggression Factor - River is less than 1.40
This is a group that starts with good hands and usually continuation bets the flop but becomes passive on later streets. This style is very prevelant with Table Coaches. Thinking "I started with a premium hand and I will call it down no matter what"

Some other things to think about and explore:
From Kwaz's notes to me, I found these all to be good esstimates.
[ QUOTE ]

Players with a solid rise in AF from flop to river are chasers. Fish. If they raise the river - fold. They made the hand they were chasing. But they will pay for it every time. These guys are nut peddlers. They won't call any river bets with less than top pair.

Players who have the turn as there highest AF with a steady drop either side are players who have read one book. They know the preflop table but they play fit or fold post flop. They give up most of their hands on the turn.

Players who start with low flop AF and rise on the turn and level out on the river. I think these guys are the best fish of all. They call on the flop, they only drop hopeless hands on the turn but will call down on the river with many second best hands.

Players who drop in AF from flop to river. Again they've read one book. They know preflop and play from the tight tables. They will follow everything up with a continuation bet but have real trouble letting anything go. They can't bring themselves to fold because they've got a "Premium Starting Hand"


[/ QUOTE ]

So look at street agression factors and see what other trends you can find.

As I said all of this is not Gospel.

I think me and Kwaz both came to the conclusion that there is no perfect way to do this, and no autorate will fit every 2+2er's style.

* You've made it this far read the disclaimer again

So I hope this opens your eyes to some of the features that the autorate can do. And I really hope people go beyond this and give a widder variety of subgroups. You can post them in here if you like so all our autrate stuff can stay in one place and not clutter the rest of the forum because as the disclaimer says, overall this is a minor thing.

One thing me and Kwaz learned from this
Do not waste all your playing time looking for the Holy Grail of autorates

Now go forth and I hope this helps. And please make updates of ideas along this line that you run accross.

* Read the disclaimer at the top again

Thank You
Thomas

[ajm36]

Great post--one thing I don't understand is this--You rated Tight Neutrals with AF between 1.5 and 2.19 and Super Tight Passives less tahn 2.19. There is clearly an overlap here and I was curious about this particular classification as Neutral in one rating but Passive in another.

[Shillx]

Hey,

I like the groups that you have setup, but there are some glaring flaws. Having said that, this is a step in the right direction. Here is what I would do if I still played LHE ring:

1) Keep 3 VPIP groups (< 22, 22-35, > 35).

2) Attach a PFR to each of them. Set the cutoff at 5% for high VPIPers and 7% for other players. Above that is aggressive and below that is passive. There are essentially 2 reasons to do this.

a) It converges faster then AF and is a good measure of aggression.

b) It tells us something more then just AF would alone.

3) Use a sliding scale as you suggest for the AF.

- For loose make the ranges (< .8, .8-1.2, > 1.2)
- For typical make them (< 1.2, 1.2-1.8, > 1.8)
- For tight make them (< 1.5, 1.5-2.5, > 2.5)

Something like that.

Good post and hopefully we can get some ideas brewing to create the best possible Autorate range.

[nomadtla]

[ QUOTE ]
Great post--one thing I don't understand is this--You rated Tight Neutrals with AF between 1.5 and 2.19 and Super Tight Passives less tahn 2.19. There is clearly an overlap here and I was curious about this particular classification as Neutral in one rating but Passive in another.

[/ QUOTE ]

[ QUOTE ]
You'll notice that the AF line between passive and aggressive varies as the VPIP goes up. This is a understandable correlation but one that is hard to quantify. But as a player's VPIP increases their AF decreases though they may be still as sound postflop because the more flops you see the more marginal situations you are put in where calling is correct. I developed an equation for an estimate of the line between passive and agressive AF for any VPIP.
[equation] 1 = (AF x VPIP)/30 [/equation]
If you plug a players VPIP and AF into this equation and it equals a # greater then 1 they leand towards aggressive, if it's less then 1 they lean towards the passive side.
As I said this is just an esstimate and not Gospel.


[/ QUOTE ]

A much tighter player must have a higher aggression factor before they reach neutral.
for instance a player with a VPIP of 10 would be at aggressive >3 If anything the line for passives for supertight is acctually a little low.

