merkur33 wrote:You're right, this can lead to endless discussions that are touching the most complex parts of poker strategy. however, by orientating my reads strictly at the average environment one actually plays in, one gets at least a decent prospect of how loose/timid/tight/whatever a person is relative to the average other villain. There are several ways to play profitable poker, with average stats or (more often) with non-average stats , but if i know this player folds more often his flop bets to raises than your average other opponent, i have something reliable to start with.
Indeed, I quite agree. But I'm sure you can see how, given that I was planning to distribute my HUD and that I didn't want to go into deep data analysis of trying to ascertain what ranges to use, why I would use linear ranges (which I'll talk more about in a bit).
merkur33 wrote:(i forgot to say that my stat averages are all from 7-10 player at the table. Shorthanded situations have other requirements; at a certain point its a more complex picture of situation (position, Ms, player mood, loosing/winning streaks, whatever...), player profile (from previous hud reads and made conclusions from those, eventual recent bahavior changes and from observed plays) is more relevant than a single hud stat alone, so in a way what one or another stat shows at a given moment, is far less important.)
All of my analysis and play is for full ring as well; I have admittedly played very little short handed poker and won't claim to know one bit what the proper adjustments are.
merkur33 wrote:Also, many stats are really typical in a certain region of the 100%-Range. By having a linear color distribution, one gives away too easily the most important advantage of coloring stats at all: you can read and understand them very easy and fast (almost as fast as these popular autorate-icons, which i don't like much, for several reasons). If one has a PFR of well above 20% in fullring-tourneys, i dont really care if its 28% or 58%, he is an probably an aggrodonk anyway. what im far more interested in is wheter it is 10, 12, 14 or 16%... Now i don't really know exactly how yo have the color distribution in your current hud, but i thought it was linear, and i think this is often shifting the attention towards regions that are not really relevant.
That's why I have two sets of ranges, and I chose my stats I use for coloring carefully. Remember, I don't color based on a stat's value, I color based on other stats' values. The stats that I use as color values are:
Limp/Fold %, PFR & Fold to 3bet %, 3bet vs steal %, LWPC (limp behind limpers %), Fold to 3-bet % after stealing, fold to flop cbet %, winning % given any postflop raise, fold to c-bet in 3bet pots, float turn % and fold to steal %.
So for example, my VP$IP is colored by a player's limp/fold %. A VP$IP 20 can have any number of colors - red if the player never limp/folds, magenta if he always limp/folds and anywhere in between. I find this particularly useful for a variety of reasons:
(1) It lets me see a bunch of other stats at all times on my HUD but without making things feel extremely cluttered. Once you get used to thinking "I can steal, what color is that number in? Oh, they fold to steal a lot, let's steal ATC", you'll find that it works really well.
(2) These stats(*) tend not to matter what the specific value is. I don't care if someone folds to steal 83% or 85%. If I have two villains folding 80%+ I can steal ATC profitably on the button.
(3) The stats tend to mesh with their color which makes the color easier to remember (e.g. Flop C-bet % is colored by Fold to Flop C-bet %, this makes sense).
(4) I don't play so many tables that I can't just read the numbers for stats when I care about the numbers. If I want to know someone's exact VP$IP, there it is right in front of me.
I find this approach works extremely well for me, but even ignoring the "how do you figure out what is a tight/loose range" question, I wanted to be able to remember what the values behind the colors were. If I have 10 different color ranges to remember, I'll never keep them straight. Creating one color specturm to use universally helped a lot, so I now always know that cyan is 80-90%. For ATS & Fold to 3-bet that means the guy gives up his steals at the drop of a hat (ie- weak play) but for Float Turn % it means that the guy bets the turn a lot if you cbet the flop then "give up" (ie- aggressive play). Remembering what the stat is helps me adjust at the table to the different color values, it does require actual thinking while I play though.
(*) Two of the above stats: 3bet vs steal and LWPC only tend to have very small values, so I used a different set of ranges for these values given that the numbers tend to be compressed. So instead of 10% blocks I went with 2% blocks. As it is, I don't really use the color for LWPC very much, so I could probably replace it with something better, the problem is, I can't figure out what to replace it with. 3-bet vs steal color is great though as you can tell very quickly which people are 3betting vs steals wider than they 3bet normally.
Yeah, I was particularly happy with the M workaround, I'm glad you like it
Setting it to every two hands won't work though unless you also divide the M value by two. You see, what PT3 is actually doing is summing up all values that match a given set of parameters, namely that the hand happened in the last X seconds. If multiple hands get in that range, you'll get double, triple or more of the actual value you want. That's why you want to refresh every hand and keep the number of seconds low - if you don't, you'll get numbers way out of whack. I did quite a bit of testing before settling on 30 seconds. It seemed to be enough time that PT3 could always get there but not so much time that two hands could ever finish within that window.
You could definitely use this plan to get recent temporal snapshots, just the thing to keep in mind is that people can very easily get a run of cards and just raise 3 hands in a row because hey they have the cards to raise. So long as you keep that in mind, feel free to play with it and see what sort of recent X minutes stats work well for you - I'd be curious how those work out if you build them.
) and im thinking about converting them all to cash and give this a try again (because maybe i dont suck as bad at cash games any longer as i did one year ago...). But the procedure to convert them manually is way to complicated (I maybe lost one month of poker just to figure out stats and huds). Is there any workaround with search&replace involved? Basically i just have to replace "tournament_" with nothing in lots of configuration files, or am i wrong? Later on i could then tweak one or antoher stat which doesn't work in cash, but it would save me lots of work. 
I will to this the next time i need to come off-tilt again. BTAICBW...