They don't overlap cause VPIP will seperate them before it judges their AF.

[ajm36]

Thanks, missed that.

[nomadtla]

Hey Shillx,
Thanks for the reply

[ QUOTE ]

2) Attach a PFR to each of them.... There are essentially 2 reasons to do this.

a) It converges faster then AF and is a good measure of aggression.

b) It tells us something more then just AF would alone.


[/ QUOTE ]

These are both very good points, and something I feel kinda lacked from our design.

The problem with dividing players by VPIP, PFR, and AF is that PT only has 14 icons. To fully divide by the method you layed out would take 18 icons. We have to give somewhere. I chose to give PFR so more categories could be used elsewhere (though I'm not sure that it's the best way).

[ QUOTE ]
3) Use a sliding scale as you suggest for the AF.

- For loose make the ranges (< .8, .8-1.2, > 1.2)
- For typical make them (< 1.2, 1.2-1.8, > 1.8)
- For tight make them (< 1.5, 1.5-2.5, > 2.5)

[/ QUOTE ]

This is a deffinite and probably the most important part of my OP. Since a neutral AF changes as VPIP goes up we can't base an autorate dividing every player on that line of 1.5, or 2, or any other set # across the board.

[ QUOTE ]
hopefully we can get some ideas brewing to create the best possible Autorate range.

[/ QUOTE ]

This was my intent more then anything to just get a discussion going and have a new thread that would encompass this discussion. I'm not looking to set anything in Stone for anyone. As I said in my OP, finding the perfect way might not be possible, and even if it were since we all have variables in our styles what helps one player might not help me and vice versa.

I was also looking to discuss other Sub-catergory ideas. If anyone had noticed a particular street by street change meaning one thing. Or used the autorate function like Wookie to weed out a select group to look for or avoid.

As far as I know the relation of street by street AF's has not been discussed as a base for finding particular charecteristics. I know some of us have street AF's up in our HUD's but wondered if anyone else had been able to define a ultra high or low AF on one street, or a relation of increasing/decreasing street AF's meant a particular style of villan.

So I just wanted to throw my thoughts out there if they help anyone, and see if anyone else had similar ideas or noticed similar trends.

[Guruman]

I like the idea of modifying the autorate based on postflop play, and have been toying with the idea myself.

I think it would be better served to rate individuals who fall into very specific archetypes, and forego the attempt to rate everyone

By identifying and labeling specific patterns, I'll bet we can get better use out of the function. I don't think its any high travesty to have 100 hands on a player and not have them autorated yet.

What is a travesty is folding to the guy with the rock icon next to him when he tends to like showing down with anythin UI once he gets into a hand.

I wish the autorate could see acutal lines like [raised two flush board and checked through the turn]. then it would rawk.

[HouseCalls]

[ QUOTE ]

2) Attach a PFR to each of them.... There are essentially 2 reasons to do this.

a) It converges faster then AF and is a good measure of aggression.

b) It tells us something more then just AF would alone.


[/ QUOTE ]

These are both very good points, and something I feel kinda lacked from our design.



[/ QUOTE ]

If you are using PAHUD you can have both PFR and the category or icon up next to each player which may give more info here.

Great work Tom and Kwaz - and bump for the pm crowd

[nomadtla]

[ QUOTE ]
I think it would be better served to rate individuals who fall into very specific archetypes, and forego the attempt to rate everyone

[/ QUOTE ]

I agree I just wanted to throw something out there to start the conversation.

This is what I was really looking for all along in an autorate. I'd love to have one that would give me better surety of folding to a river raise. I would rather have 12-14 solid groups even if they were only 5% my DB per group if it meant I had more surety in their postflop action.

Bumped to accompany Cold Cash's post.
And more on how we came up with our post flop focus.

I can't be bothered to write it out again.

http://forumserver.twoplustwo.com/showfl...e=0#Post4189456

Both Nomadtla and I spent a lot of time going through our databases looking for a better rating system. As he says, there is no holy grail, but hopefully we came a little closer.

The relative aggression equation that we hammered out, which we both use slightly differently (see linky) is very useful.

I've also saved myself quite a few dollars with the "chaser" category. And this is what it's all about.


Keep value betting the calling stations [img]/images/graemlins/grin.gif[/img]